{"id":204385,"date":"2025-10-06T04:12:10","date_gmt":"2025-10-06T04:12:10","guid":{"rendered":"https:\/\/www.newsbeep.com\/us\/204385\/"},"modified":"2025-10-06T04:12:10","modified_gmt":"2025-10-06T04:12:10","slug":"tip-of-the-iceberg-understanding-the-full-depth-of-big-techs-contribution-to-us-innovation-and-competitiveness","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/us\/204385\/","title":{"rendered":"Tip of the Iceberg: Understanding the Full Depth of Big Tech\u2019s Contribution to US Innovation and Competitiveness"},"content":{"rendered":"<p class=\"TocHeading\">Contents<\/p>\n<p class=\"Toc1\"><a href=\"#_Toc210372951\">Key Takeaways 1<\/a><\/p>\n<p class=\"Toc1\"><a href=\"#_Toc210372952\">Introduction. 2<\/a><\/p>\n<p class=\"Toc1\"><a href=\"#_Toc210372953\">Societal Benefits 4<\/a><\/p>\n<p class=\"Toc1\"><a href=\"#_Toc210372954\">Recognize and Account for the \u201cFull Iceberg\u201d 33<\/a><\/p>\n<p class=\"Toc1\"><a href=\"#_Toc210372955\">Conclusion. 34<\/a><\/p>\n<p class=\"Toc1\"><a href=\"#_Toc210372956\">Endnotes 34<\/a><\/p>\n<p class=\"Normal\">\u00a0<\/p>\n<p class=\"Normal\">As late as the early 2010s, large U.S. information technology companies were celebrated by politicians, media, and academics. Now they attack big tech as too powerful and greedy. Liberals complain that they are surveilling society, spreading harmful content, and undermining democracy. Conservatives say that they are censoring free speech, influencing elections, and undermining traditional American values. Both sides believe that they are harming children and both right and left populists want to break them up.<a href=\"#_edn1\" title=\"\" id=\"_ednref1\">[1]<\/a><\/p>\n<p class=\"Normal\">This \u201cvoguish\u201d critique is based on at best a superficial understanding of what these companies do: Apple makes iPhones; Amazon makes an online store; Alphabet makes a search engine; Microsoft makes Word, Excel, and PowerPoint; and Meta makes social networks. Especially for those who dismiss social media as at best trivial entertainment, this view sees only the tip of the iceberg of the massive benefits large technology companies provide to the world and the United States.<\/p>\n<p class=\"Normal\">At the same time critics attack big tech, they also argue that U.S. corporations don\u2019t invest enough in the long term\u2014in research and development (R&amp;D), in capital investment, in efforts that have high risks and long-term returns. If only these \u201cquarterly capitalism\u201d companies would act better, everyone would be better off. And yet, big tech is doing the opposite of quarterly capitalism, and those efforts should be praised rather than excoriated.<\/p>\n<p class=\"Normal\">While many major U.S. technology firms\u2014including Cisco, Qualcomm, and Nvidia\u2014demonstrate similar innovation patterns and positive spillovers, space constraints limit this report to examining five: Apple, Amazon, Alphabet, Meta, and Microsoft.<\/p>\n<p class=\"Normal\">R&amp;D: In 2024, the five largest U.S. technology companies invested $227 billion in R&amp;D, exceeding the federal government\u2019s non-defense R&amp;D budget and outpacing the annual R&amp;D expenditures of most countries.<a href=\"#_edn2\" title=\"\" id=\"_ednref2\">[2]<\/a> These investments were not limited to core business lines. Research by the Information Technology and Innovation Foundation (ITIF) finds that approximately 78 percent of non-manufacturing R&amp;D growth came from just four industries: electronic shipping and electronic auctions; data processing, hosting, and related services (e.g., cloud providers); software publishers; and other information (e.g., Internet companies).<a href=\"#_edn3\" title=\"\" id=\"_ednref3\">[3]<\/a> Big tech companies also fund long-horizon research into artificial intelligence (AI), quantum computing, and semiconductors; they underwrite experimental \u201cmoonshot\u201d projects that may fail commercially but advance technical capabilities; and they generate spillovers that diffuse across academia, startups, and industry more broadly.<a name=\"OLE_LINK1\" id=\"OLE_LINK1\"\/><a href=\"#_edn4\" title=\"\" id=\"_ednref4\">[4]<\/a><\/p>\n<p class=\"Normal\">Capital Investment: In 2024, these companies invested over $250 billion in capital expenditures, and cumulatively, they have invested over $1 trillion in the 21st century.<a href=\"#_edn5\" title=\"\" id=\"_ednref5\">[5]<\/a> Training a competitive large language model can cost more than $100 million in compute resources.<a href=\"#_edn6\" title=\"\" id=\"_ednref6\">[6]<\/a> Developing quantum processors requires billion-dollar fabrication facilities and programs extending over a decade or more.<a href=\"#_edn7\" title=\"\" id=\"_ednref7\">[7]<\/a> Building global satellite constellations or hyperscale cloud infrastructure involves up-front investments in the tens of billions of dollars.<a href=\"#_edn8\" title=\"\" id=\"_ednref8\">[8]<\/a> Amazon, Alphabet, Microsoft, Meta, and Apple finance data centers, subsea cables, renewable energy projects, and custom semiconductor platforms that underpin not only their own operations but also the broader digital economy.<a href=\"#_edn9\" title=\"\" id=\"_ednref9\">[9]<\/a><\/p>\n<p class=\"Normal\">Public Knowledge Generation: Big tech firms publish thousands of research papers annually, release open source frameworks, and make large datasets and models available to universities and startups.<a href=\"#_edn10\" title=\"\" id=\"_ednref10\">[10]<\/a> Initiatives such as Google DeepMind\u2019s AlphaFold, which mapped 200 million protein structures and released them freely, accelerate biomedical research worldwide.<a href=\"#_edn11\" title=\"\" id=\"_ednref11\">[11]<\/a> This openness distinguishes U.S. firms from more traditional corporate R&amp;D models, wherein proprietary knowledge rarely circulates outside company boundaries.<\/p>\n<p class=\"Normal\">STEM (Science, Technology, Engineering, and Math) Skill Development: These companies serve as training grounds for highly skilled researchers and engineers who later move into startups, universities, and government laboratories. Alums have founded thousands of companies, raising tens of billions of dollars in venture funding and diffusing organizational knowledge about building scalable systems.<a href=\"#_edn12\" title=\"\" id=\"_ednref12\">[12]<\/a> In sectors such as AI, where global competition for talent is intense, this clustering effect strengthens the U.S. ecosystem as a whole.<a href=\"#_edn13\" title=\"\" id=\"_ednref13\">[13]<\/a><\/p>\n<p class=\"Normal\">Funding for Startups: Acquisitions and partnerships link large firms to the startup ecosystem, providing capital, technical infrastructure, and global reach. While critics describe this as anticompetitive, empirical evidence shows that acquisition activity often validates markets and attracts further venture investment.<a href=\"#_edn14\" title=\"\" id=\"_ednref14\">[14]<\/a> High-profile cases, such as that of YouTube and Android, demonstrate how integration with larger platforms can rescue promising technologies from collapse and scale them to have a global impact.<a href=\"#_edn15\" title=\"\" id=\"_ednref15\">[15]<\/a> Blocking acquisitions, by contrast, has at times produced the opposite of the intended result, leading to firm failure and strategic losses to foreign competitors.<a href=\"#_edn16\" title=\"\" id=\"_ednref16\">[16]<\/a><\/p>\n<p class=\"Normal\">Innovations Beyond the Information Sector: Big Tech is primarily about information\u2014generation, storage, analysis, and transmission. But it is about much more than that. These companies are innovating in a range of important areas, including national defense, transportation, agriculture, education, health care, space travel, and more.<\/p>\n<p class=\"Normal\">The United States finds itself in an existential techno-economic war with China, where the Chinese Communist Party (CCP) is seeking to dominate all advanced industries, including manufacturing and information.<a href=\"#_edn17\" title=\"\" id=\"_ednref17\">[17]<\/a> And it has core advantages, including a closed domestic market, massive government subsidies, and favorable antitrust rules that encourage size and scale. Given the severe U.S. government fiscal constraints, it is not realistic to expect the U.S. government to match the financial largess the CCP bestows on its big tech champions. If the United States does not want to lose to China, it needs to strengthen, not weaken, its large technology companies.<\/p>\n<p class=\"Normal\">Too many have defaulted to adopting the European model of hobbling, Gulliver-like, America\u2019s large technology companies through taxes, regulations, fines, antitrust breakups, and other restrictions.<a href=\"#_edn18\" title=\"\" id=\"_ednref18\">[18]<\/a>The idea that doing so will boost U.S. innovation and competitors is so farcical that it\u2019s amazing that anyone can believe it. If we continue down that Brussels path, the result will be clear: an America where, in a decade or two\u2019s time, today\u2019s big tech firms are looked at nostalgically the way we look at past tech leaders such as Xerox, RCA, Lucent, and Kodak.<\/p>\n<p class=\"Normal\">That is why, at the very least, what the U.S. policy can do with regard to its own big tech companies is no harm. Doing so will certainly strengthen America\u2019s global competitiveness and the jobs big tech creates at home. It will also support U.S national security.<a href=\"#_edn19\" title=\"\" id=\"_ednref19\">[19]<\/a> American advances in AI, semiconductors, and quantum computing are overwhelmingly driven by private R&amp;D. The Pentagon adapts these capabilities for defense applications, benefiting from billions in commercial investment that government contracts alone would not reproduce at a similar scale or speed.<a href=\"#_edn20\" title=\"\" id=\"_ednref20\">[20]<\/a> And it will help enable innovation across a range of traditional sectors where traditional firms are underinvesting in high-risk activities.<\/p>\n<p class=\"Normal\">Overall, the prevailing discourse reduces these firms to consumer platforms and market power, overlooking the broader system of innovation, infrastructure, and security that depends on their scale. Policies that treat big tech primarily as a problem risk weakening assets that underpin U.S. competitiveness. A more complete assessment requires recognizing the positive externalities they generate.<\/p>\n<p class=\"Normal\">R&amp;D determines which nations lead in technological advancement and economic competitiveness.<a href=\"#_edn21\" title=\"\" id=\"_ednref21\">[21]<\/a> The source and scale of R&amp;D funding shape what technologies get developed, how quickly they reach market, and which countries capture the resulting economic benefits. Today, private corporations have emerged as the dominant funders of R&amp;D, with five U.S. technology companies alone investing more than the entire federal government and most other countries.<\/p>\n<p class=\"Normal\">This concentration of research capacity creates distinctive advantages, namely the ability to sustain decade-long losses on experimental technologies, pursue innovations requiring billions in up-front investment, maintain portfolio approaches to high-risk research, and compete with state-backed entities without direct government support.<\/p>\n<p class=\"Normal\">These investments are fundamentally profit driven. But while companies pursue R&amp;D to maintain a competitive advantage and capture future markets, they also generate substantial positive externalities. The technologies, knowledge, and capabilities developed through corporate R&amp;D frequently diffuse beyond the investing firms, creating benefits for competitors, academia, and society at large that exceed the private returns captured by the companies themselves.<\/p>\n<p>Unique Financial Capacity for High-Risk R&amp;D<\/p>\n<p class=\"Normal\">Transformative technologies typically require extended periods of investment, with highly uncertain returns. The transistor took 23 years from its invention at Bell Labs to reach widespread commercial adoption.<a href=\"#_edn22\" title=\"\" id=\"_ednref22\">[22]<\/a> The Internet evolved over decades with government support before private commercialization.<a href=\"#_edn23\" title=\"\" id=\"_ednref23\">[23]<\/a> These extended timelines, combined with high failure rates, don\u2019t make it very desirable to fund research that may take decades to pay off, if ever. Traditional capital markets demand returns within 5\u201310 years. Government grants operate on even shorter cycles. This mismatch between innovation timelines and funding horizons constrains breakthrough research.<\/p>\n<p class=\"Caption\"><a id=\"_Ref209711203\">Figure<\/a> 1: Big tech companies\u2019 R&amp;D spending, 2015\u20132024 (billions)<a href=\"#_edn24\" title=\"\" id=\"_ednref24\">[24]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image001.png\" width=\"768\" height=\"480\" alt=\"image\"\/><\/p>\n<p class=\"Normal\">The leading five U.S. technology companies\u2014Amazon, Apple, Alphabet, Meta, and Microsoft\u2014invested $227 billion in R&amp;D in 2024.<a href=\"#_edn25\" title=\"\" id=\"_ednref25\">[25]<\/a> This exceeds the $172 billion U.S. federal R&amp;D budget and the total annual R&amp;D expenditure of most countries in 2023.<a href=\"#_edn26\" title=\"\" id=\"_ednref26\">[26]<\/a> The scale represents not just large numbers but also a qualitative difference in innovation capacity. When individual companies deploy tens of billions of dollars annually on research, they can sustain projects that would exhaust traditional funding sources.<\/p>\n<p class=\"Caption\">Figure 2: U.S. private sector R&amp;D spending, 2015\u20132024 (billions)<\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image002.png\" width=\"768\" height=\"480\" alt=\"image\"\/><\/p>\n<p class=\"Normal\">Google\u2019s parent company, Alphabet, demonstrates this capacity through its experimental X division. In 2016, the division reported quarterly losses of $859 million pursuing \u201cmoonshot\u201d projects, with estimated annual losses between $3 billion and $4 billion.<a href=\"#_edn27\" title=\"\" id=\"_ednref27\">[27]<\/a> These projects included Internet-delivery balloons, glucose-monitoring contact lenses, and autonomous vehicles\u2014technologies with uncertain commercial prospects but potentially transformative impacts. Despite the losses, Alphabet\u2019s stock rose 6 percent after the earnings announcement, reflecting investor acceptance of long-term research investments supported by profitable core businesses.<\/p>\n<p class=\"Caption\">Figure 3: U.S. big tech companies\u2019 R&amp;D spending vs. select countries\u2019 totals, 2015\u20132024 (billions)<a href=\"#_edn28\" title=\"\" id=\"_ednref28\">[28]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image003.png\" width=\"768\" height=\"480\" alt=\"image\"\/><\/p>\n<p class=\"Callout\">Alphabet\u2019s experimental X division reported quarterly losses of $859 million pursuing moonshot projects. Despite the losses, Alphabet\u2019s stock rose 6 percent, reflecting investor acceptance of long-term research investments.<\/p>\n<p class=\"Normal\">Amazon\u2019s approach to hardware development illustrates sustained commitment despite commercial failure. The company\u2019s devices division, centered on Alexa-enabled products, lost over $25 billion between 2017 and 2021.<a href=\"#_edn29\" title=\"\" id=\"_ednref29\">[29]<\/a> In just the first quarter of 2022, the division lost another $3 billion.<a href=\"#_edn30\" title=\"\" id=\"_ednref30\">[30]<\/a> The company sold Echo devices at or below cost, hoping to generate revenue through voice commerce that never materialized. Yet, Amazon sustained these losses for nearly a decade, advancing voice recognition and natural language processing technologies in the process.<\/p>\n<p class=\"Normal\">Custom semiconductor development represents another domain where only the largest technology companies can compete. Apple\u2019s transition to proprietary silicon required years of development and billions in investment, culminating in chips that now power its entire product line. This vertical integration, designing processors specifically optimized for their software and AI workloads, requires sustained investment that few companies can match.<\/p>\n<p class=\"Caption\">Figure 4: Subset of big tech companies\u2019 losses on high-risk, high-reward moonshot projects<a href=\"#_edn31\" title=\"\" id=\"_ednref31\">[31]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image004.png\" width=\"768\" height=\"384\" alt=\"image\"\/><\/p>\n<p class=\"Normal\">This financial capacity derives from profitable core businesses that generate massive cash flows. Amazon Web Services (AWS) produced $62 billion in revenue in 2021 with operating margins exceeding 30 percent.<a href=\"#_edn32\" title=\"\" id=\"_ednref32\">[32]<\/a> Alphabet\u2019s advertising business generates over $200 billion annually.<a href=\"#_edn33\" title=\"\" id=\"_ednref33\">[33]<\/a> Big tech companies funnel significant shares of profits to research portfolios that no other entities can easily match in scale or duration. The private sector cross-subsidization model bypasses capital market constraints that would terminate unprofitable research in traditional settings.<\/p>\n<p>Scale Creates Fundamentally Different Innovation Possibilities<\/p>\n<p class=\"Normal\">Frontier technologies exhibit escalating capital requirements that create natural barriers to participation. Breakthroughs in AI, quantum computing, and autonomous systems demand massive up-front investments. These requirements stem from fundamental technical constraints rather than artificial barriers.<\/p>\n<p class=\"Normal\">Training large language models exemplifies the scale. OpenAI\u2019s GPT-3 required over $4.6 million in compute costs alone for a single training run.<a href=\"#_edn34\" title=\"\" id=\"_ednref34\">[34]<\/a> GPT-4 pushed costs above $100 million.<a href=\"#_edn35\" title=\"\" id=\"_ednref35\">[35]<\/a> Industry estimates suggest that next-generation models will require billions of dollars in infrastructure investment. The exponential relationship between model capability and computational requirements, wherein each order of magnitude improvement demands roughly two orders of magnitude more compute, creates mounting barriers to entry.<a href=\"#_edn36\" title=\"\" id=\"_ednref36\">[36]<\/a><\/p>\n<p class=\"Caption\">Figure 5: R&amp;D and infrastructure costs for frontier technologies<a href=\"#_edn37\" title=\"\" id=\"_ednref37\">[37]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image005.png\" width=\"768\" height=\"576\" alt=\"image\"\/><\/p>\n<p class=\"Normal\">Only organizations with exceptional resources can develop AI models that rely on dedicated access to hundreds of thousands of specialized processors. When OpenAI needed to scale GPT-4 training, it secured resources through an expanded partnership with Microsoft, which provided both funding and exclusive access to Azure infrastructure.<a href=\"#_edn38\" title=\"\" id=\"_ednref38\">[38]<\/a> Smaller organizations, regardless of technical talent, cannot access comparable computational resources because the cost is simply too great.<\/p>\n<p class=\"Normal\">Autonomous vehicle development presents similar scale requirements. Alphabet\u2019s Waymo has invested over $5 billion since its inception, funding not just research but also extensive infrastructure.<a href=\"#_edn39\" title=\"\" id=\"_ednref39\">[39]<\/a> The program requires custom sensor development, high-definition mapping of entire metropolitan areas, sophisticated simulation environments processing millions of virtual miles, and real-world testing fleets. Each component demands specialized expertise and sustained funding that few organizations can provide.<\/p>\n<p class=\"Normal\">Quantum computing pushes scale requirements even further. Alphabet\u2019s 2019 achievement of quantum supremacy marked the culmination of 13 years of sustained investment.<a href=\"#_edn40\" title=\"\" id=\"_ednref40\">[40]<\/a> The program required custom fabrication facilities for quantum processors, cryogenic systems maintaining temperatures near absolute zero, and interdisciplinary teams spanning physics, computer science, and engineering.<a href=\"#_edn41\" title=\"\" id=\"_ednref41\">[41]<\/a> Amazon, IBM, and Microsoft are pursuing similar programs, each investing billions without near-term revenue prospects.<a href=\"#_edn42\" title=\"\" id=\"_ednref42\">[42]<\/a><\/p>\n<p class=\"Normal\">These examples illustrate how technological complexity drives capital concentration. As innovations require integration across multiple disciplines, longer development cycles, and more sophisticated infrastructure, the minimum viable investment rises correspondingly. This creates a natural filter where only organizations with exceptional financial capacity can pursue frontier technologies.<\/p>\n<p>Risk-Taking That Drives Breakthrough Innovation<\/p>\n<p class=\"Normal\">Innovation inherently involves exploring multiple technical paths with uncertain outcomes. Historical analysis of major technologies reveals consistent patterns: numerous failed approaches preceded each breakthrough, success often emerged from unexpected directions, and the ultimate applications differed from initial visions.<a href=\"#_edn43\" title=\"\" id=\"_ednref43\">[43]<\/a> Effective innovation systems must therefore support portfolio approaches, pursuing multiple competing solutions simultaneously while maintaining tolerance for failure.<\/p>\n<p class=\"Normal\">Business scholar Clayton Christensen and others long criticized U.S. corporations for focusing on what he termed incremental, rather than breakthrough, innovation.<a href=\"#_edn44\" title=\"\" id=\"_ednref44\">[44]<\/a> The former is making small, gradual improvements to existing products for current customers rather than pursuing transformative technologies. And he was right. Too many companies became too risk-averse, losing market share to foreign companies often backed by the deep pockets of the state.<\/p>\n<p class=\"Normal\">Large technology companies have focused significantly on breakthrough innovation and incorporated significant risk-taking practices, which are difficult for traditional organizations to match. Many of these projects fail and never make it to full commercialization, but at least these companies took the risk. Other projects do succeed. Microsoft has pursued topological quantum computing since 2005, investing billions of dollars in an unproven approach that could revolutionize the field or prove impossible.<a href=\"#_edn45\" title=\"\" id=\"_ednref45\">[45]<\/a> Meta funds neural interface research through its Reality Labs, exploring electromyography (EMG) wristband technology for human-computer interaction despite long development horizons.<a href=\"#_edn46\" title=\"\" id=\"_ednref46\">[46]<\/a> Microsoft has committed to purchasing electricity from Helion Power\u2019s fusion reactors starting in 2028, although no company has yet successfully produced commercial fusion power.<a href=\"#_edn47\" title=\"\" id=\"_ednref47\">[47]<\/a> This tolerance for extreme technical risk enables the pursuit of transformative technologies wherein theoretical feasibility remains unproven.<a href=\"#_edn48\" title=\"\" id=\"_ednref48\">[48]<\/a><\/p>\n<p class=\"Normal\">The portfolio strategy extends beyond simple failure tolerance. These companies pursue multiple competing technical approaches simultaneously, hedging against technical uncertainty. In AI, major firms maintain investments across different model architectures, training approaches, and application domains rather than commit to single paths. Microsoft invests in both its own AI research and the OpenAI partnership. Alphabet both develops Tensor Processing Units (TPUs) and maintains GPU infrastructure.<a href=\"#_edn49\" title=\"\" id=\"_ednref49\">[49]<\/a> This parallel exploration increases the probability that at least one approach succeeds.<\/p>\n<p class=\"Normal\">Nvidia exemplifies this strategy through its multi-track chip development, which runs three design teams simultaneously on staggered 18- to 24-month cycles, allowing for continuous innovation while hedging against the failure of any single architecture.<a href=\"#_edn50\" title=\"\" id=\"_ednref50\">[50]<\/a> The company invested nearly $10 billion developing its B200 Blackwell chip alone, while maintaining parallel investments in its software ecosystem: over 900 CUDA (Compute Unified Device Architecture) libraries that ensure its hardware advantages translate into practical applications.<a href=\"#_edn51\" title=\"\" id=\"_ednref51\">[51]<\/a><\/p>\n<p class=\"Normal\">Financial resilience enables countercyclical R&amp;D investment, increasing research spending during economic downturns when competitors retreat. Academic research demonstrates that companies that maintain R&amp;D investment through recessions emerge with significant advantages during recovery.<a href=\"#_edn52\" title=\"\" id=\"_ednref52\">[52]<\/a> While smaller firms must cut research spending in order to survive downturns, and larger, more traditional firms cut to keep their share price up, large technology companies can accelerate investment when resources (talent, equipment, facilities) become more available and affordable.<\/p>\n<p class=\"Normal\">The risk capacity manifests in the willingness to pursue technologies with extreme uncertainty. Google Brain began as a speculative project with unclear commercial applications.<a href=\"#_edn53\" title=\"\" id=\"_ednref53\">[53]<\/a> The team explored whether deep learning techniques, largely abandoned by the AI community, could achieve breakthrough results with sufficient scale. The project consumed massive computational resources with no guarantee of success. Today, the technologies developed generate over $10 billion annually through search improvements and cloud services.<\/p>\n<p>Competing With State-Backed Chinese Competitors<\/p>\n<p class=\"Normal\">China\u2019s innovation model poses a systematic challenge to market-based systems through the coordinated use of state resources. The Chinese government allocated over $450 billion to R&amp;D in 2023, representing an 8.1 percent increase aimed at sustaining growth above 7 percent annually.<a href=\"#_edn54\" title=\"\" id=\"_ednref54\">[54]<\/a> This spending operates within an integrated system in which state-owned enterprises, research institutes, and nominally private companies align with national objectives. Recent initiatives include a $138 billion fund for quantum computing, AI, and semiconductors.<a href=\"#_edn55\" title=\"\" id=\"_ednref55\">[55]<\/a><\/p>\n<p class=\"Normal\">These state subsidies constitute more than aggressive competition; they violate fundamental principles of international trade. As ITIF has documented extensively, China\u2019s systematic use of below-market financing, production subsidies, and state-directed overcapacity represents predatory innovation mercantilism designed to capture entire industries rather than compete on merit.<a href=\"#_edn56\" title=\"\" id=\"_ednref56\">[56]<\/a> These strategies violate World Trade Organization (WTO) subsidy rules, antidumping principles, and the basic reciprocity underlying global trade.<a href=\"#_edn57\" title=\"\" id=\"_ednref57\">[57]<\/a><\/p>\n<p class=\"Callout\">Against China\u2019s coordinated state-capital model, traditional market competition proves insufficient. Startups and mid-sized companies lack the resources to compete with entities backed by sovereign wealth and fight decade-long price wars.<\/p>\n<p class=\"Normal\">The state-directed approach demonstrates effectiveness through market dominance in targeted sectors. Chinese solar manufacturers control over 80 percent of global production after sustained below-cost pricing drove out international competitors.<a href=\"#_edn58\" title=\"\" id=\"_ednref58\">[58]<\/a> Government subsidies enabled this strategy, accepting losses that would bankrupt private firms. In telecommunications, state backing allowed Huawei to underbid Western competitors by 30\u201340 percent while investing more in R&amp;D than did Nokia and Ericsson combined.<a href=\"#_edn59\" title=\"\" id=\"_ednref59\">[59]<\/a> The pattern repeats across strategic industries from batteries to drones.<a href=\"#_edn60\" title=\"\" id=\"_ednref60\">[60]<\/a><\/p>\n<p class=\"Normal\">Against this coordinated model, traditional market competition proves insufficient. Startups and mid-sized companies lack the resources to compete with entities backed by sovereign wealth and fight decade-long price wars.<a href=\"#_edn61\" title=\"\" id=\"_ednref61\">[61]<\/a> Government research grants, while valuable, operate at scales that are inadequate and inconsistent for competing with national industrial strategies.<a href=\"#_edn62\" title=\"\" id=\"_ednref62\">[62]<\/a> The structural mismatch between dispersed market actors and coordinated state capital threatens technological leadership.<\/p>\n<p class=\"Normal\">American technology companies provide the essential counterweight through concentrated private resources. The $227 billion in R&amp;D spending by five firms approaches Chinese state investment while maintaining market discipline.<a href=\"#_edn63\" title=\"\" id=\"_ednref63\">[63]<\/a> These companies possess the scale to sustain long-term competition, absorb temporary losses, and invest in next-generation technologies without government direction.<\/p>\n<p class=\"Normal\">The competition extends beyond financial resources to innovation models. While China excels at scaling proven technologies through state support, U.S. corporate laboratories maintain advantages in breakthrough innovation. Private companies alone cannot counter China\u2019s state-coordinated approach; the United States requires more comprehensive policy responses, from strategic industrial support to coordinated trade enforcement. Yet, large technology companies provide critical leverage in this competition.<\/p>\n<p class=\"Normal\">Another complaint levied against U.S. corporations is that too much of their research focuses on the development side of the continuum rather than on earlier-stage research. In fact, leading American tech companies spend considerable amounts on basic research and openly publish their research findings. They also release tools freely available to developers and provide computing infrastructure to universities and independent researchers. This openness creates knowledge spillovers and enables research capabilities across the broader innovation ecosystem.<\/p>\n<p>Open Research Model<\/p>\n<p class=\"Normal\">Open research speeds up innovation. When companies publish their findings, other researchers can build on that work immediately instead of starting from scratch. This creates faster progress across the entire field. The benefits extend throughout the economy\u2014startups can use published AI models without developing their own, universities can access datasets they couldn\u2019t afford to create, and engineers can adopt proven methods rather than reinvent solutions. Economic analyses show that open data generates 0.5 percent more gross domestic product (GDP) annually compared with data requiring payment, with global economic impact estimates ranging from $3 trillion to $5 trillion per year.<a href=\"#_edn64\" title=\"\" id=\"_ednref64\">[64]<\/a><\/p>\n<p class=\"Caption\">Figure 6: Big tech vs other companies\u2019 contributions to AI advancement<a href=\"#_edn65\" title=\"\" id=\"_ednref65\">[65]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image006.png\" width=\"768\" height=\"384\" alt=\"image\"\/><\/p>\n<p class=\"Caption\">Figure 7: U.S. companies\u2019 AI models by training compute (FLOPS, log scale)<a href=\"#_edn66\" title=\"\" id=\"_ednref66\">[66]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image007.png\" width=\"768\" height=\"480\" alt=\"image\"\/><\/p>\n<p class=\"Caption\">Figure 8: Top 10 critical tech patent contributors (excluding life sciences)<a href=\"#_edn67\" title=\"\" id=\"_ednref67\">[67]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image008.png\" width=\"768\" height=\"480\" alt=\"image\"\/><\/p>\n<p class=\"Caption\">Figure 9: Top 10 patent contributors to U.S. defense tech<a href=\"#_edn68\" title=\"\" id=\"_ednref68\">[68]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image009.png\" width=\"768\" height=\"480\" alt=\"image\"\/><\/p>\n<p class=\"Normal\">NeurIPS, ICML, and ICLR are the top three influential AI conferences based on Google Scholar metrics, with acceptance rates usually between 20 and 25 percent.<a href=\"#_edn69\" title=\"\" id=\"_ednref69\">[69]<\/a> At these top-tier conferences, Alphabet released over 500 research papers in 2023, with DeepMind researchers presenting more than 80 papers at ICML 2023 and Microsoft submitting over 100 accepted papers at NeurIPS 2024.<a href=\"#_edn70\" title=\"\" id=\"_ednref70\">[70]<\/a> Industry research produced 51 notable machine learning models in 2023, compared with 15 from academia.<a href=\"#_edn71\" title=\"\" id=\"_ednref71\">[71]<\/a> This volume establishes tech companies as the largest contributors to top-tier AI conferences alongside leading universities.<\/p>\n<p class=\"Callout\">At top-tier AI conferences, industry research produced 51 notable machine learning models in 2023, compared with 15 from academia.<\/p>\n<p class=\"Normal\">The real economic impact comes through open source infrastructure. TensorFlow and PyTorch, developed by Alphabet and Meta, respectively, are software frameworks that enable developers to build and deploy AI applications<a name=\"OLE_LINK3\" id=\"OLE_LINK3\">.<\/a><a href=\"#_edn72\" title=\"\" id=\"_ednref72\">[72]<\/a> These frameworks provide the essential building blocks, from basic mathematical operations to complex neural network architectures, that developers need to create AI systems. Without them, each organization would need to write thousands of lines of foundational code before even beginning its actual AI work. TensorFlow commands 37 percent market share and PyTorch 25 percent among machine learning tools.<a href=\"#_edn73\" title=\"\" id=\"_ednref73\">[73]<\/a> PyTorch\u2019s adoption by 80 percent of research papers at major conferences demonstrates how quickly open tools can become industry standards.<a href=\"#_edn74\" title=\"\" id=\"_ednref74\">[74]<\/a> Developing a production-grade machine learning framework requires substantial investment in terms of engineering teams, infrastructure, and years of refinement. By providing these tools freely, big tech companies eliminate redundant development efforts across thousands of organizations.<\/p>\n<p class=\"Caption\">Figure 10: Big tech contributions to critical open source infrastructure (indexed scores by CSET)<a href=\"#_edn75\" title=\"\" id=\"_ednref75\">[75]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image010.png\" width=\"768\" height=\"576\" alt=\"image\"\/><\/p>\n<p class=\"Normal\">AlphaFold is a prime example of the benefits of open research. Google DeepMind\u2019s protein structure prediction system, cited over 20,000 times since release, accelerates research across pharmaceuticals, agriculture, and materials science.<a href=\"#_edn76\" title=\"\" id=\"_ednref76\">[76]<\/a> The Drugs for Neglected Diseases Initiative uses it to develop treatments for diseases that affect billions but generate little profit.<a href=\"#_edn77\" title=\"\" id=\"_ednref77\">[77]<\/a> Academic labs apply it to antibiotic resistance and cancer research.<a href=\"#_edn78\" title=\"\" id=\"_ednref78\">[78]<\/a> Industrial researchers use it to design new enzymes and materials.<a href=\"#_edn79\" title=\"\" id=\"_ednref79\">[79]<\/a> This single open release may accelerate hundreds of drug development programs by years.<\/p>\n<p class=\"Normal\">Big tech companies naturally maintain significant proprietary research operations because their core products and competitive advantages depend on it. Yet, compared with traditional corporate R&amp;D models, wherein virtually all research remains internal, big tech has normalized publishing academic research and releasing foundational tools while competing fiercely in product markets. This balance distinguishes their contribution to the innovation landscape.<\/p>\n<p>Talent Multiplier Effects<a id=\"_Toc206356662\"\/><\/p>\n<p class=\"Normal\">Advances in AI, semiconductors, and complex software require large teams of specialized researchers and engineers. Companies and countries compete intensely for this talent. Success depends not just on recruiting individuals but also on building talent clusters in which people can move between organizations, share knowledge, and start new ventures.<\/p>\n<p class=\"Normal\">These clusters extend far beyond Silicon Valley. In Pittsburgh, Pennsylvania, Carnegie Mellon has attracted research labs and offices from prominent companies such as Apple, Intel, Alphabet, Microsoft, Disney, Facebook, IBM, and others to its campus, creating a hub that city officials describe as \u201cthe nation\u2019s most concentrated AI hub outside of Silicon Valley.\u201d<a href=\"#_edn80\" title=\"\" id=\"_ednref80\">[80]<\/a> In Chicago, the University of Illinois Systems Discovery Partners Institute runs programs that include partnerships with AWS, Alphabet, Microsoft, and dozens of other tech companies, training nearly 30 students from universities including University of Illinois Urbana-Champaign, University of Illinois Chicago, and City Colleges of Chicago.<a href=\"#_edn81\" title=\"\" id=\"_ednref81\">[81]<\/a>These geographic expansions demonstrate how big tech companies anchor innovation ecosystems nationwide, training talent who later work across the economy.<\/p>\n<p class=\"Normal\">Amazon\u2019s robotics investments further demonstrate this ecosystem-building effect. The company employs over 750,000 robots across its operations and has hired thousands of robotics engineers, creating a critical mass of automation talent.<a href=\"#_edn82\" title=\"\" id=\"_ednref82\">[82]<\/a> These engineers develop expertise that later diffuses throughout the economy as they move to startups and other industries, helping establish U.S. leadership in industrial automation.<a href=\"#_edn83\" title=\"\" id=\"_ednref83\">[83]<\/a><\/p>\n<p class=\"Normal\">Big tech\u2019s role in talent development extends well beyond its direct employees. Companies invest heavily in K-12 and university STEM education programs that build the broader talent pipeline. Microsoft\u2019s TEALS (Technology Education and Literacy in Schools) program exemplifies this commitment.<a href=\"#_edn84\" title=\"\" id=\"_ednref84\">[84]<\/a> It pairs over 1,650 volunteers from more than 700 companies with high school teachers to co-teach computer science, operating in 31 states and serving 17,000 students annually. <a href=\"#_edn85\" title=\"\" id=\"_ednref85\">[85]<\/a> Microsoft Research\u2019s fellowship program has supported over 700 Ph.D. students globally.<a href=\"#_edn86\" title=\"\" id=\"_ednref86\">[86]<\/a> Thousands more cycle through internships, postdoctoral positions, and visiting researcher programs annually. These positions provide not just funding but also access to mentorship, datasets, and infrastructure unavailable in academia.<\/p>\n<p class=\"Normal\">The Nvidia-University of Florida (UF) partnership demonstrates how industry investments create ecosystem-wide benefits. Through $25 million in hardware, software, and training support, Nvidia enabled UF to share AI resources with all 12 Florida state universities.<a href=\"#_edn87\" title=\"\" id=\"_ednref87\">[87]<\/a> The partnership trained over 800 students through certified instructors and extended advanced computing access to institutions traditionally excluded from such resources.<a href=\"#_edn88\" title=\"\" id=\"_ednref88\">[88]<\/a><\/p>\n<p class=\"Normal\">The entrepreneurial impact is also substantial. Former Alphabet employees founded 14 of the top 50 AI startups that have collectively raised $14.7 billion at a combined valuation of $71.61 billion as of October 2024.<a href=\"#_edn89\" title=\"\" id=\"_ednref89\">[89]<\/a> The broader ecosystem extends far beyond AI\u20142,300 companies founded by Google alumni have raised $98.2 billion across all sectors, while Microsoft alumni have created 5,923 companies, raising $69 billion in total.<a href=\"#_edn90\" title=\"\" id=\"_ednref90\">[90]<\/a> Meta demonstrates the highest funding success rate at 62.9 percent despite fewer total startups, with alumni founding critical AI companies including OpenAI and Harvey.<a href=\"#_edn91\" title=\"\" id=\"_ednref91\">[91]<\/a><\/p>\n<p class=\"Normal\">Big tech companies actively support this talent circulation through structured programs that maintain connections with the entrepreneurial ecosystem. Google\u2019s AI Futures Fund provides equity investment and early access to DeepMind models, Microsoft for Startups offers up to $100,000 in Azure credits with technical mentorship through its GrowthX Accelerator, and Meta\u2019s Llama Startup Program provides $6,000 monthly in cloud inference costs.<a href=\"#_edn92\" title=\"\" id=\"_ednref92\">[92]<\/a><\/p>\n<p class=\"Normal\">Leading American tech companies build hyperscale data centers, submarine cables, and renewable energy infrastructure. This infrastructure strengthens U.S. economic competitiveness, enhances national security capabilities, and creates positive spillovers for other industries.<\/p>\n<p>Private Capital Fills Critical Infrastructure Gaps<\/p>\n<p class=\"Normal\">Digital infrastructure has become as essential to the modern economy as roads and electricity, yet the scale of investment required exceeds what governments can provide alone. While the benefits of digital infrastructure are clear, the U.S. government lacks the necessary funds to lead.<a href=\"#_edn93\" title=\"\" id=\"_ednref93\">[93]<\/a> The federal government invests $208 billion annually in R&amp;D, but this funding primarily supports basic research rather than the massive capital deployments needed for data centers, subsea cables, and computational infrastructure.<a href=\"#_edn94\" title=\"\" id=\"_ednref94\">[94]<\/a> A report from the Center for Strategic and International Studies emphasizes that \u201ccompanies\u2019 participation and collaboration with governments are essential,\u201d particularly where \u201cgovernments alone often cannot provide sufficient technological advancement.\u201d<a href=\"#_edn95\" title=\"\" id=\"_ednref95\">[95]<\/a><\/p>\n<p class=\"Callout\">Digital infrastructure has become as essential to the modern economy as roads and electricity, yet the scale of investment required exceeds what governments can provide alone.<\/p>\n<p class=\"Normal\">The U.S. technology sector now invests over $300 billion annually in AI infrastructure alone, filling gaps that traditional government programs cannot address due to budget constraints, procurement delays, and the rapid pace of technological change.<a href=\"#_edn96\" title=\"\" id=\"_ednref96\">[96]<\/a> Amazon committed $150 billion to data center expansion over 15 years, exceeding the federal government\u2019s entire annual R&amp;D budget.<a href=\"#_edn97\" title=\"\" id=\"_ednref97\">[97]<\/a> Alphabet owns or leases stakes in at least 16 subsea cables, creating redundant data pathways that enhance both commercial efficiency and national security.<a href=\"#_edn98\" title=\"\" id=\"_ednref98\">[98]<\/a> These investments represent strategic bets on unproven technologies rather than conventional capital expenditures.<\/p>\n<p class=\"Normal\">Modern digital infrastructure requires technical capabilities beyond traditional engineering. Hyperscale data centers need breakthrough innovations in cooling, power management, and network architecture. When commercial semiconductors proved inadequate for AI workloads, companies developed custom silicon\u2014Alphabet\u2019s Tensor Processing Units, Amazon\u2019s Graviton processors, and Microsoft\u2019s Azure Maia chips\u2014investing billions without guaranteed returns. Meta built AI training clusters with 24,576 H100 GPUs, computational resources that exceed most government capabilities.<a href=\"#_edn99\" title=\"\" id=\"_ednref99\">[99]<\/a><\/p>\n<p class=\"Normal\">Technology companies pursue infrastructure investments in emerging technologies with no clear path to profitability. Alphabet achieved quantum supremacy in 2019, following a decade of investment that yielded no immediate commercial applications. AWS provides access to quantum computing through Braket, subsidizing million-dollar systems for researchers. Microsoft experiments with underwater data centers and novel cooling technologies that may take decades to yield a return. These long-term bets on frontier technologies would fail traditional investment criteria, but prove essential for maintaining technological leadership.<\/p>\n<p>Infrastructure Ownership as Geopolitical Leverage<\/p>\n<p class=\"Normal\">Digital infrastructure control translates directly into geopolitical influence. When nations and businesses integrate their operations with global technology platforms, they align with the infrastructure owners\u2019 technical standards, governance models, and operational practices. These relationships deepen over time as economies build innovative capabilities on these platforms, creating technical interconnections and shared interests that strengthen international partnerships. Infrastructure ownership thus generates influence through technological integration.<\/p>\n<p class=\"Normal\">American cloud providers dominate with 62\u201367 percent of global cloud infrastructure, enabling them to establish technical standards through market adoption rather than regulatory mandates.<a href=\"#_edn100\" title=\"\" id=\"_ednref100\">[100]<\/a> When AWS implements security requirements for 90 percent of Fortune 500 companies, these protocols become global best practices, as companies align their operations with the platform\u2019s capabilities.<a href=\"#_edn101\" title=\"\" id=\"_ednref101\">[101]<\/a> Singapore\u2019s Smart Nation initiative leverages American cloud platforms, with government agencies using AWS, Microsoft Azure, and Google Cloud to deliver advanced urban services and digital governance.<a href=\"#_edn102\" title=\"\" id=\"_ednref102\">[102]<\/a> Estonia has built its pioneering e-government system through these platforms, enabling everything from digital identity verification to seamless tax collection.<a href=\"#_edn103\" title=\"\" id=\"_ednref103\">[103]<\/a><\/p>\n<p class=\"Callout\">American cloud providers dominate with 62\u201367 percent of global cloud infrastructure, enabling them to establish technical standards through market adoption rather than regulatory mandates.<\/p>\n<p class=\"Normal\">Platform capabilities enable coordinated responses during geopolitical crises through commercial mechanisms. Following Russia\u2019s 2022 invasion of Ukraine, Microsoft, Google Cloud, and AWS restricted services to sanctioned entities while ensuring continued support for Ukrainian agencies and their allies.<a href=\"#_edn104\" title=\"\" id=\"_ednref104\">[104]<\/a> These actions required no government infrastructure or military resources. Private companies enforced terms of service that align with international sanctions, demonstrating how commercial decisions support diplomatic objectives.<\/p>\n<p class=\"Normal\">The business logic driving platform expansion creates these strategic advantages through commercial competition. Cloud providers are pursuing international markets to capture projected spending of $679 billion in 2024, necessitating a global presence to achieve economies of scale.<a href=\"#_edn105\" title=\"\" id=\"_ednref105\">[105]<\/a> Amazon plans $100 billion in capital expenditure for 2025, primarily for AWS infrastructure that will serve both commercial and government customers worldwide.<a href=\"#_edn106\" title=\"\" id=\"_ednref106\">[106]<\/a> Competition for enterprise clients drives continuous security improvements that benefit all users, while platform companies maintain high standards to protect brand reputation. These commercial imperatives consistently produce outcomes that align with democratic governance principles and strengthen America\u2019s partnerships with allies.<\/p>\n<p class=\"Caption\">Figure 11: GPU cluster market share<a href=\"#_edn107\" title=\"\" id=\"_ednref107\">[107]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image011.png\" width=\"768\" height=\"480\" alt=\"image\"\/><\/p>\n<p>Strategic Competition for Global Digital Infrastructure<\/p>\n<p class=\"Normal\">China\u2019s aggressive expansion of its digital infrastructure threatens American technological leadership and access to emerging markets. Beijing has weaponized infrastructure investment to create exclusive economic zones where American companies cannot compete\u2014when Huawei builds a nation\u2019s 4G network, U.S. firms lose that market for mobile services, cloud computing, and digital payments. This infrastructure competition determines whether American companies can access the 3 billion users coming online in developing markets and whether U.S. technical standards will govern the global digital economy. Losing this competition means ceding control of the world\u2019s fastest-growing economies to China.<\/p>\n<p class=\"Normal\">China has invested heavily through its Digital Silk Road initiative, with $79 billion deployed by 2018 and continued significant investments since.<a href=\"#_edn108\" title=\"\" id=\"_ednref108\">[108]<\/a> Huawei controls 70 percent of Africa\u2019s 4G networks, creating Chinese-dominated markets across the continent.<a href=\"#_edn109\" title=\"\" id=\"_ednref109\">[109]<\/a> These deployments require exclusive Chinese suppliers and mandate routing data to be processed through Beijing\u2019s systems. Zimbabwe agreed to share facial data with CloudWalk to help train its AI systems.<a href=\"#_edn110\" title=\"\" id=\"_ednref110\">[110]<\/a> Each Chinese infrastructure win locks out American companies for decades\u2014once a country builds on Huawei equipment, switching costs make changing to U.S. systems economically impossible.<\/p>\n<p class=\"Normal\">American companies counter through commercial innovation that keeps markets open. Alphabet\u2019s $1 billion Equiano cable provides 20 times more West African capacity, creating $7 billion in GDP growth that American businesses can capture.<a href=\"#_edn111\" title=\"\" id=\"_ednref111\">[111]<\/a> AWS and Microsoft cloud services let countries build digital economies that remain accessible to U.S. firms.<a href=\"#_edn112\" title=\"\" id=\"_ednref112\">[112]<\/a> Meta\u2019s Llama and Google\u2019s TensorFlow establish American AI standards globally through voluntary adoption by millions of developers.<a href=\"#_edn113\" title=\"\" id=\"_ednref113\">[113]<\/a> Meta\u2019s expansion of Llama access to NATO, EU institutions, and key allied nations for defense applications creates a shared AI infrastructure that excludes China while strengthening democratic technological cooperation.<a href=\"#_edn114\" title=\"\" id=\"_ednref114\">[114]<\/a> These investments ensure that American companies can compete in emerging markets rather than being locked out by Chinese infrastructure monopolies.<\/p>\n<p class=\"Normal\">Business incentives drive American success more effectively than government programs do. Companies invest to capture growth in markets such as Africa, which is projected to reach $712 billion by 2050.<a href=\"#_edn115\" title=\"\" id=\"_ednref115\">[115]<\/a> Commercial competition pushes firms to offer superior technology and transparent terms. Open sourcing creates global developer communities trained on American platforms. While companies pursue profits, their infrastructure investments secure U.S. economic interests by keeping emerging markets open to American business and preventing Chinese monopolization of the digital future.<\/p>\n<p class=\"Normal\">U.S. technology companies invest over $200 billion annually in R&amp;D for AI, quantum computing, and advanced semiconductors\u2014technologies that define modern military advantage.<a href=\"#_edn116\" title=\"\" id=\"_ednref116\">[116]<\/a> This commercial investment creates a dual-use dividend: the Pentagon gains access to capabilities developed through market competition rather than government contracts alone, while companies expand their innovations into defense applications. The result strengthens both U.S. economic competitiveness and national security without requiring separate development tracks.<\/p>\n<p>The Commercial Technology Foundation for Defense Innovation<\/p>\n<p class=\"Normal\">AI enables military systems to process vast amounts of data and identify threats more quickly than humans ever could. AI algorithms can scan satellite imagery covering thousands of square miles to detect military vehicles, analyze patterns across millions of intercepted communications to predict attacks, and identify potential equipment failures before they occur.<a href=\"#_edn117\" title=\"\" id=\"_ednref117\">[117]<\/a> In practice, this means U.S. forces can detect Chinese naval movements across the Pacific, track terrorist networks through digital signatures, and maintain aircraft with unprecedented efficiency. The same AI capabilities drive drug discovery for biodefense and optimize military supply chains.<\/p>\n<p class=\"Normal\">Quantum computing poses a threat to break the encryption protecting military communications, financial systems, and critical infrastructure. Current encryption relies on mathematical problems that would take classical computers billions of years to solve. Quantum computers could crack these codes in hours by evaluating multiple solutions simultaneously.<a href=\"#_edn118\" title=\"\" id=\"_ednref118\">[118]<\/a> China has invested $15 billion in quantum research and claims quantum advances that could challenge U.S. technological superiority.<a href=\"#_edn119\" title=\"\" id=\"_ednref119\">[119]<\/a> The United States allocated $1.2 billion through the National Quantum Initiative Act of 2018, but commercial development by IBM, Alphabet, and others provides the bulk of American quantum capabilities.<a href=\"#_edn120\" title=\"\" id=\"_ednref120\">[120]<\/a> Beyond code breaking, quantum computers will design new materials for hypersonic vehicles, optimize logistics across global supply chains, and enable quantum encryption that cannot be broken even by other quantum computers.<\/p>\n<p class=\"Normal\">While the United States lost semiconductor manufacturing leadership to Asia decades ago, American technology companies maintain dominance in chip design\u2014the high-value intellectual property that drives innovation.<a href=\"#_edn121\" title=\"\" id=\"_ednref121\">[121]<\/a> Companies such as Nvidia, Apple, and AMD invest billions in developing architectures that advance computing capabilities, creating innovations the military adapts for defense applications.<a href=\"#_edn122\" title=\"\" id=\"_ednref122\">[122]<\/a><\/p>\n<p class=\"Normal\">American technology companies drive advancement in all three domains through commercial R&amp;D.<a href=\"#_edn123\" title=\"\" id=\"_ednref123\">[123]<\/a> The military adapts these commercially developed technologies rather than funding them from scratch.<a href=\"#_edn124\" title=\"\" id=\"_ednref124\">[124]<\/a> This arrangement benefits both sides: companies gain defense contracts and technical challenges that push innovation boundaries, while the military accesses capabilities that defense budgets alone could never create.<a href=\"#_edn125\" title=\"\" id=\"_ednref125\">[125]<\/a> Unlike China\u2019s military-civil fusion strategy that mandates technology transfer from companies to military institutions, the United States relies on voluntary commercial innovation in which technology companies\u2019 market-driven R&amp;D creates capabilities the military can adapt.<a href=\"#_edn126\" title=\"\" id=\"_ednref126\">[126]<\/a><\/p>\n<p class=\"Caption\">Table 1: Major big tech defense contracts<a href=\"#_edn127\" title=\"\" id=\"_ednref127\">[127]<\/a><\/p>\n<p class=\"TableHeadingBold\" align=\"left\">Company<\/p>\n<p class=\"TableHeadingBold\" align=\"left\">Contract or Program<\/p>\n<p class=\"TableHeadingBold\" align=\"left\">Value<\/p>\n<p class=\"TableHeadingBold\" align=\"left\">Duration<\/p>\n<p class=\"TableHeadingBold\" align=\"left\">Status<\/p>\n<p class=\"TableHeadingBold\" align=\"left\">Key Technologies<\/p>\n<p class=\"TableHeadingBold\">Microsoft<\/p>\n<p class=\"TableBody\">IVAS (AR Headsets)<\/p>\n<p class=\"TableBody\">$22B<\/p>\n<p class=\"TableBody\">10 years<\/p>\n<p class=\"TableBody\">Transferring to Anduril<\/p>\n<p class=\"TableBody\">HoloLens AR, mixed reality<\/p>\n<p class=\"TableBody\">JWCC (Cloud)<\/p>\n<p class=\"TableBody\">Share of $9B<\/p>\n<p class=\"TableBody\">2022\u20132028<\/p>\n<p class=\"TableBody\">Active<\/p>\n<p class=\"TableBody\">Azure cloud services<\/p>\n<p class=\"TableBody\">CIA C2E (Cloud)<\/p>\n<p class=\"TableBody\">Share of \u201ctens of billions\u201d<\/p>\n<p class=\"TableBody\">15 years<\/p>\n<p class=\"TableBody\">Active<\/p>\n<p class=\"TableBody\">Multi-cloud infrastructure<\/p>\n<p class=\"TableHeadingBold\">Amazon (AWS)<\/p>\n<p class=\"TableBody\">NSA Wild &amp; Stormy<\/p>\n<p class=\"TableBody\">$10B<\/p>\n<p class=\"TableBody\">10 years<\/p>\n<p class=\"TableBody\">Active (sole source)<\/p>\n<p class=\"TableBody\">Top-secret cloud infrastructure<\/p>\n<p class=\"TableBody\">JWCC<\/p>\n<p class=\"TableBody\">Share of $9B<\/p>\n<p class=\"TableBody\">2022\u20132028<\/p>\n<p class=\"TableBody\">Active<\/p>\n<p class=\"TableBody\">Cloud computing<\/p>\n<p class=\"TableBody\">CIA C2E<\/p>\n<p class=\"TableBody\">Share of \u201ctens of billions\u201d<\/p>\n<p class=\"TableBody\">15 years<\/p>\n<p class=\"TableBody\">Active<\/p>\n<p class=\"TableBody\">Intelligence cloud services<\/p>\n<p class=\"TableBody\">CIA C2S (Legacy)<\/p>\n<p class=\"TableBody\">$600M<\/p>\n<p class=\"TableBody\">2013\u20132023<\/p>\n<p class=\"TableBody\">Completed<\/p>\n<p class=\"TableBody\">First major IC cloud contract<\/p>\n<p class=\"TableHeadingBold\">Alphabet<\/p>\n<p class=\"TableBody\">JWCC<\/p>\n<p class=\"TableBody\">Share of $9B<\/p>\n<p class=\"TableBody\">2022\u20132028<\/p>\n<p class=\"TableBody\">Active<\/p>\n<p class=\"TableBody\">Cloud services<\/p>\n<p class=\"TableBody\">CIA C2E<\/p>\n<p class=\"TableBody\">Share of \u201ctens of billions\u201d<\/p>\n<p class=\"TableBody\">15 years<\/p>\n<p class=\"TableBody\">Active<\/p>\n<p class=\"TableBody\">Multi-cloud platform<\/p>\n<p class=\"TableBody\">CDAO AI Contract<\/p>\n<p class=\"TableBody\">$200M<\/p>\n<p class=\"TableBody\">2025\u20132026<\/p>\n<p class=\"TableBody\">Active<\/p>\n<p class=\"TableBody\">Frontier AI capabilities<\/p>\n<p class=\"TableHeadingBold\">Meta<\/p>\n<p class=\"TableBody\">Anduril Partnership<\/p>\n<p class=\"TableBody\">TBD<\/p>\n<p class=\"TableBody\">Development<\/p>\n<p class=\"TableBody\">In negotiation<\/p>\n<p class=\"TableBody\">AR\/VR combat systems<\/p>\n<p class=\"TableBody\">Llama for Defense<\/p>\n<p class=\"TableBody\">Open source<\/p>\n<p class=\"TableBody\">Ongoing<\/p>\n<p class=\"TableBody\">Available<\/p>\n<p class=\"TableBody\">AI language models<\/p>\n<p class=\"TableHeadingBold\">Apple<\/p>\n<p class=\"TableBody\">FlexTech Alliance<\/p>\n<p class=\"TableBody\">Minimal<\/p>\n<p class=\"TableBody\">2015-present<\/p>\n<p class=\"TableBody\">Limited involvement<\/p>\n<p class=\"TableBody\">Flexible electronics R&amp;D<\/p>\n<p>    Military Adaptation of Commercial Platforms<\/p>\n<p class=\"Normal\">The transformation of commercial technologies for military use has evolved from an opportunistic practice to a strategic necessity. Modern military systems increasingly incorporate technologies originally developed for civilian markets, adapted through specialized integration and hardening processes. This approach leverages billions of dollars in commercial R&amp;D while adding military-specific capabilities that traditional defense contractors typically develop.<\/p>\n<p class=\"Normal\">Cloud computing infrastructure exemplifies the successful transfer of commercial technology to the military. The Joint Warfighting Cloud Capability contract, valued at up to $9 billion through 2028, relies exclusively on commercial providers: AWS, Microsoft Azure, Google Cloud, and Oracle Cloud Infrastructure.<a href=\"#_edn128\" title=\"\" id=\"_ednref128\">[128]<\/a> Each vendor brings distinct capabilities shaped by their commercial focus\u2014AWS\u2019s experience hosting the CIA\u2019s classified systems since 2013, Azure\u2019s enterprise integration tools refined through serving Fortune 500 companies, Alphabet\u2019s AI and machine learning services developed for consumer applications, and Oracle\u2019s database technologies proven in commercial financial systems.<a href=\"#_edn129\" title=\"\" id=\"_ednref129\">[129]<\/a> The multivendor approach provides resilience and prevents single-point failures while maintaining competition for task orders.<\/p>\n<p class=\"Normal\">Microsoft\u2019s Integrated Visual Augmentation System demonstrates both the potential and challenges of militarizing consumer technology. Based on the HoloLens 2 augmented reality headset, IVAS aims to provide soldiers with enhanced situational awareness through thermal imaging, navigation aids, and tactical overlays.<a href=\"#_edn130\" title=\"\" id=\"_ednref130\">[130]<\/a> The program\u2019s evolution reflects the iterative nature of commercial-to-military adaptation. Initial versions faced significant challenges\u2014soldiers reported discomfort, reliability issues, and concerns about combat effectiveness. However, the commercial foundation enabled rapid iteration, with version 1.2 incorporating soldier feedback to improve ergonomics and performance.<a href=\"#_edn131\" title=\"\" id=\"_ednref131\">[131]<\/a> Despite ongoing challenges that led to a program restructuring in 2025, the ability to leverage billions in commercial AR investment provided capabilities that purpose-built military development could not have achieved within similar timeframes.<\/p>\n<p class=\"Callout\">Modern military platforms contain thousands of processors manufactured in commercial facilities. For example, the F-35 fighter jet\u2019s systems depend entirely on chips produced alongside commercial electronics.<\/p>\n<p class=\"Normal\">AI represents perhaps the most significant domain of commercial-to-military technology transfer. Project Maven, which applies AI to analyze drone footage and identify targets, has evolved from initial controversy over Alphabet\u2019s participation to a multi-vendor effort involving Alphabet, Microsoft, Amazon, Palantir, and numerous specialized firms.<a href=\"#_edn132\" title=\"\" id=\"_ednref132\">[132]<\/a> These companies adapt computer vision algorithms originally developed for autonomous vehicles, medical imaging, and consumer photography to military intelligence applications. The commercial AI ecosystem provides not just algorithms but entire development frameworks, training methodologies, and computational infrastructure that accelerate military AI deployment.<\/p>\n<p class=\"Normal\">The dependence on commercial semiconductor supply chains underscores how thoroughly military capabilities rely on commercial technology. Modern military platforms contain thousands of processors manufactured in commercial facilities. For example, the F-35 fighter jet\u2019s systems depend entirely on chips produced alongside commercial electronics.<a href=\"#_edn133\" title=\"\" id=\"_ednref133\">[133]<\/a> When military systems require specialized components such as radiation-hardened processors for space applications, commercial foundries modify their standard processes rather than establish dedicated military production lines. This approach provides access to cutting-edge manufacturing capabilities while maintaining economic viability.<\/p>\n<p>Spillovers Into Other Strategic Sectors<\/p>\n<p class=\"Normal\">Avoiding the advice of former GE CEO Jack Welch, large technology companies are willing to engage in businesses in which they are not the dominant one or two players. In fact, they actively pursue research in health care, materials science, transportation, space, agriculture, and climate\u2014fields far from their core businesses. These initiatives leverage computational infrastructure, AI expertise, and data processing capabilities to tackle problems for which these resources can accelerate progress. The pattern reveals companies directing significant R&amp;D investments toward challenges beyond their commercial domains, creating spillover effects that benefit entire industries through both discoveries and open knowledge sharing.<\/p>\n<p>Health Care and Biomedical Research<\/p>\n<p class=\"Normal\">Biological systems involve interactions among thousands of molecules, each with complex three-dimensional structures. Traditional structure determination methods such as X-ray crystallography require extensive time and resources\u2014the Protein Structure Initiative spent $1.4 billion over a decade to determine just 13,000 structures.<a href=\"#_edn134\" title=\"\" id=\"_ednref134\">[134]<\/a> Drug development faces similar constraints, with the average new drug taking 10\u201315 years and $2.6 billion to bring to market, largely because preclinical screening cannot accurately predict human outcomes.<a href=\"#_edn135\" title=\"\" id=\"_ednref135\">[135]<\/a><\/p>\n<p class=\"Normal\">The computational approach changes these economics. AlphaFold predicted structures for 200 million proteins in 18 months\u2014work that would have required centuries of traditional crystallography based on historical productivity rates.<a href=\"#_edn136\" title=\"\" id=\"_ednref136\">[136]<\/a> By making the database freely available, Google DeepMind has enabled researchers worldwide to skip structure determination entirely. Over 500,000 researchers now use the resource, with studies citing AlphaFold appearing in journals at a rate of 17 per day by 2024.<a href=\"#_edn137\" title=\"\" id=\"_ednref137\">[137]<\/a><\/p>\n<p class=\"Callout\">Microsoft\u2019s AI evaluated 32.6 million potential battery materials in 80 hours\u2014screening that would traditionally take decades.<\/p>\n<p class=\"Normal\">Microsoft\u2019s approach to drug discovery addresses the screening bottleneck. Working with Pacific Northwest National Laboratory, Microsoft\u2019s AI evaluated 32.6 million potential battery materials in 80 hours\u2014screening that would traditionally take decades.<a href=\"#_edn138\" title=\"\" id=\"_ednref138\">[138]<\/a> Applied to tuberculosis research, its AI system achieved a 125-fold improvement in identifying effective inhibitors compared with random screening.<a href=\"#_edn139\" title=\"\" id=\"_ednref139\">[139]<\/a> This demonstrates how computational methods can transform trial-and-error processes into targeted searches.<\/p>\n<p class=\"Normal\">Apple\u2019s contribution comes through scale and accessibility. The Apple Heart Study enrolled 400,000 participants in eight months, detecting atrial fibrillation with an 84 percent positive predictive value.<a href=\"#_edn140\" title=\"\" id=\"_ednref140\">[140]<\/a> Published in the New England Journal of Medicine, the study demonstrates how consumer devices enable population-scale medical research.<a href=\"#_edn141\" title=\"\" id=\"_ednref141\">[141]<\/a> Traditional cardiac arrhythmia clinical trials typically enroll hundreds to thousands of participants over multiple years, making Apple\u2019s approach a significant methodological advance.<a href=\"#_edn142\" title=\"\" id=\"_ednref142\">[142]<\/a><\/p>\n<p>Manufacturing and Materials Science<\/p>\n<p class=\"Normal\">Materials discovery has historically proceeded through iterative experimentation. Synthesizing and characterizing a new compound typically takes months, with computational studies suggesting that researchers test thousands of candidates to find one viable material.<a href=\"#_edn143\" title=\"\" id=\"_ednref143\">[143]<\/a> The vast space of possible element combinations means most efforts eliminate nonviable options rather than develop useful materials.<\/p>\n<p class=\"Normal\">Computational modeling bypasses physical synthesis entirely. Google DeepMind\u2019s GNoME discovered 2.2 million crystal structures by simulating atomic interactions, identifying stable configurations without creating them physically.<a href=\"#_edn144\" title=\"\" id=\"_ednref144\">[144]<\/a> The 380,000 stable materials include 528 lithium-ion conductors\u2014critical for battery technology.<a href=\"#_edn145\" title=\"\" id=\"_ednref145\">[145]<\/a> External validation came when laboratories synthesized 736 materials, confirming computational predictions.<a href=\"#_edn146\" title=\"\" id=\"_ednref146\">[146]<\/a> By adding discoveries to the open Materials Project database, the work expanded the searchable space of materials tenfold.<\/p>\n<p class=\"Normal\">Microsoft\u2019s MatterGen takes a different approach, generating materials with specified properties rather than searching existing possibilities.<a href=\"#_edn147\" title=\"\" id=\"_ednref147\">[147]<\/a> Creating structures harder than diamond (bulk modulus exceeding 400 gigapascals) demonstrates that computational design can exceed natural materials. <a href=\"#_edn148\" title=\"\" id=\"_ednref148\">[148]<\/a> Open-sourcing these models democratizes materials discovery\u2014institutions without supercomputers can now participate in materials innovation.<\/p>\n<p>Agriculture and Food Security<\/p>\n<p class=\"Normal\">Agricultural productivity faces information asymmetry: expertise concentrates in research institutions while farming happens in rural areas with limited access to specialists. According to the Food and Agriculture Organization (FAO), plant diseases cause global crop losses of 20\u201340 percent annually, costing the global economy over $220 billion.<a href=\"#_edn149\" title=\"\" id=\"_ednref149\">[149]<\/a> McKinsey research indicates that precision agriculture technologies can improve yields by 1520 percent through optimized resource use, but deploying sensors and analytics across millions of small farms presents economic challenges.<a href=\"#_edn150\" title=\"\" id=\"_ednref150\">[150]<\/a><\/p>\n<p class=\"Normal\">Smartphones change this equation. Google\u2019s AI detects cassava diseases with 98 percent accuracy using phone cameras\u2014but the innovation isn\u2019t the accuracy; it\u2019s making the model run offline on basic phones.<a href=\"#_edn151\" title=\"\" id=\"_ednref151\">[151]<\/a> TensorFlow Lite versions work without Internet connectivity, crucial for rural deployment.<a href=\"#_edn152\" title=\"\" id=\"_ednref152\">[152]<\/a> Agricultural ministries in Kenya, Tanzania, and Uganda have prevented $60 million in crop losses using these free tools.<a href=\"#_edn153\" title=\"\" id=\"_ednref153\">[153]<\/a><\/p>\n<p class=\"Normal\">Microsoft\u2019s FarmBeats demonstrates how publishing complete technical specifications multiplies impact. The system reduces water usage by 30 percent while maintaining yields, but Microsoft hasn\u2019t commercialize it.<a href=\"#_edn154\" title=\"\" id=\"_ednref154\">[154]<\/a> Instead, publishing sensor designs and data pipelines have enabled agricultural universities to build systems locally. Technology transfer programs have reached 10,000 farms in India and Africa\u2014scale achieved through replication rather than direct deployment.<a href=\"#_edn155\" title=\"\" id=\"_ednref155\">[155]<\/a><\/p>\n<p>Environmental Science and Climate<\/p>\n<p class=\"Normal\">Climate science requires computational infrastructure beyond most institutions\u2019 reach. The European Centre for Medium-Range Weather Forecasts (ECMWF) operates high-performance computing facilities with procurement costs exceeding \u20ac50 million (~$58.7 million) and substantial annual operating expenses.<a href=\"#_edn156\" title=\"\" id=\"_ednref156\">[156]<\/a> Processing satellite data for deforestation tracking requires petabyte-scale storage and computing power. This concentration of resources in wealthy nations creates gaps in climate monitoring for regions most vulnerable to climate change.<\/p>\n<p class=\"Normal\">Google DeepMind\u2019s GraphCast breaks this constraint. The model produces 10-day forecasts that are more accurate than ECMWF but runs on a single machine in under one minute.<a href=\"#_edn157\" title=\"\" id=\"_ednref157\">[157]<\/a> By open sourcing GraphCast, Alphabet enables any institution with a workstation to generate world-class weather predictions.<a href=\"#_edn158\" title=\"\" id=\"_ednref158\">[158]<\/a> For developing nations, this transforms disaster preparedness\u2014accurate cyclone predictions become possible without massive infrastructure investment.<\/p>\n<p class=\"Normal\">Data center energy optimization demonstrates how sharing methodologies creates compound benefits. Alphabet\u2019s AI has reduced cooling requirements by 40 percent, achieving 1.1 Power Usage Effectiveness.<a href=\"#_edn159\" title=\"\" id=\"_ednref159\">[159]<\/a> Publishing detailed methodologies has enabled competitors to achieve 20\u201330 percent reductions.<a href=\"#_edn160\" title=\"\" id=\"_ednref160\">[160]<\/a> With data centers responsible for approximately 1 percent of global electricity consumption according to the International Energy Agency, industry-wide efficiency improvements have a significant climate impact.<a href=\"#_edn161\" title=\"\" id=\"_ednref161\">[161]<\/a><\/p>\n<p>Nuclear Energy<\/p>\n<p class=\"Normal\">Technology companies have emerged as critical catalysts for nuclear energy development, with Alphabet, Microsoft, and Amazon signing major agreements to develop new reactors and restart decommissioned ones.<a href=\"#_edn162\" title=\"\" id=\"_ednref162\">[162]<\/a> Alphabet signed an agreement with Kairos Power to deliver six or seven small modular reactors (SMRs) between 2030 and 2035, Amazon anchored a $500 million investment round in X-energy, and Microsoft\u2019s 20-year power purchase agreement enabled the restart of a reactor at Three Mile Island.<a href=\"#_edn163\" title=\"\" id=\"_ednref163\">[163]<\/a> These investments provide crucial early market demand for SMRs, which, according to ITIF research, are \u201cthe only realistic path to significantly scale nuclear power in the United States\u201d and could become \u201can important strategic export industry\u201d over the next two decades.<a href=\"#_edn164\" title=\"\" id=\"_ednref164\">[164]<\/a><\/p>\n<p class=\"Callout\">Technology companies\u2019 willingness to sign long-term power purchase agreements at premium prices provides the financial certainty needed for SMR manufacturers to achieve production\u00a0scale.<\/p>\n<p class=\"Normal\">The spillover effects extend well beyond powering data centers. SMRs can provide thermal energy for industrial processes at temperatures suitable for chemicals production, petroleum refining, and desalination, creating decarbonization opportunities across multiple industries.<a href=\"#_edn165\" title=\"\" id=\"_ednref165\">[165]<\/a>Technology companies\u2019 willingness to sign long-term power purchase agreements at premium prices provides the financial certainty needed for SMR manufacturers to achieve production scale, while their expertise in simulation and modeling accelerates regulatory approvals. This private sector leadership is particularly important, as China currently has 58 reactors built and 29 under construction, with Russia\u2019s Rosatom expanding into 13 countries, making technology companies\u2019 investments essential for maintaining U.S. competitiveness in advanced nuclear technology.<a href=\"#_edn166\" title=\"\" id=\"_ednref166\">[166]<\/a><\/p>\n<p>Transportation and Urban Mobility<\/p>\n<p class=\"Normal\">Autonomous vehicles face a testing paradox: proving safety requires billions of miles of real-world driving, but deploying vehicles without proof creates liability. No single company can generate sufficient data. Traditional automotive testing\u2014controlled tracks and simulations\u2014doesn\u2019t capture edge cases that cause accidents.<\/p>\n<p class=\"Normal\">Waymo\u2019s decision to share safety methodologies after 71 million autonomous miles has created industry-wide benefits.<a href=\"#_edn167\" title=\"\" id=\"_ednref167\">[167]<\/a> Demonstrating 85 percent fewer injury-causing crashes than human drivers matters less than publishing complete data and methodologies.<a href=\"#_edn168\" title=\"\" id=\"_ednref168\">[168]<\/a> Competitors, including Cruise and Aurora, have adopted these standards, accelerating their development by years. The spillover effect has accelerated industry-wide progress toward reducing traffic fatalities, which claimed 38,824 lives in the United States in 2020, according to NHTSA.<a href=\"#_edn169\" title=\"\" id=\"_ednref169\">[169]<\/a><\/p>\n<p class=\"Normal\">Last-mile delivery represents 28 percent of total transportation costs, according to supply chain analyses, but has resisted optimization due to computational complexity.<a href=\"#_edn170\" title=\"\" id=\"_ednref170\">[170]<\/a> Amazon\u2019s routing algorithms solve problems with millions of variables\u2014which packages go on which trucks in what order to minimize time and fuel.<a href=\"#_edn171\" title=\"\" id=\"_ednref171\">[171]<\/a> Publishing these algorithms has enabled UPS, FedEx, and regional carriers to achieve 15\u201320 percent time reduction and 12 percent fuel savings.<a href=\"#_edn172\" title=\"\" id=\"_ednref172\">[172]<\/a> The aggregate impact across the logistics industry exceeds Amazon\u2019s own efficiency gains.<\/p>\n<p>Quantum Computing Breakthroughs<\/p>\n<p class=\"Normal\">Quantum computing is still in its early stages because building stable quantum systems with low error rates requires solving novel physics problems. The challenge isn\u2019t just engineering but also discovering new states of matter and developing advanced error correction for quantum decoherence.<\/p>\n<p class=\"Normal\">Alphabet\u2019s creation of time crystals\u2014matter in perpetual motion without energy input\u2014represents new physics with implications for quantum memory.<a href=\"#_edn173\" title=\"\" id=\"_ednref173\">[173]<\/a> Publishing complete protocols in nature has enabled replication at Stanford and Oxford, advancing the field collectively.<a href=\"#_edn174\" title=\"\" id=\"_ednref174\">[174]<\/a> Microsoft\u2019s multidecade investment in topological qubits based on Majorana zero modes, shared through academic publications rather than patents, provides alternative approaches to quantum stability.<a href=\"#_edn175\" title=\"\" id=\"_ednref175\">[175]<\/a> These parallel paths increase the probability of achieving fault-tolerant quantum computing, with applications from drug discovery to cryptography.<\/p>\n<p class=\"Normal\">There is a pattern across these contributions: technology companies address computational and data processing bottlenecks that limit progress in other fields. By sharing discoveries and methodologies rather than monopolizing them, they create spillover effects such that the collective benefit exceeds private returns\u2014strengthening the broader innovation ecosystem beyond direct business interests.<\/p>\n<p>Nurturing the Startup Ecosystem<\/p>\n<p class=\"Normal\">The U.S. technology ecosystem thrives on complementary capabilities between startups and established firms. While big tech companies conduct substantial internal R&amp;D, certain types of innovation emerge more naturally from startups unburdened by existing business models and organizational structures. Conversely, breakthrough technologies often require capital intensity and infrastructure scale that only large platforms can provide. The median startup now requires approximately 11 years to reach IPO (Initial Public Offering), up from 5 years in 2000, during which time most need multiple funding rounds and operational support.<a href=\"#_edn176\" title=\"\" id=\"_ednref176\">[176]<\/a> This extended timeline creates opportunities for productive collaboration, wherein startups bring agility and fresh perspectives while large firms provide patient capital, global distribution, and technical infrastructure. The result is an ecosystem wherein strategic acquisitions have become the dominant exit path, with only 11 percent of unicorn exits occurring through IPOs in 2024, down from 83 percent in 2010.<a href=\"#_edn177\" title=\"\" id=\"_ednref177\">[177]<\/a><\/p>\n<p class=\"Caption\">Figure 12: Unicorn exit composition: IPOs vs. acquisitions, 2010\u20132024<a href=\"#_edn178\" title=\"\" id=\"_ednref178\">[178]<\/a><\/p>\n<p class=\"Normal\"><img loading=\"lazy\" decoding=\"async\" id=\"Chart1\" src=\"https:\/\/itif-publications-production.s3.amazonaws.com\/2025-big-tech-contributions_for%20HTML_files\/image012.png\" width=\"768\" height=\"480\" alt=\"image\"\/><\/p>\n<p>Creating Exit Markets and Providing Capital<\/p>\n<p class=\"Normal\">The traditional IPO pathway has constricted dramatically. Annual technology IPOs in the United States declined from a 1980\u20132019 median of 158 to just 72 in 2024, despite venture investment reaching $209 billion that year.<a href=\"#_edn179\" title=\"\" id=\"_ednref179\">[179]<\/a> The median time from first venture funding to IPO exit has extended from 3.8 years in 2000 to over 8 years today.<a href=\"#_edn180\" title=\"\" id=\"_ednref180\">[180]<\/a> This creates a liquidity challenge: with 741 U.S. unicorns currently awaiting exits and only 15 achieving them in the past 12 months, the theoretical backlog would take 49 years to clear at current rates.\u2075 When capital remains trapped in illiquid positions, venture funds cannot return money to limited partners, constraining future investment capacity.<\/p>\n<p class=\"Normal\">Large technology companies have evolved into essential liquidity providers for the venture ecosystem. Alphabet\u2019s portfolio through GV includes over 700 investments, while Microsoft\u2019s M12 has backed more than 100 companies since 2016.<a href=\"#_edn181\" title=\"\" id=\"_ednref181\">[181]<\/a> These corporate venture arms don\u2019t just provide capital\u2014they establish valuation benchmarks that enable subsequent funding rounds. Facebook\u2019s $1 billion Instagram acquisition in 2012, though controversial at the time, validated social media platform valuations.<a href=\"#_edn182\" title=\"\" id=\"_ednref182\">[182]<\/a> Beyond direct investment, infrastructure support programs such as AWS Activate have provided over $7 billion in credits to startups since 2013, with more than 330,000 startups participating globally.<a href=\"#_edn183\" title=\"\" id=\"_ednref183\">[183]<\/a> This enables companies to delay fundraising while proving product-market fit, with some qualifying for up to $100,000 in credits to offset cloud infrastructure costs.<a href=\"#_edn184\" title=\"\" id=\"_ednref184\">[184]<\/a><\/p>\n<p class=\"Normal\">Critics argue that big tech creates \u201ckill zones\u201d wherein venture capitalists avoid funding potential competitors.<a href=\"#_edn185\" title=\"\" id=\"_ednref185\">[185]<\/a> However, empirical evidence challenges this narrative. Research shows that sectors experiencing recent big tech acquisitions actually attract increased venture investment, as successful exits demonstrate achievable returns and validate market opportunities.<a href=\"#_edn186\" title=\"\" id=\"_ednref186\">[186]<\/a> The Instagram acquisition illustrates this dynamic: rather than deterring investment, it has catalyzed increased funding for social platforms by proving that billion-dollar valuations were achievable. Similarly, Google\u2019s YouTube acquisition didn\u2019t eliminate video platform competition but instead validated the market, leading to the emergence of TikTok, Twitch, and other successful platforms. The concern about predatory acquisitions remains valid for specific cases, but aggregate data suggests that acquisition activity generally correlates with increased rather than decreased startup formation.<\/p>\n<p>Rescuing and Scaling Promising Technologies<\/p>\n<p class=\"Normal\">Strategic acquisitions often provide struggling startups with the resources needed to achieve their potential. YouTube faced existential challenges when Google acquired it for $1.65 billion in 2006\u2014the company was burning through $1 million monthly in bandwidth costs while facing copyright lawsuits that threatened its survival.<a href=\"#_edn187\" title=\"\" id=\"_ednref187\">[187]<\/a> Google\u2019s infrastructure and legal resources enabled YouTube to reduce costs dramatically while resolving content disputes, growing into a platform that generated $50 billion in combined advertising and subscription revenue over the four quarters ending Q3 2024.<a href=\"#_edn188\" title=\"\" id=\"_ednref188\">[188]<\/a><\/p>\n<p class=\"Normal\">Android presents an even more dramatic transformation. Google acquired the struggling startup for just $50 million in 2005 when Android was nearly out of funding.<a href=\"#_edn189\" title=\"\" id=\"_ednref189\">[189]<\/a> That investment enabled Android to become the operating system powering 71 percent of global smartphones, creating an app economy worth hundreds of billions annually.<a href=\"#_edn190\" title=\"\" id=\"_ednref190\">[190]<\/a><\/p>\n<p class=\"Normal\">Recent blocked acquisitions demonstrate how regulatory intervention can produce outcomes opposite to intended goals. When European regulators blocked Amazon\u2019s $1.7 billion iRobot acquisition in January 2024 to preserve competition in robotic vacuums, iRobot collapsed within months\u2014eliminating 572 jobs (51 percent of workforce), posting $145.5 million in losses, and forcing the company to explore strategic alternatives, including potential bankruptcy.<a href=\"#_edn191\" title=\"\" id=\"_ednref191\">[191]<\/a> The company that pioneered consumer robotics and held 64 percent market share in 2016 now trades below $20 per share, while Chinese competitors Ecovacs and Roborock have captured the premium segments iRobot created.<a href=\"#_edn192\" title=\"\" id=\"_ednref192\">[192]<\/a><\/p>\n<p class=\"Callout\">When European regulators blocked Amazon\u2019s iRobot acquisition to preserve competition, iRobot collapsed within months\u2014eliminating 572 jobs and exploring bankruptcy.<\/p>\n<p class=\"Normal\">ARM Holdings immediately laid off 1,000 employees (15 percent of the workforce) when regulators blocked Nvidia\u2019s $40 billion acquisition in 2022, abandoning growth investments the merger would have enabled during the AI boom.<a href=\"#_edn193\" title=\"\" id=\"_ednref193\">[193]<\/a> China\u2019s strategic use of merger review demonstrates the geopolitical dimension: blocking Qualcomm\u2019s $44 billion NXP acquisition after eight other jurisdictions approved it, preventing American semiconductor consolidation while Chinese competitors strengthened their position in automotive chips.<a href=\"#_edn194\" title=\"\" id=\"_ednref194\">[194]<\/a><\/p>\n<p class=\"Normal\">Not every acquisition succeeds\u2014Google\u2019s $12.5 billion Motorola purchase resulted in significant losses. However, the documented destruction from blocked deals raises questions about current regulatory frameworks designed for 20th-century industrial markets. Modern technology competition often requires scale for fundamental technical reasons: AI model training demands massive computational resources, semiconductor manufacturing requires billion-dollar fabrication facilities, and global platforms need extensive infrastructure to serve billions of users reliably.<\/p>\n<p class=\"Normal\">The challenge lies in distinguishing genuinely anticompetitive acquisitions from partnerships that enable innovation requiring patient capital and global scale. When promising technologies fail because they lack resources to compete globally, when American semiconductor technology transfers to China through forced divestitures, and when regulatory protection leads to bankruptcy rather than independence, the framework itself requires reconsideration. A more nuanced approach would recognize that while vigilance against anticompetitive conduct remains essential, reflexively blocking consolidation while global competitors advance through strategic integration may ultimately harm the innovation ecosystem regulators seek to protect.<\/p>\n<p class=\"Normal\">\u201cSoft power\u201d proved decisive in America\u2019s Cold War victory, with jazz ambassadors, Hollywood films, and consumer culture winning hearts and minds behind the Iron Curtain more effectively than military might could alone.<a href=\"#_edn195\" title=\"\" id=\"_ednref195\">[195]<\/a> The State Department\u2019s cultural diplomacy programs\u2014from Dizzy Gillespie\u2019s concerts in Eastern Europe to Radio Free Europe\u2019s broadcasts\u2014demonstrated that influence through attraction could transform global alignments without firing a shot.<a href=\"#_edn196\" title=\"\" id=\"_ednref196\">[196]<\/a><\/p>\n<p class=\"Normal\">Today, the United States faces a similar ideological competition with China, yet it has reduced foreign aid spending by 92 percent while China invests heavily in digital infrastructure across the developing world.<a href=\"#_edn197\" title=\"\" id=\"_ednref197\">[197]<\/a> American technology companies, pursuing commercial expansion in emerging markets, now incidentally project U.S. influence in ways that echo Cold War cultural diplomacy. When countries adopt digital platforms, they import more than technology; they adopt the governance models, economic systems, and cultural norms embedded within those platforms. American companies pursuing profits in these markets create dependencies that favor open markets, free expression, and the rule of law\u2014values essential to their business models and also aligned with U.S. strategic interests.<a href=\"#_edn198\" title=\"\" id=\"_ednref198\">[198]<\/a> This commercial expansion generates two primary forms of soft power: cultural influence through platforms that shape global discourse and development outcomes that would otherwise require government investment.<\/p>\n<p class=\"Callout\">Today, the United States faces a similar ideological competition with China, yet it has reduced foreign aid spending by 92 percent while China invests heavily in digital infrastructure across the developing world.<\/p>\n<p>    Culture and Narrative<\/p>\n<p class=\"Normal\">Cultural influence represents a critical dimension of soft power, shaping how billions worldwide perceive American values, consume information, and organize economically. In the competition with China for global influence, control over information flows and cultural production translates directly into strategic advantage. When populations adopt American platforms for communication, entertainment, and education, they integrate into systems that embed American assumptions about free expression, individual agency, and market economics.<a href=\"#_edn199\" title=\"\" id=\"_ednref199\">[199]<\/a><\/p>\n<p class=\"Normal\">American technology platforms have achieved unprecedented global cultural reach through purely commercial expansion. Facebook\u2019s 3.07 billion users, YouTube\u2019s 2.54 billion, and Instagram\u2019s 2 billion dwarf Chinese alternatives\u2014WhatsApp maintains 73 percent usage in middle-income countries compared with 36 percent for TikTok.<a href=\"#_edn200\" title=\"\" id=\"_ednref200\">[200]<\/a> The creator economy is large and growing\u2014market researchers estimate that it will reach approximately $150 billion in 2024, with projections of nearly $1.1 trillion by 2034. Separately, analysis of platform engagement indicates that U.S.\/English language content commands outsized global reach, reinforcing American production norms and monetization models.<a href=\"#_edn201\" title=\"\" id=\"_ednref201\">[201]<\/a> YouTube alone contributes $35 billion to the U.S. GDP while training millions of international creators in American production methods and business practices.<a href=\"#_edn202\" title=\"\" id=\"_ednref202\">[202]<\/a><\/p>\n<p class=\"Normal\">Platform algorithms and monetization structures reinforce American cultural and economic norms without government intervention. English dominates, accounting for approximately 50 percent of online content, as creators worldwide optimize their content for English-speaking audiences to maximize revenue.<a href=\"#_edn203\" title=\"\" id=\"_ednref203\">[203]<\/a> Netflix\u2019s 277.65 million subscribers globally dedicate 60 percent of their viewing to U.S.-produced content.<a href=\"#_edn204\" title=\"\" id=\"_ednref204\">[204]<\/a> These commercial systems project influence more efficiently than government programs do, though their strategic value depends on continued alignment between corporate profits and national interests.<\/p>\n<p class=\"Normal\">These benefits exist because commercial incentives align with U.S. interests. TikTok\u2019s rapid growth demonstrates how quickly cultural influence can shift, and the United States cannot depend entirely on corporate platforms for cultural influence\u2014but given competition with state-backed Chinese alternatives, failing to recognize their strategic value would be shortsighted.<\/p>\n<p>Development Substitution<\/p>\n<p class=\"Normal\">American foreign aid has historically served strategic purposes beyond humanitarian goals\u2014preventing instability that breeds terrorism, building partnerships, and creating markets for U.S. goods. With the potential elimination of over 90 percent of USAID programs, removing $60 billion across 120 countries, a significant tool of American influence might disappear. Technology companies pursuing commercial opportunities in emerging markets now incidentally achieve development objectives that would otherwise require government funding.<\/p>\n<p class=\"Normal\">Alphabet\u2019s health care initiatives demonstrate this dynamic. The company has screened over 600,000 people for diabetic retinopathy while building relationships with health systems for future commercialization, with plans for 3 million screenings in India and Thailand.<a href=\"#_edn205\" title=\"\" id=\"_ednref205\">[205]<\/a> Microsoft has invested $165 million in AI for Good across 72 countries, creating Microsoft-trained workforces while addressing health challenges.<a href=\"#_edn206\" title=\"\" id=\"_ednref206\">[206]<\/a> Digital payment platforms seeking transaction fees have lifted hundreds of thousands from poverty\u2014in Kenya alone, mobile money has enabled 185,000 women to leave farming for business, reducing extreme poverty by 22 percent.<a href=\"#_edn207\" title=\"\" id=\"_ednref207\">[207]<\/a><\/p>\n<p class=\"Normal\">These corporate programs cannot replace government aid\u2014they follow commercial logic, ignore unprofitable markets, and may abandon regions when business models change. Yet, their activities achieve certain strategic objectives at zero taxpayer cost, filling gaps where government programs have been eliminated. Supporting American platforms competing globally costs less than traditional aid while generating development outcomes that benefit U.S. interests.<\/p>\n<p class=\"Normal\">Current tech policy debates focus overwhelmingly on potential harms while ignoring documented benefits that large technology companies generate beyond their core business activities. Regulatory impact assessments measure market concentration and privacy concerns without weighing them against contributions to economic dynamism and technological progress, leading to policies that may address narrow concerns while undermining broader U.S. competitive advantages.<\/p>\n<p class=\"Normal\">Big tech companies generate substantial positive spillovers: their R&amp;D produces knowledge that diffuses throughout the economy, their infrastructure investments create capabilities that thousands of businesses rely on, and their workforce programs train millions in critical digital skills. These benefits accrue to competitors, startups, and unrelated industries, yet rarely factor into policy decisions.<\/p>\n<p class=\"Normal\">Policymakers need frameworks that capture these ecosystem-wide effects. A regulation that reduces big tech R&amp;D spending by 10 percent might achieve its narrow objective (e.g., collecting revenue through fines) while eliminating billions in spillover benefits. Without measuring these trade-offs, policymakers operate with incomplete information. The same applies to infrastructure investments\u2014when regulations increase compliance costs, companies may scale back investments in data centers, submarine cables, and edge computing facilities that serve as critical infrastructure for thousands of other businesses.<\/p>\n<p class=\"Normal\">The global context makes recognizing these contributions urgent. China explicitly leverages its tech champions\u2019 spillover benefits for national development, while European policies focus primarily on holding down American tech companies so EU ones can hopefully emerge. In this environment, failing to account for positive externalities means systematically undervaluing a critical source of American competitive advantage. Policy decisions should reflect complete cost-benefit analyses that include innovation diffusion, infrastructure provision, and workforce development alongside traditional regulatory concerns.<\/p>\n<p class=\"Callout\">Regulations designed to address narrow concerns may inadvertently eliminate billions in spillover benefits, undermine critical infrastructure development, and weaken the very ecosystem advantages that distinguish the United States from competitors.<\/p>\n<p class=\"Normal\">The current regulatory approach to large technology companies systematically undervalues their contributions to American economic and technological leadership. While policymakers focus on market concentration and privacy concerns, they overlook the documented positive spillovers from R&amp;D that diffuses across industries to infrastructure investments that enable thousands of businesses, with workforce training that builds critical national capabilities.<\/p>\n<p class=\"Normal\">This incomplete accounting creates real risks. Regulations designed to address narrow concerns may inadvertently eliminate billions in spillover benefits, undermine critical infrastructure development, and weaken the very ecosystem advantages that distinguish the United States from competitors. China leverages its tech champions as engines of national development while Europe pursues restrictive policies aimed at constraining American companies. In this competitive environment, failing to recognize the full scope of big tech contributions amounts to unilateral economic disarmament.<\/p>\n<p class=\"Normal\">Effective policy requires measuring the complete impact of regulatory decisions. Cost-benefit analyses must include innovation diffusion, infrastructure provision, and human capital development alongside traditional regulatory metrics. Without this comprehensive approach, the United States risks dismantling competitive advantages that took decades to build and cannot be easily replaced. The choice is not between regulation and laissez-faire, but between informed policy that preserves beneficial spillovers and reflexive restrictions that sacrifice long-term competitiveness for short-term political gains.<\/p>\n<p class=\"BoxHeading1\">About the Author<\/p>\n<p class=\"BoxBody\">Hilal Aka is a policy analyst at ITIF focusing on U.S. technology competitiveness. Previously, she interned with the Center for a New American Security\u2019s technology and national security program and was an economic consultant on antitrust and competition matters at Charles River Associates. She holds a bachelor\u2019s degree from Wellesley College with a double major in economics and mathematics.<\/p>\n<p class=\"BoxHeading1\">About ITIF<\/p>\n<p class=\"BoxBody\">The Information Technology and Innovation Foundation (ITIF) is an independent 501(c)(3) nonprofit, nonpartisan research and educational institute that has been recognized repeatedly as the world\u2019s leading think tank for science and technology policy. Its mission is to formulate, evaluate, and promote policy solutions that accelerate innovation and boost productivity to spur growth, opportunity, and progress. For more information, visit <a href=\"https:\/\/itif.org\/about\/\" rel=\"nofollow noopener\" target=\"_blank\">itif.org\/about<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref1\" title=\"\" id=\"_edn1\">[1]<\/a>.\u00a0\u00a0\u00a0\u00a0 Ash Johnson, \u201cThe Facts Behind Allegations of Political Bias on Social Media\u201d (Information Technology and Innovation Foundation, October 26, 2023), <a href=\"https:\/\/itif.org\/publications\/2023\/10\/26\/the-facts-behind-allegations-of-political-bias-on-social-media\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/itif.org\/publications\/2023\/10\/26\/the-facts-behind-allegations-of-political-bias-on-social-media\/<\/a>; Lauren Feiner, \u201cHow Facebook and Big Tech Gained D.C. Scrutiny in the 2010s,\u201d CNBC, December 24, 2019; updated December 26, 2019, <a href=\"https:\/\/www.cnbc.com\/2019\/12\/24\/how-facebook-and-big-tech-gained-dc-scrutiny-in-the-2010s.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.cnbc.com\/2019\/12\/24\/how-facebook-and-big-tech-gained-dc-scrutiny-in-the-2010s.html<\/a>; Elizabeth Warren, \u201cHere\u2019s How We Can Break Up Big Tech\u201d (Medium, March 8, 2019), <a href=\"https:\/\/medium.com\/@teamwarren\/heres-how-we-can-break-up-big-tech-9ad9e0da324c\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/medium.com\/@teamwarren\/heres-how-we-can-break-up-big-tech-9ad9e0da324c<\/a>; Auguste Meyrat, \u201cBreak Up Big Tech Before It\u2019s Too Late,\u201d American Greatness, July 26, 2021, <a href=\"https:\/\/amgreatness.com\/2021\/07\/26\/break-up-big-tech-before-its-too-late\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/amgreatness.com\/2021\/07\/26\/break-up-big-tech-before-its-too-late<\/a>; Adam Thierer, \u201cCan Trump Hold the Tech Right and Populist Right Together?\u201d (R Street Institute, February 7, 2025), <a href=\"https:\/\/www.rstreet.org\/commentary\/can-trump-hold-the-tech-right-and-populist-right-together\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.rstreet.org\/commentary\/can-trump-hold-the-tech-right-and-populist-right-together\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref2\" title=\"\" id=\"_edn2\">[2]<\/a>.\u00a0\u00a0\u00a0\u00a0 European Commission, Joint Research Centre, 2024 EU Industrial R&amp;D Investment Scoreboard, (accessed August 3, 2025), <a href=\"https:\/\/iri.jrc.ec.europa.eu\/scoreboard\/2024-eu-industrial-rd-investment-scoreboard\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/iri.jrc.ec.europa.eu\/scoreboard\/2024-eu-industrial-rd-investment-scoreboard<\/a>; Macrotrends, \u201cAmazon Research and Development Expenses 2003\u20132025\u201d (accessed August 3, 2025), <a href=\"https:\/\/www.macrotrends.net\/stocks\/charts\/AMZN\/amazon\/research-development-expenses\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.macrotrends.net\/stocks\/charts\/AMZN\/amazon\/research-development-expenses<\/a>; Amazon doesn\u2019t report a standalone R&amp;D line. The commonly used proxy\u2014\u201dTechnology &amp; infrastructure\u201d\u2014bundles true R&amp;D with large non-R&amp;D costs (e.g., servers, data-center depreciation, ops). Subtracting \u201ctotal video and music expense\u201d removes content, but the remainder still contains infrastructure. Treat Amazon\u2019s reported figure (or ex-content variant) as an upper bound on \u201ctrue R&amp;D.\u201d See U.S. Securities and Exchange Commission, \u201cAmazon.com, Inc. Form 10-K for the Fiscal Year Ended December 31, 2024\u201d (filed February 2, 2025), <a href=\"https:\/\/www.sec.gov\/Archives\/edgar\/data\/1018724\/000101872425000004\/amzn-20241231.htm\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.sec.gov\/Archives\/edgar\/data\/1018724\/000101872425000004\/amzn-20241231.htm<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref4\" title=\"\" id=\"_edn4\">[4]<\/a>.\u00a0\u00a0\u00a0\u00a0 Brad Smith, \u201cThe Golden Opportunity for American AI,\u201d Microsoft On the Issues, January 3, 2025, <a href=\"https:\/\/blogs.microsoft.com\/on-the-issues\/2025\/01\/03\/the-golden-opportunity-for-american-ai\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/blogs.microsoft.com\/on-the-issues\/2025\/01\/03\/the-golden-opportunity-for-american-ai\/<\/a>; \u201cThe Year of Quantum: From concept to reality in 2025\u201d (McKinsey &amp; Company, June 2025), <a href=\"https:\/\/www.mckinsey.com\/capabilities\/mckinsey-digital\/our-insights\/the-year-of-quantum-from-concept-to-reality-in-2025\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.mckinsey.com\/capabilities\/mckinsey-digital\/our-insights\/the-year-of-quantum-from-concept-to-reality-in-2025<\/a>; Christian Schoeberl and Jack Corrigan, \u201cFunding the AI Cloud \u2014 Amazon, Alphabet, and Microsoft\u2019s Cloud Computing Investments, Part 3\u201d (Center for Security and Emerging Technology, October 2024), <a href=\"https:\/\/cset.georgetown.edu\/publication\/funding-the-ai-cloud-amazon-alphabet-and-microsofts-cloud-computing-investments-part-3\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/cset.georgetown.edu\/publication\/funding-the-ai-cloud-amazon-alphabet-and-microsofts-cloud-computing-investments-part-3\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref7\" title=\"\" id=\"_edn7\">[7]<\/a>.\u00a0\u00a0\u00a0\u00a0 PsiQuantum, \u201cPsiQuantum Raises $1 Billion to Build Million-Qubit Scale, Fault-Tolerant Quantum Computers,\u201d news release, 2025, <a href=\"https:\/\/www.psiquantum.com\/news-import\/psiquantum-1b-fundraise\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.psiquantum.com\/news-import\/psiquantum-1b-fundraise<\/a>; IQM Quantum Computers, \u201cIQM Quantum Computers Raises over $300 Million in Series B Funding Round Led by U.S. Investor Ten Eleven Ventures with strong support from Tesi,\u201d news release, September 3, 2025, <a href=\"https:\/\/meetiqm.com\/press-releases\/iqm-quantum-computers-raises-over-300-million-in-series-b-funding-round\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/meetiqm.com\/press-releases\/iqm-quantum-computers-raises-over-300-million-in-series-b-funding-round\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref10\" title=\"\" id=\"_edn10\">[10]<\/a>.\u00a0\u00a0 Sandro Mendon\u00e7a, Eduardo Silva, and Bruno Dam\u00e1sio, \u201cThe Knowledge Base of Big Tech: Research as a Source of Informational Leadership by the Dominant U.S. Digital Platforms,\u201d Telecommunications Policy 49, no. 7 (August 2025): 102908, <a href=\"https:\/\/doi.org\/10.1016\/j.telpol.2025.102908\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/doi.org\/10.1016\/j.telpol.2025.102908<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref15\" title=\"\" id=\"_edn15\">[15]<\/a>.\u00a0\u00a0 Amy Leong, \u201cGoogle\u2019s Acquisition of YouTube (2006): A Big Moment in Online Media,\u201d Medium, December 14, 2024, <a href=\"https:\/\/medium.com\/@meiyee715\/googles-acquisition-of-youtube-2006-a-big-moment-in-online-media-4ed83a3f5f13\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/medium.com\/@meiyee715\/googles-acquisition-of-youtube-2006-a-big-moment-in-online-media-4ed83a3f5f13<\/a>; Siddhant Sinha, \u201cNearly 17 years ago, Google secured its most significant acquisition: Have you any idea what it was?\u201d Medium, July 2, 2022, <a href=\"https:\/\/medium.com\/@Blank_Misfit\/nearly-17-years-ago-google-secured-its-most-significant-acquisition-have-you-any-idea-what-it-3e3d9bf1d87d\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/medium.com\/@Blank_Misfit\/nearly-17-years-ago-google-secured-its-most-significant-acquisition-have-you-any-idea-what-it-3e3d9bf1d87d<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref24\" title=\"\" id=\"_edn24\">[24]<\/a>.\u00a0\u00a0 European Commission, \u201c2024 EU Industrial R&amp;D Investment Scoreboard\u201d; Macrotrends, \u201cAmazon Research and Development Expenses 2003\u20132025\u201d; Amazon doesn\u2019t report a standalone R&amp;D line. The commonly used proxy\u2014\u201dTechnology &amp; infrastructure\u201d\u2014bundles true R&amp;D with large non-R&amp;D costs (e.g., servers, data-center depreciation, ops). Subtracting \u201ctotal video and music expense\u201d removes content, but the remainder still contains infrastructure. Treat Amazon\u2019s reported figure (or ex-content variant) as an upper bound on \u201ctrue R&amp;D.\u201d See U.S. Securities and Exchange Commission, \u201cAmazon.com, Inc. Form 10-K for the Fiscal Year Ended December 31, 2024.\u201d<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref25\" title=\"\" id=\"_edn25\">[25]<\/a>.\u00a0\u00a0 European Commission, 2024 EU Industrial R&amp;D Investment Scoreboard; Macrotrends, \u201cAmazon Research and Development Expenses.\u201d<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref31\" title=\"\" id=\"_edn31\">[31]<\/a>.\u00a0\u00a0 \u201cMeta\u2019s Reality Labs Posts $5 Billion Loss in Fourth Quarter,\u201d (CNBC, January 29, 2025), <a href=\"https:\/\/www.cnbc.com\/2025\/01\/29\/metas-reality-labs-posts-5-billion-loss-in-fourth-quarter.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.cnbc.com\/2025\/01\/29\/metas-reality-labs-posts-5-billion-loss-in-fourth-quarter.html<\/a>; TJ Denzer, \u201cFacebook (META) Reality Labs Lost $17.7 Billion in 2024,\u201d (Shacknews, January 30, 2025), <a href=\"https:\/\/www.shacknews.com\/article\/142903\/facebook-meta-fy-2024-reality-labs\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.shacknews.com\/article\/142903\/facebook-meta-fy-2024-reality-labs<\/a>; \u201cMeta\u2019s Reality Labs Posts $5 Billion Loss for Q4 2024,\u201d (Digital Music News, January 31, 2025), <a href=\"https:\/\/www.digitalmusicnews.com\/2025\/01\/30\/metas-reality-labs-posts-5-billion-loss-for-q4-2024\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.digitalmusicnews.com\/2025\/01\/30\/metas-reality-labs-posts-5-billion-loss-for-q4-2024\/<\/a>; Grayson Brulte, \u201cAlphabet Q2 2023 Earnings, Waymo,\u201d (The Road to Autonomy, July 8, 2024), <a href=\"https:\/\/www.roadtoautonomy.com\/alphabet-q2-2023-earnings-waymo\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.roadtoautonomy.com\/alphabet-q2-2023-earnings-waymo\/<\/a>; Akash Sriram et al., \u201cAlphabet\u2019s Waymo Cuts More Than 100 Jobs in Second Round of Layoffs,\u201d (Reuters via Yahoo Finance, March 1, 2023), <a href=\"https:\/\/finance.yahoo.com\/news\/alphabets-waymo-cuts-137-jobs-181510391.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/finance.yahoo.com\/news\/alphabets-waymo-cuts-137-jobs-181510391.html<\/a>; Blathnaid O\u2019Dea, \u201cAlphabet Cuts Jobs at Two of Its \u2018Other Bets\u2019 Divisions,\u201d (Silicon Republic, January 13, 2023), <a href=\"https:\/\/www.siliconrepublic.com\/business\/alphabet-cuts-jobs-other-bets-verily-intrinsic\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.siliconrepublic.com\/business\/alphabet-cuts-jobs-other-bets-verily-intrinsic<\/a>; \u201cAmazon Lost $25B on Alexa-Powered Devices,\u201d (Mobile World Live, July 24, 2024), <a href=\"https:\/\/www.mobileworldlive.com\/devices\/amazon-lost-25b-on-alexa-powered-devices\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.mobileworldlive.com\/devices\/amazon-lost-25b-on-alexa-powered-devices\/<\/a>; \u201cAmazon Alexa on Track to Lose $10 Billion This Year, Described as \u2018Colossal Failure\u2019 in New Report,\u201d (Fox Business, November 22, 2022), <a href=\"https:\/\/www.foxbusiness.com\/technology\/amazon-alexa-track-lose-10-billion-year-described-colossal-failure-new-report\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.foxbusiness.com\/technology\/amazon-alexa-track-lose-10-billion-year-described-colossal-failure-new-report<\/a>; Ron Amadeo, \u201cAmazon Alexa Is a \u2018Colossal Failure,\u2019 On Pace To Lose $10 Billion This Year,\u201d (Ars Technica, November 21, 2022), <a href=\"https:\/\/arstechnica.com\/gadgets\/2022\/11\/amazon-alexa-is-a-colossal-failure-on-pace-to-lose-10-billion-this-year\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/arstechnica.com\/gadgets\/2022\/11\/amazon-alexa-is-a-colossal-failure-on-pace-to-lose-10-billion-this-year\/<\/a>; Joe Rossignol, \u201cApple Spent More Than $10 Billion on Apple Car Before Canceling Project,\u201d (MacRumors, February 29, 2024), <a href=\"https:\/\/www.macrumors.com\/2024\/02\/28\/apple-car-10-billion-spent\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.macrumors.com\/2024\/02\/28\/apple-car-10-billion-spent\/<\/a>; Bobby Allyn, \u201cAfter 10 Years of Development, Apple Abruptly Cancels Its Electric Car Project,\u201d (NPR, February 28, 2024), <a href=\"https:\/\/www.npr.org\/2024\/02\/27\/1234315814\/apple-cancels-electric-car\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.npr.org\/2024\/02\/27\/1234315814\/apple-cancels-electric-car<\/a>; Ashley Stewart, \u201cMicrosoft Lost Billions of Dollars on HoloLens, and Its Huge IVAS Military Contract Is in Trouble,\u201d (Business Insider via DNyuz, October 14, 2024), <a href=\"https:\/\/dnyuz.com\/2024\/10\/14\/microsoft-lost-billions-of-dollars-on-hololens-and-its-huge-ivas-military-contract-is-in-trouble\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/dnyuz.com\/2024\/10\/14\/microsoft-lost-billions-of-dollars-on-hololens-and-its-huge-ivas-military-contract-is-in-trouble\/<\/a>; Luke Jones, \u201cMicrosoft Ends HoloLens 2 Production, Losing Billions, And Pulling Back Military IVAS Project,\u201d (WinBuzzer, October 15, 2024), <a href=\"https:\/\/winbuzzer.com\/2024\/10\/15\/microsoft-ends-hololens-2-production-losing-billions-and-pulling-back-military-ivas-project-xcxwbn\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/winbuzzer.com\/2024\/10\/15\/microsoft-ends-hololens-2-production-losing-billions-and-pulling-back-military-ivas-project-xcxwbn\/<\/a>; \u201cAnduril Takes Over $22B US Army Combat Headset Contract from Microsoft,\u201d (Straight Arrow News, February 12, 2025), <a href=\"https:\/\/san.com\/cc\/anduril-takes-over-22b-us-army-combat-headset-contract-from-microsoft\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/san.com\/cc\/anduril-takes-over-22b-us-army-combat-headset-contract-from-microsoft\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref34\" title=\"\" id=\"_edn34\">[34]<\/a>.\u00a0\u00a0 Ben Dickson, \u201cThe GPT-3 Economy,\u201d TechTalks, September 21, 2020, <a href=\"https:\/\/bdtechtalks.com\/2020\/09\/21\/gpt-3-economy-business-model\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/bdtechtalks.com\/2020\/09\/21\/gpt-3-economy-business-model\/<\/a>; David Meyer, \u201cWhy the Cost of Training AI Could Soon Become Too Much to Bear,\u201d Fortune, April 4, 2024, <a href=\"https:\/\/fortune.com\/2024\/04\/04\/ai-training-costs-how-much-is-too-much-openai-gpt-anthropic-microsoft\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/fortune.com\/2024\/04\/04\/ai-training-costs-how-much-is-too-much-openai-gpt-anthropic-microsoft\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref37\" title=\"\" id=\"_edn37\">[37]<\/a>.\u00a0\u00a0 Taiwan Semiconductor Manufacturing Company Limited, \u201cTSMC and U.S. Department of Commerce Announce up to US$6.6 Billion in Proposed CHIPS Act Direct Funding\u201d (TSMC, April 8, 2024), <a href=\"https:\/\/pr.tsmc.com\/english\/news\/3122\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/pr.tsmc.com\/english\/news\/3122<\/a>; Samsung Global Newsroom, \u201cSamsung Electronics Announces New Advanced Semiconductor Fab Site in Taylor, Texas\u201d (Samsung Electronics, November 23, 2021), <a href=\"https:\/\/news.samsung.com\/global\/samsung-electronics-announces-new-advanced-semiconductor-fab-site-in-taylor-texas\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/news.samsung.com\/global\/samsung-electronics-announces-new-advanced-semiconductor-fab-site-in-taylor-texas<\/a>; Tim De Chant, \u201cNvidia, Google, and Bill Gates help Commonwealth Fusion Systems raise $863M,\u201d TechCrunch, August 28, 2025, <a href=\"https:\/\/techcrunch.com\/2025\/08\/28\/nvidia-google-and-bill-gates-help-commonwealth-fusion-systems-raise-863m\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/techcrunch.com\/2025\/08\/28\/nvidia-google-and-bill-gates-help-commonwealth-fusion-systems-raise-863m\/<\/a>; Charles Seife, \u201cWorld\u2019s Largest Fusion Project Is in Big Trouble, New Documents Reveal,\u201d Scientific American, February 20, 2024, <a href=\"https:\/\/www.scientificamerican.com\/article\/worlds-largest-fusion-project-is-in-big-trouble-new-documents-reveal\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.scientificamerican.com\/article\/worlds-largest-fusion-project-is-in-big-trouble-new-documents-reveal\/<\/a>; \u201cStarlink,\u201d Wikipedia, last modified September 24, 2025, <a href=\"https:\/\/en.wikipedia.org\/wiki\/Starlink\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/en.wikipedia.org\/wiki\/Starlink<\/a>; Amazon, \u201cAmazon receives FCC approval for Project Kuiper satellite constellation\u201d (Amazon, July 30, 2020), <a href=\"https:\/\/www.aboutamazon.com\/news\/company-news\/amazon-receives-fcc-approval-for-project-kuiper-satellite-constellation\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.aboutamazon.com\/news\/company-news\/amazon-receives-fcc-approval-for-project-kuiper-satellite-constellation<\/a>; Rebecca Bellan, \u201cAlphabet\u2019s self-driving unit Waymo closes $5.6 billion funding round as robotaxi race heats up in the U.S.,\u201d CNBC, October 25, 2024, <a href=\"https:\/\/www.cnbc.com\/2024\/10\/25\/alphabets-self-driving-unit-waymo-closes-5point6-billion-funding-round.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.cnbc.com\/2024\/10\/25\/alphabets-self-driving-unit-waymo-closes-5point6-billion-funding-round.html<\/a>; \u201cGeneral Motors to shut down autonomous car developer Cruise despite $10 billion spend,\u201d Robotics and Automation News, December 13, 2024, <a href=\"https:\/\/roboticsandautomationnews.com\/2024\/12\/13\/general-motors-to-shut-down-autonomous-car-developer-cruise-despite-10-billion-spend\/87685\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/roboticsandautomationnews.com\/2024\/12\/13\/general-motors-to-shut-down-autonomous-car-developer-cruise-despite-10-billion-spend\/87685\/<\/a>; \u201cGigafactory Nevada,\u201d Wikipedia, last modified August 1, 2025, <a href=\"https:\/\/en.wikipedia.org\/wiki\/Gigafactory_Nevada\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/en.wikipedia.org\/wiki\/Gigafactory_Nevada<\/a>; Rebecca Tinucci, \u201cTesla invests $3.6B in two new Nevada factories to build Semis and cells,\u201d TechCrunch, January 24, 2023, <a href=\"https:\/\/techcrunch.com\/2023\/01\/24\/tesla-invests-3-6b-in-two-new-nevada-factories-to-build-semis-and-cells\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/techcrunch.com\/2023\/01\/24\/tesla-invests-3-6b-in-two-new-nevada-factories-to-build-semis-and-cells\/<\/a>; \u201cElon Musk\u2019s xAI Flips The Switch On 100K Nvidia H100 GPUs Worth Up To $4B In Memphis Supercluster: \u2018Most Powerful AI Training Cluster In The World\u2019,\u201d Benzinga, July 22, 2024, <a href=\"https:\/\/www.benzinga.com\/news\/24\/07\/39881748\/elon-musks-xai-flips-the-switch-on-100k-nvidia-h100-gpus-worth-up-to-4b-in-memphis-supercluster-most\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.benzinga.com\/news\/24\/07\/39881748\/elon-musks-xai-flips-the-switch-on-100k-nvidia-h100-gpus-worth-up-to-4b-in-memphis-supercluster-most<\/a>; \u201cAWS Has Spent $35 Billion on its Northern Virginia Data Centers,\u201d Data Center Frontier, <a href=\"https:\/\/www.datacenterfrontier.com\/cloud\/article\/11427911\/aws-has-spent-35-billion-on-its-northern-virginia-data-centers\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.datacenterfrontier.com\/cloud\/article\/11427911\/aws-has-spent-35-billion-on-its-northern-virginia-data-centers<\/a>; Lindsey Wilkinson, \u201cGoogle Cloud pours more than $25B into domestic AI infrastructure,\u201d CIO Dive, July 15, 2025, <a href=\"https:\/\/www.ciodive.com\/news\/google-cloud-blackstone-aws-us-ai-data-center-buildouts\/753147\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.ciodive.com\/news\/google-cloud-blackstone-aws-us-ai-data-center-buildouts\/753147\/<\/a>; \u201cBig Tech\u2019s Quantum Computing Investments: Google, IBM, and Microsoft by the Numbers,\u201d PatentPC, August 2025, <a href=\"https:\/\/patentpc.com\/blog\/big-techs-quantum-computing-investments-google-ibm-and-microsoft-by-the-numbers\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/patentpc.com\/blog\/big-techs-quantum-computing-investments-google-ibm-and-microsoft-by-the-numbers<\/a>; \u201cIBM Reports $1B in Cumulative Quantum Computing Revenue,\u201d IoT World Today, February 6, 2025, <a href=\"https:\/\/www.iotworldtoday.com\/quantum\/ibm-reports-1b-in-cumulative-quantum-computing-revenue\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.iotworldtoday.com\/quantum\/ibm-reports-1b-in-cumulative-quantum-computing-revenue<\/a>; \u201cDiving Deep into Submarine Cables: The Undersea Lifelines of Internet Connectivity,\u201d Kentik Blog, <a href=\"https:\/\/www.kentik.com\/blog\/diving-deep-into-submarine-cables-undersea-lifelines-of-internet-connectivity\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.kentik.com\/blog\/diving-deep-into-submarine-cables-undersea-lifelines-of-internet-connectivity\/<\/a>; \u201cThe Economics of Submarine Cables,\u201d TeleGeography, <a href=\"https:\/\/blog.telegeography.com\/the-economics-of-submarine-cables\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/blog.telegeography.com\/the-economics-of-submarine-cables<\/a>; \u201cHow Much Does It Cost to Train Frontier AI Models?\u201d Epoch AI, January 13, 2025, <a href=\"https:\/\/epoch.ai\/blog\/how-much-does-it-cost-to-train-frontier-ai-models\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/epoch.ai\/blog\/how-much-does-it-cost-to-train-frontier-ai-models<\/a>; Ben Cottier et al., \u201cThe rising costs of training frontier AI models,\u201d arXiv preprint arXiv:2405.21015, 2024, <a href=\"https:\/\/arxiv.org\/abs\/2405.21015\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/arxiv.org\/abs\/2405.21015<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref41\" title=\"\" id=\"_edn41\">[41]<\/a>.\u00a0\u00a0 \u201cQuantum Computer,\u201d Google Quantum AI, accessed September 24, 2025, <a href=\"https:\/\/quantumai.google\/quantumcomputer\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/quantumai.google\/quantumcomputer<\/a>; \u201cOur Lab,\u201d Google Quantum AI, accessed September 24, 2025, <a href=\"https:\/\/quantumai.google\/lab\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/quantumai.google\/lab<\/a>; \u201cInside Google\u2019s Quantum Computing Data Center,\u201d Data Center Frontier, November 22, 2021, <a href=\"https:\/\/www.datacenterfrontier.com\/featured\/article\/11427786\/inside-googles-quantum-computing-data-center\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.datacenterfrontier.com\/featured\/article\/11427786\/inside-googles-quantum-computing-data-center<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref51\" title=\"\" id=\"_edn51\">[51]<\/a>.\u00a0\u00a0 Anthony Garreffa, \u201cNVIDIA spent $10 billion on developing its next-generation Blackwell GPU,\u201d TweakTown, March 30, 2024, <a href=\"https:\/\/www.tweaktown.com\/news\/96987\/nvidia-spent-10-billion-on-developing-its-next-generation-blackwell-gpu\/index.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.tweaktown.com\/news\/96987\/nvidia-spent-10-billion-on-developing-its-next-generation-blackwell-gpu\/index.html<\/a>; Sanjiv Kumar et al., \u201cNVIDIA Accelerates Science and Engineering With CUDA-X Libraries Powered by GH200 and GB200 Superchips,\u201d NVIDIA Blog, March 26, 2025, <a href=\"https:\/\/blogs.nvidia.com\/blog\/cuda-x-grace-hopper-blackwell\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/blogs.nvidia.com\/blog\/cuda-x-grace-hopper-blackwell\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref63\" title=\"\" id=\"_edn63\">[63]<\/a>.\u00a0\u00a0 European Commission, 2024 EU Industrial R&amp;D Investment Scoreboard; Macrotrends, \u201cAmazon Research and Development Expenses.\u201d<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref65\" title=\"\" id=\"_edn65\">[65]<\/a>.\u00a0\u00a0 PARAT (Private-Sector AI-Related Activity Tracker),\u201d Emerging Technology Observatory, Center for Security and Emerging Technology, Georgetown University, accessed October 1, 2025, <a href=\"https:\/\/parat.eto.tech\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/parat.eto.tech\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref69\" title=\"\" id=\"_edn69\">[69]<\/a>.\u00a0\u00a0 \u201cICLR, NeurIPS, and ICML are the top three Publications for Artificial Intelligence, According to Google\u2019s Scholar Metrics 2022,\u201d Analytics Drift, June 30, 2022, <a href=\"https:\/\/analyticsdrift.com\/iclr-neurips-and-icml-are-top-three-publications-for-artificial-intelligence-according-to-googles-scholar-metrics-2022\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/analyticsdrift.com\/iclr-neurips-and-icml-are-top-three-publications-for-artificial-intelligence-according-to-googles-scholar-metrics-2022\/<\/a>; lixin4ever, \u201cConference-Acceptance-Rate: Acceptance rates for the major AI conferences,\u201d GitHub repository, accessed January 2025, <a href=\"https:\/\/github.com\/lixin4ever\/Conference-Acceptance-Rate\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/github.com\/lixin4ever\/Conference-Acceptance-Rate<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref72\" title=\"\" id=\"_edn72\">[72]<\/a>.\u00a0\u00a0 Adam Paszke et al., \u201cPyTorch: An Imperative Style, High-Performance Deep Learning Library,\u201d in Proceedings of the 33rd International Conference on Neural Information Processing Systems (NeurIPS 2019), <a href=\"https:\/\/arxiv.org\/abs\/1912.01703\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/arxiv.org\/abs\/1912.01703<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref75\" title=\"\" id=\"_edn75\">[75]<\/a>.\u00a0\u00a0 The foundational importance score measures how essential a project is to other software systems based on factors like dependency chains and downstream impact. The community adoption score measures usage intensity through metrics including downloads, contributors, and developer engagement. Both are composite indices scaled from low to high rather than absolute counts; Emerging Technology Observatory, \u201cOpen-source software Research and Community Activity (ORCA),\u201d Center for Security and Emerging Technology, Georgetown University, accessed September 2025, <a href=\"https:\/\/orca.eto.tech\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/orca.eto.tech\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref82\" title=\"\" id=\"_edn82\">[82]<\/a>. Amazon, \u201c10 Years of Amazon Robotics: How Robots Help Sort Your Holiday Packages,\u201d Amazon News, November 3, 2022, <a href=\"https:\/\/www.amazon.com\/news\/operations\/10-years-of-amazon-robotics-how-robots-help-sort-your-holiday-packages\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.amazon.com\/news\/operations\/10-years-of-amazon-robotics-how-robots-help-sort-your-holiday-packages<\/a>; Karen Weise, \u201cAmazon\u2019s Robot Workforce Has Increased by 50% in Just Two Years,\u201d The Seattle Times, December 21, 2023, <a href=\"https:\/\/www.seattletimes.com\/business\/amazon\/amazons-robot-workforce-has-increased-50-percent\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.seattletimes.com\/business\/amazon\/amazons-robot-workforce-has-increased-50-percent<\/a>; Eugene Kim, \u201cAmazon Has 750,000 Robots Working in Its Warehouses,\u201d Business Insider, October 18, 2023, <a href=\"https:\/\/www.businessinsider.com\/amazon-warehouses-robots-workforce-2023-10\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.businessinsider.com\/amazon-warehouses-robots-workforce-2023-10<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref92\" title=\"\" id=\"_edn92\">[92]<\/a>.\u00a0\u00a0 \u201cAI Futures Fund,\u201d Google Labs, accessed September 24, 2025, <a href=\"https:\/\/labs.google\/aifuturesfund\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/labs.google\/aifuturesfund\/<\/a>; \u201cMicrosoft for Startups Investor Offer Benefits,\u201d Microsoft Learn, accessed September 24, 2025, <a href=\"https:\/\/learn.microsoft.com\/en-us\/microsoft-for-startups\/benefits\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/learn.microsoft.com\/en-us\/microsoft-for-startups\/benefits<\/a>; \u201cMeta Launches Llama Startup Program to Fuel Generative AI Innovation,\u201d OpenTools AI News, May 22, 2025, <a href=\"https:\/\/opentools.ai\/news\/meta-launches-llama-startup-program-to-fuel-generative-ai-innovation\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/opentools.ai\/news\/meta-launches-llama-startup-program-to-fuel-generative-ai-innovation<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref99\" title=\"\" id=\"_edn99\">[99]<\/a>.\u00a0\u00a0 Engineering at Meta, \u201cBuilding Meta\u2019s GenAI Infrastructure,\u201d May 14, 2024, <a href=\"https:\/\/engineering.fb.com\/2024\/03\/12\/data-center-engineering\/building-metas-genai-infrastructure\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/engineering.fb.com\/2024\/03\/12\/data-center-engineering\/building-metas-genai-infrastructure\/<\/a>; John Martinis and Sergio Boixo, \u201cQuantum Supremacy Using a Programmable Superconducting Processor,\u201d Google AI Blog, October 23, 2019, <a href=\"http:\/\/blog.research.google\/2019\/10\/quantum-supremacy-using-programmable.html\" rel=\"nofollow noopener\" target=\"_blank\">http:\/\/blog.research.google\/2019\/10\/quantum-supremacy-using-programmable.html<\/a>; Amazon Web Services, \u201cAmazon Braket \u2013 Quantum Cloud Computing Service,\u201d <a href=\"https:\/\/aws.amazon.com\/braket\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/aws.amazon.com\/braket\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref117\" title=\"\" id=\"_edn117\">[117]<\/a>. Stacie Pettyjohn and Tate Nurkin, \u201cHow Modern Militaries Are Leveraging AI\u201d (Atlantic Council, July 2023), <a href=\"https:\/\/www.atlanticcouncil.org\/in-depth-research-reports\/report\/how-modern-militaries-are-leveraging-ai\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.atlanticcouncil.org\/in-depth-research-reports\/report\/how-modern-militaries-are-leveraging-ai\/<\/a>; \u201cMilitary Applications of AI in 2024\u201d (Sentient Digital, Inc., June 2024), <a href=\"https:\/\/sdi.ai\/blog\/the-most-useful-military-applications-of-ai\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/sdi.ai\/blog\/the-most-useful-military-applications-of-ai\/<\/a>; Andrew P. Barlow and Allison Bender, \u201cInnovating Defense: Generative AI\u2019s Role in Military Evolution,\u201d Infantry Magazine, Summer 2025, <a href=\"https:\/\/www.army.mil\/article\/286707\/innovating_defense_generative_ais_role_in_military_evolution\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.army.mil\/article\/286707\/innovating_defense_generative_ais_role_in_military_evolution<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref122\" title=\"\" id=\"_edn122\">[122]<\/a>. \u201cSemiconductor Industry R&amp;D Spending: Who\u2019s Investing the Most?\u201d PatentPC, July 15, 2025, <a href=\"https:\/\/patentpc.com\/blog\/semiconductor-industry-rd-spending-whos-investing-the-most\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/patentpc.com\/blog\/semiconductor-industry-rd-spending-whos-investing-the-most<\/a>; Tom\u2019s Hardware, \u201cIntel spends more on R&amp;D than Nvidia and AMD combined, yet continues to lag in market cap\u201d (Yahoo Finance, October 15, 2024), <a href=\"https:\/\/finance.yahoo.com\/news\/intel-spends-more-r-d-154648983.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/finance.yahoo.com\/news\/intel-spends-more-r-d-154648983.html<\/a>; \u201cMorgan Stanley Research: With $16 Billion In R&amp;D Spending This Year Alone, NVIDIA Can Maintain A 4-5 Year Development Cycle\u201d (Wccftech, February 14, 2025), <a href=\"https:\/\/wccftech.com\/morgan-stanley-research-with-16-billion-in-rd-spending-this-year-alone-nvidia-can-maintain-a-4-5%E2%80%91year-development-cycle-by-running-three-design-teams-sequentially-each\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/wccftech.com\/morgan-stanley-research-with-16-billion-in-rd-spending-this-year-alone-nvidia-can-maintain-a-4-5%E2%80%91year-development-cycle-by-running-three-design-teams-sequentially-each\/<\/a>; \u201cSilicon Island: Assessing Taiwan\u2019s Importance to U.S. Economic Growth and Security\u201d (Center for Strategic and International Studies, January 10, 2025), <a href=\"https:\/\/www.csis.org\/analysis\/silicon-island-assessing-taiwans-importance-us-economic-growth-and-security\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.csis.org\/analysis\/silicon-island-assessing-taiwans-importance-us-economic-growth-and-security<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref127\" title=\"\" id=\"_edn127\">[127]<\/a>. Ashley Roque and Sydney J. Freedberg Jr., \u201cMicrosoft announces plan to slide $22 billion IVAS contract over to Anduril,\u201d Breaking Defense, February 11, 2025, <a href=\"https:\/\/breakingdefense.com\/2025\/02\/microsoft-announces-plan-to-slide-22-billion-ivas-contract-over-to-anduril\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/breakingdefense.com\/2025\/02\/microsoft-announces-plan-to-slide-22-billion-ivas-contract-over-to-anduril\/<\/a>; Jaspreet Singh, \u201cPentagon awards Google, Amazon, Microsoft, Oracle $9B cloud contract,\u201d Breaking Defense, December 7, 2022, <a href=\"https:\/\/breakingdefense.com\/2022\/12\/pentagon-awards-google-amazon-microsoft-oracle-9b-cloud-contract\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/breakingdefense.com\/2022\/12\/pentagon-awards-google-amazon-microsoft-oracle-9b-cloud-contract\/<\/a>; Billy Mitchell, \u201cCIA Awards Secret Multibillion-Dollar Cloud Contract,\u201d Nextgov, November 20, 2020, <a href=\"https:\/\/www.nextgov.com\/modernization\/2020\/11\/exclusive-cia-awards-secret-multibillion-dollar-cloud-contract\/170227\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.nextgov.com\/modernization\/2020\/11\/exclusive-cia-awards-secret-multibillion-dollar-cloud-contract\/170227\/<\/a>; Frank Konkel, \u201cNSA Re-awards Secret $10 Billion Contract to Amazon,\u201d Nextgov, April 27, 2022, <a href=\"https:\/\/www.nextgov.com\/emerging-tech\/2022\/04\/nsa-re-awards-secret-10-billion-contract-amazon\/366184\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.nextgov.com\/emerging-tech\/2022\/04\/nsa-re-awards-secret-10-billion-contract-amazon\/366184\/<\/a>; Brandi Vincent, \u201cPentagon Awards $9B Cloud Contract to Amazon, Google, Microsoft, Oracle,\u201d Nextgov, December 7, 2022, <a href=\"https:\/\/www.nextgov.com\/digital-government\/2022\/12\/amazon-google-microsoft-oracle-awarded-9b-pentagon-cloud-contract\/380596\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.nextgov.com\/digital-government\/2022\/12\/amazon-google-microsoft-oracle-awarded-9b-pentagon-cloud-contract\/380596\/<\/a>; Nick Wakeman, \u201cCIA makes awards for intelligence community\u2019s next massive cloud contract,\u201d Washington Technology, November 20, 2020, <a href=\"https:\/\/www.washingtontechnology.com\/2020\/11\/cia-makes-awards-for-intelligence-communitys-next-massive-cloud-contract\/355018\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.washingtontechnology.com\/2020\/11\/cia-makes-awards-for-intelligence-communitys-next-massive-cloud-contract\/355018\/<\/a>; Frank Konkel, \u201cA decade-old risk led to \u2018phenomenal partnership\u2019 between AWS and the intel community,\u201d Nextgov, December 14, 2024, <a href=\"https:\/\/www.nextgov.com\/acquisition\/2024\/12\/decades-old-risk-led-phenomenal-partnership-between-aws-and-intel-community\/401649\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.nextgov.com\/acquisition\/2024\/12\/decades-old-risk-led-phenomenal-partnership-between-aws-and-intel-community\/401649\/<\/a>; FedScoop Staff, \u201cCIA quietly awards C2E cloud contract possibly worth billions,\u201d FedScoop, November 20, 2020, <a href=\"https:\/\/fedscoop.com\/cia-quietly-awards-billion-dollar-c2e-cloud-contract\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/fedscoop.com\/cia-quietly-awards-billion-dollar-c2e-cloud-contract\/<\/a>; Jim Kelly, \u201cGoogle Public Sector Awarded $200M DoD CDAO Contract for AI &amp; Cloud Acceleration,\u201d Google Cloud Blog, July 14, 2025, <a href=\"https:\/\/cloud.google.com\/blog\/topics\/public-sector\/google-public-sector-awarded-200-million-contract-to-accelerate-ai-and-cloud-capabilities-across-department-of-defenses-chief-digital-and-artificial-intelligence-office-cdao\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/cloud.google.com\/blog\/topics\/public-sector\/google-public-sector-awarded-200-million-contract-to-accelerate-ai-and-cloud-capabilities-across-department-of-defenses-chief-digital-and-artificial-intelligence-office-cdao<\/a>; Ashley Roque, \u201cAnduril, Meta team up for Army IVAS recompete,\u201d Breaking Defense, May 29, 2025, <a href=\"https:\/\/breakingdefense.com\/2025\/05\/anduril-meta-team-up-for-army-ivas-recompete\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/breakingdefense.com\/2025\/05\/anduril-meta-team-up-for-army-ivas-recompete\/<\/a>; Nick Clegg, \u201cOpen Source AI Can Help America Lead in AI and Strengthen Global Security,\u201d Meta, November 4, 2024, <a href=\"https:\/\/about.fb.com\/news\/2024\/11\/open-source-ai-america-global-security\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/about.fb.com\/news\/2024\/11\/open-source-ai-america-global-security\/<\/a>; \u201cFlexTech Alliance Receives $75 Million Department of Defense Award To Create and Manage a Flexible Hybrid Electronics Manufacturing Facility,\u201d PR Newswire, August 28, 2015, <a href=\"https:\/\/www.prnewswire.com\/news-releases\/flextech-alliance-receives-75-million-department-of-defense-award-to-create-and-manage-a-flexible-hybrid-electronics-manufacturing-facility-300134578.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.prnewswire.com\/news-releases\/flextech-alliance-receives-75-million-department-of-defense-award-to-create-and-manage-a-flexible-hybrid-electronics-manufacturing-facility-300134578.html<\/a>. \u00a0<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref128\" title=\"\" id=\"_edn128\">[128]<\/a>. Department of Defense, \u201cDepartment of Defense Announces Joint Warfighting Cloud Capability Procurement,\u201d news release, December 7, 2022, <a href=\"https:\/\/www.defense.gov\/News\/Releases\/Release\/Article\/3239378\/department-of-defense-announces-joint-warfighting-cloud-capability-procurement\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.defense.gov\/News\/Releases\/Release\/Article\/3239378\/department-of-defense-announces-joint-warfighting-cloud-capability-procurement\/<\/a>; Lauren C. Williams, \u201cJWCC 1 year in: Military branches test the waters as DoD envisions cloud service 2.0,\u201d Breaking Defense, February 5, 2024, <a href=\"https:\/\/breakingdefense.com\/2024\/02\/jwcc-1-year-in-military-branches-test-the-waters-as-dod-envsions-cloud-service-2-0\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/breakingdefense.com\/2024\/02\/jwcc-1-year-in-military-branches-test-the-waters-as-dod-envsions-cloud-service-2-0\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref129\" title=\"\" id=\"_edn129\">[129]<\/a>. Nicole de Haay, \u201cCIA Awards Secret Multibillion-Dollar Cloud Contract,\u201d Nextgov\/FCW, November 20, 2020, <a href=\"https:\/\/www.nextgov.com\/modernization\/2020\/11\/exclusive-cia-awards-secret-multibillion-dollar-cloud-contract\/170227\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.nextgov.com\/modernization\/2020\/11\/exclusive-cia-awards-secret-multibillion-dollar-cloud-contract\/170227\/<\/a>; Frank Konkel, \u201cA decade-old risk led to \u2018phenomenal partnership\u2019 between AWS and the intel community,\u201d Nextgov\/FCW, December 14, 2024, <a href=\"https:\/\/www.nextgov.com\/acquisition\/2024\/12\/decades-old-risk-led-phenomenal-partnership-between-aws-and-intel-community\/401649\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.nextgov.com\/acquisition\/2024\/12\/decades-old-risk-led-phenomenal-partnership-between-aws-and-intel-community\/401649\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref130\" title=\"\" id=\"_edn130\">[130]<\/a>. David Marra et al., \u201cU.S. Army to use HoloLens technology in high-tech headsets for soldiers,\u201d Microsoft News, May 11, 2023, <a href=\"https:\/\/news.microsoft.com\/source\/features\/digital-transformation\/u-s-army-to-use-hololens-technology-in-high-tech-headsets-for-soldiers\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/news.microsoft.com\/source\/features\/digital-transformation\/u-s-army-to-use-hololens-technology-in-high-tech-headsets-for-soldiers\/<\/a>; Ashley Roque, \u201cArmy kickstarts possible recompete of Microsoft\u2019s $22 billion IVAS production deal,\u201d Breaking Defense, January 22, 2025, <a href=\"https:\/\/breakingdefense.com\/2025\/01\/army-kickstarts-possible-recompete-of-microsofts-22-billion-ivas-production-deal\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/breakingdefense.com\/2025\/01\/army-kickstarts-possible-recompete-of-microsofts-22-billion-ivas-production-deal\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref133\" title=\"\" id=\"_edn133\">[133]<\/a>. U.S. Government Accountability Office, Defense Industrial Base: Actions Needed to Address Risks Posed by Dependence on Foreign Suppliers, GAO-25-107283 (Washington DC: GAO, January 2025), <a href=\"https:\/\/www.gao.gov\/products\/gao-25-107283\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.gao.gov\/products\/gao-25-107283<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref134\" title=\"\" id=\"_edn134\">[134]<\/a>. Helen M. Berman et al., \u201cOutcome of a Workshop on Archiving Structural Models of Biological Macromolecules,\u201d Structure 14, no. 8 (2006): 1211\u20131217; National Institute of General Medical Sciences, \u201cProtein Structure Initiative: Achievements and Future Goals\u201d (Bethesda, MD: NIH, 2015).<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref135\" title=\"\" id=\"_edn135\">[135]<\/a>. Joseph A. DiMasi, Henry G. Grabowski, and Ronald W. Hansen, \u201cInnovation in the Pharmaceutical Industry: New Estimates of R&amp;D Costs,\u201d Journal of Health Economics 47 (2016): 20\u201333.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref136\" title=\"\" id=\"_edn136\">[136]<\/a>. Jumper et al., \u201cHighly Accurate Protein Structure Prediction with AlphaFold<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref137\" title=\"\" id=\"_edn137\">[137]<\/a>. \u201cAlphaFold Protein Structure Database,\u201d European Bioinformatics Institute, accessed January 2025; Samantha Robertson et al., \u201cThe Impact of AlphaFold on Structural Biology Research,\u201d Nature Reviews Drug Discovery 23 (2024): 112\u2013128.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref138\" title=\"\" id=\"_edn138\">[138]<\/a>. Nathan Baker et al., \u201cAccelerated Materials Discovery Using AI,\u201d Microsoft Azure Quantum Blog, January 9, 2024.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref139\" title=\"\" id=\"_edn139\">[139]<\/a>. \u201cAI Breakthroughs in Drug Discovery,\u201d Microsoft Research, January 12, 2024.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref140\" title=\"\" id=\"_edn140\">[140]<\/a>. Marco V. Perez et al., \u201cLarge-Scale Assessment of a Smartwatch to Identify Atrial Fibrillation,\u201d New England Journal of Medicine 381 (2019): 1909-1917.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref142\" title=\"\" id=\"_edn142\">[142]<\/a>. Benjamin A. Steinberg et al., \u201cThirty-Second Gold Standard Definition of Atrial Fibrillation and Its Relationship With Subsequent Arrhythmia Patterns,\u201d Circulation: Arrhythmia and Electrophysiology 11 (2018): e006274.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref143\" title=\"\" id=\"_edn143\">[143]<\/a>. Gerbrand Ceder and Kristin Persson, \u201cHow Supercomputers Will Yield a Golden Age of Materials Science,\u201d Scientific American, November 30, 2013.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref144\" title=\"\" id=\"_edn144\">[144]<\/a>. Amil Merchant et al., \u201cScaling Deep Learning for Materials Discovery,\u201d Nature 624 (2023): 80\u201385.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref146\" title=\"\" id=\"_edn146\">[146]<\/a>. \u201cExternal Validation of GNoME Predictions,\u201d Berkeley Lab Materials Project, November 29, 2023.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref147\" title=\"\" id=\"_edn147\">[147]<\/a>. Tian Xie et al., \u201cMatterGen: A Generative Model for Materials Design,\u201d Microsoft Research, January 2025.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref148\" title=\"\" id=\"_edn148\">[148]<\/a>. \u201cMatterSim: Materials Under Real-World Conditions,\u201d Microsoft Research Blog, January 15, 2025.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref149\" title=\"\" id=\"_edn149\">[149]<\/a>. Food and Agriculture Organization of the United Nations, \u201cThe State of Food and Agriculture 2023\u201d (Rome: FAO, 2023).<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref150\" title=\"\" id=\"_edn150\">[150]<\/a>. Peter Breugel et al., \u201cAgriculture\u2019s Connected Future: How Technology Can Yield New Growth\u201d (McKinsey &amp; Company, October 2020).<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref151\" title=\"\" id=\"_edn151\">[151]<\/a>. Amanda Ramcharan et al., \u201cDeep Learning for Image-Based Cassava Disease Detection,\u201d Frontiers in Plant Science 8 (2017): 1852.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref152\" title=\"\" id=\"_edn152\">[152]<\/a>. \u201cTensorFlow Lite for Mobile Deployment,\u201d Google AI Blog, 2023.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref153\" title=\"\" id=\"_edn153\">[153]<\/a>. \u201cImpact Assessment of AI Crop Disease Detection in East Africa,\u201d FAO Report, 2024.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref154\" title=\"\" id=\"_edn154\">[154]<\/a>. Ranveer Chandra et al., \u201cFarmBeats: An IoT Platform for Data-Driven Agriculture,\u201d NSDI \u201817, USENIX, 2017.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref155\" title=\"\" id=\"_edn155\">[155]<\/a>.\u201dFarmBeats Open Source Release,\u201d Microsoft Research, 2023.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref156\" title=\"\" id=\"_edn156\">[156]<\/a>. \u201cECMWF\u2019s New Supercomputer,\u201d European Centre for Medium-Range Weather Forecasts, 2022.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref157\" title=\"\" id=\"_edn157\">[157]<\/a>. Remi Lam et al., \u201cGraphCast: Learning Skillful Medium-Range Global Weather Forecasting,\u201d Science 382 (2023): 1416-1421.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref158\" title=\"\" id=\"_edn158\">[158]<\/a>. \u201cGraphCast Model Release,\u201d Google DeepMind Blog, December 14, 2023.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref159\" title=\"\" id=\"_edn159\">[159]<\/a>. \u201cDeepMind AI Reduces Google Data Centre Cooling Bill by 40%,\u201d Google DeepMind Blog, July 20, 2016.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref160\" title=\"\" id=\"_edn160\">[160]<\/a>. Jim Gao and Richard Evans, \u201cSafety-First AI for Autonomous Data Centre Cooling,\u201d Google DeepMind Blog, August 17, 2018.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref161\" title=\"\" id=\"_edn161\">[161]<\/a>. International Energy Agency, \u201cData Centres and Data Transmission Networks\u201d (Paris: IEA, 2023).<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref163\" title=\"\" id=\"_edn163\">[163]<\/a>. Michael Terrell, \u201cGoogle signs advanced nuclear clean energy agreement with Kairos Power,\u201d Google: The Keyword (blog), October 14, 2024, <a href=\"https:\/\/blog.google\/outreach-initiatives\/sustainability\/google-kairos-power-nuclear-energy-agreement\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/blog.google\/outreach-initiatives\/sustainability\/google-kairos-power-nuclear-energy-agreement\/<\/a>; X-Energy Reactor Company, \u201cAmazon Invests in X-energy to Support Advanced Small Modular Nuclear Reactors and Expand Carbon-Free Power,\u201d news release, October 16, 2024, <a href=\"https:\/\/x-energy.com\/media\/news-releases\/amazon-invests-in-x-energy-to-support-advanced-small-modular-nuclear-reactors-and-expand-carbon-free-power\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/x-energy.com\/media\/news-releases\/amazon-invests-in-x-energy-to-support-advanced-small-modular-nuclear-reactors-and-expand-carbon-free-power<\/a>; Constellation Energy, \u201cConstellation to Launch Crane Clean Energy Center, Restoring Jobs and Carbon-Free Power to The Grid,\u201d news release, September 20, 2024, <a href=\"https:\/\/www.constellationenergy.com\/newsroom\/2024\/Constellation-to-Launch-Crane-Clean-Energy-Center-Restoring-Jobs-and-Carbon-Free-Power-to-The-Grid.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.constellationenergy.com\/newsroom\/2024\/Constellation-to-Launch-Crane-Clean-Energy-Center-Restoring-Jobs-and-Carbon-Free-Power-to-The-Grid.html<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref166\" title=\"\" id=\"_edn166\">[166]<\/a>. \u201cNuclear Power in Russia\u201d (World Nuclear Association, accessed September 2025), <a href=\"https:\/\/world-nuclear.org\/information-library\/country-profiles\/countries-o-s\/russia-nuclear-power\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/world-nuclear.org\/information-library\/country-profiles\/countries-o-s\/russia-nuclear-power<\/a>; Xi Jinping and Vladimir Putin, \u201cLargest Sino-Russia nuclear project seals the deal in Beijing\u201d (China Atomic Energy Authority, June 8, 2018), <a href=\"https:\/\/www.caea.gov.cn\/english\/n6759361\/n6759362\/c6801897\/content.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.caea.gov.cn\/english\/n6759361\/n6759362\/c6801897\/content.html<\/a>; Matthew Kaminski, \u201cThe Other Nuclear Race: America Is Falling Behind China and Russia on Energy Innovation,\u201d Foreign Affairs, April 30, 2025, <a href=\"https:\/\/www.foreignaffairs.com\/china\/other-nuclear-race\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.foreignaffairs.com\/china\/other-nuclear-race<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref167\" title=\"\" id=\"_edn167\">[167]<\/a>. \u201cWaymo Safety Report,\u201d Waymo, 2023.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref168\" title=\"\" id=\"_edn168\">[168]<\/a>. Timothy Lee, \u201cWaymo\u2019s Safety Performance Over 7 Million Miles,\u201d Waymo Blog, December 20, 2023.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref169\" title=\"\" id=\"_edn169\">[169]<\/a>. National Highway Traffic Safety Administration, \u201cTraffic Safety Facts 2020\u201d (Washington, DC: NHTSA, 2022).<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref170\" title=\"\" id=\"_edn170\">[170]<\/a>. Matthias Winkenbach et al., \u201cThe Last-Mile Challenge in Urban Logistics,\u201d MIT Center for Transportation &amp; Logistics, 2021.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref171\" title=\"\" id=\"_edn171\">[171]<\/a>. \u201cLast Mile Routing Research Challenge,\u201d Amazon Science, Transportation Science 56, no. 1 (2022): 162-181.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref172\" title=\"\" id=\"_edn172\">[172]<\/a>. \u201cIndustry Adoption of Amazon\u2019s Routing Algorithms,\u201d MIT Sloan Management Review, October 15, 2023.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref173\" title=\"\" id=\"_edn173\">[173]<\/a>. Matteo Ippoliti et al., \u201cMany-Body Physics in the NISQ Era: Quantum Programming a Discrete Time Crystal,\u201d PRX Quantum 2 (2021): 030346.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref174\" title=\"\" id=\"_edn174\">[174]<\/a>. \u201cTime Crystal Created in Google Quantum Computer,\u201d Nature 601 (2021): 531\u2013536.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref175\" title=\"\" id=\"_edn175\">[175]<\/a>. Chetan Nayak et al., \u201cNon-Abelian Anyons and Topological Quantum Computation,\u201d Reviews of Modern Physics 80 (2008): 1083; Microsoft Quantum Team, \u201cAzure Quantum Demonstrates Path to Topological Qubits,\u201d Microsoft Research Blog, 2023.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref178\" title=\"\" id=\"_edn178\">[178]<\/a>. Stanford University\u2019s Venture Capital Initiative, Professor Ilya Strebulaev at Stanford Graduate School of Business.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref181\" title=\"\" id=\"_edn181\">[181]<\/a>. \u201cGoogle Ventures &#8211; 2025 Investor Profile, Portfolio, Team &amp; Investment Trends,\u201d Tracxn, accessed September 2025, <a href=\"https:\/\/tracxn.com\/d\/venture-capital\/google-ventures\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/tracxn.com\/d\/venture-capital\/google-ventures<\/a>; M12, \u201cPortfolio\u201d (M12 &#8211; Microsoft\u2019s Venture Fund, accessed September 2025), <a href=\"https:\/\/m12.vc\/portfolio\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/m12.vc\/portfolio\/<\/a>; Michelle Gonzalez, \u201cM12 ventures down a new path,\u201d The Official Microsoft Blog, January 5, 2023, <a href=\"https:\/\/blogs.microsoft.com\/blog\/2023\/01\/05\/m12-ventures-down-a-new-path\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/blogs.microsoft.com\/blog\/2023\/01\/05\/m12-ventures-down-a-new-path\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref186\" title=\"\" id=\"_edn186\">[186]<\/a>. Tiago S. Prado and Johannes M. Bauer, \u201cBig Tech Platform Acquisitions of Start-ups and Venture Capital Funding for Innovation,\u201d Information Economics and Policy (May 2022), <a href=\"http:\/\/dx.doi.org\/10.1016\/j.infoecopol.2022.100973\" rel=\"nofollow noopener\" target=\"_blank\">http:\/\/dx.doi.org\/10.1016\/j.infoecopol.2022.100973<\/a>; Salom\u00e9 Baslandze, Geoffrey A. Manne, and Silvana Krasteva, \u201cThe Signaling Value of Technology Venture Acquisitions\u201d (International Center for Law &amp; Economics, September 2024), <a href=\"https:\/\/papers.ssrn.com\/sol3\/papers.cfm?abstract_id=4943381\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/papers.ssrn.com\/sol3\/papers.cfm?abstract_id=4943381<\/a>; Joe Kennedy, \u201cMonopoly Myths: Is Big Tech Creating \u2018Kill Zones\u2019?\u201d (ITIF, November 2020), <a href=\"https:\/\/itif.org\/publications\/2020\/11\/09\/monopoly-myths-big-tech-creating-kill-zones\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/itif.org\/publications\/2020\/11\/09\/monopoly-myths-big-tech-creating-kill-zones\/<\/a>; Ian Hathaway, \u201cPlatform Giants and Venture-Backed Startups,\u201d October 12, 2018, <a href=\"http:\/\/www.ianhathaway.org\/blog\/2018\/10\/12\/platform-giants-and-venture-backed-startups\" rel=\"nofollow noopener\" target=\"_blank\">http:\/\/www.ianhathaway.org\/blog\/2018\/10\/12\/platform-giants-and-venture-backed-startups<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref187\" title=\"\" id=\"_edn187\">[187]<\/a>. \u201cGoogle buys YouTube for $1.65 billion,\u201d NBC News (from The Associated Press), October 10, 2006, <a href=\"https:\/\/www.nbcnews.com\/id\/wbna15196982\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.nbcnews.com\/id\/wbna15196982<\/a>; \u201cHistory of YouTube,\u201d Wikipedia, accessed January 2025, <a href=\"https:\/\/en.wikipedia.org\/wiki\/History_of_YouTube\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/en.wikipedia.org\/wiki\/History_of_YouTube<\/a>; Peter Forret, \u201cYoutube bandwidth: terabytes per day,\u201d May 4, 2006, <a href=\"https:\/\/blog.forret.com\/2006\/05\/04\/youtube-bandwidth-terabytes-per-day\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/blog.forret.com\/2006\/05\/04\/youtube-bandwidth-terabytes-per-day\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref191\" title=\"\" id=\"_edn191\">[191]<\/a>. \u201cAmazon and iRobot agree to terminate pending acquisition,\u201d iRobot Corporation and Amazon.com, Inc., news release, January 29, 2024, <a href=\"https:\/\/www.prnewswire.com\/news-releases\/amazon-and-irobot-agree-to-terminate-pending-acquisition-302046311.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.prnewswire.com\/news-releases\/amazon-and-irobot-agree-to-terminate-pending-acquisition-302046311.html<\/a>; \u201ciRobot Reports Fourth-Quarter and Full-Year 2024 Financial Results,\u201d iRobot Corporation, news release, March 12, 2025, <a href=\"https:\/\/www.prnewswire.com\/news-releases\/irobot-reports-fourth-quarter-and-full-year-2024-financial-results-302399203.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.prnewswire.com\/news-releases\/irobot-reports-fourth-quarter-and-full-year-2024-financial-results-302399203.html<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref192\" title=\"\" id=\"_edn192\">[192]<\/a>. Roborock vs iRobot vs Ecovacs &#8211; The Robot Vacuum Wars,\u201d Nanalyze, June 4, 2022, <a href=\"https:\/\/www.nanalyze.com\/2022\/05\/roborock-vs-irobot-vs-ecovacs\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.nanalyze.com\/2022\/05\/roborock-vs-irobot-vs-ecovacs\/<\/a>; \u201cChinese Robots Seize Overseas Market Share,\u201d EqualOcean, May 8, 2023, <a href=\"https:\/\/equalocean.com\/analysis\/2023050819714\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/equalocean.com\/analysis\/2023050819714<\/a>; \u201cRobot Vacuum Cleaner Market Accelarating [sic] at 17.2% CAGR,\u201d Statzon, December 18, 2024, <a href=\"https:\/\/statzon.com\/insights\/global-robotic-vacuum-cleaner-market\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/statzon.com\/insights\/global-robotic-vacuum-cleaner-market<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref193\" title=\"\" id=\"_edn193\">[193]<\/a>. Jake Blumgart, \u201cArm will reportedly lay off 15% of its workers following a failed Nvidia acquisition,\u201d PC Gamer, March 16, 2022, <a href=\"https:\/\/www.pcgamer.com\/arm-to-layoff-hundreds-of-employees-after-nvidia-buyout-failed\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.pcgamer.com\/arm-to-layoff-hundreds-of-employees-after-nvidia-buyout-failed\/<\/a>; Samuel Axon, \u201cARM will reportedly lay off up to 1,000 employees after NVIDIA sale falls through,\u201d Engadget, March 15, 2022, <a href=\"https:\/\/www.engadget.com\/arm-layoff-up-to-1000-employees-120604633.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.engadget.com\/arm-layoff-up-to-1000-employees-120604633.html<\/a>; \u201cNVIDIA and SoftBank Group Announce Termination of NVIDIA\u2019s Acquisition of Arm Limited,\u201d NVIDIA Corporation, news release, February 7, 2022, <a href=\"https:\/\/nvidianews.nvidia.com\/news\/nvidia-and-softbank-group-announce-termination-of-nvidias-acquisition-of-arm-limited\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/nvidianews.nvidia.com\/news\/nvidia-and-softbank-group-announce-termination-of-nvidias-acquisition-of-arm-limited<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref194\" title=\"\" id=\"_edn194\">[194]<\/a>. Jackie Wattles and Jill Disis, \u201cChina kills Qualcomm\u2019s $44 billion deal for NXP,\u201d CNN Business, July 26, 2018, <a href=\"https:\/\/money.cnn.com\/2018\/07\/26\/technology\/qualcomm-nxp-merger-china\/index.html\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/money.cnn.com\/2018\/07\/26\/technology\/qualcomm-nxp-merger-china\/index.html<\/a>; Ian King and Debby Wu, \u201cQualcomm Scraps NXP Deal Amid U.S.-China Trade Tensions,\u201d Bloomberg, July 26, 2018, <a href=\"https:\/\/www.bloomberg.com\/news\/articles\/2018-07-26\/qualcomm-to-scrap-nxp-deal-as-deadline-passes-for-china-approval\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.bloomberg.com\/news\/articles\/2018-07-26\/qualcomm-to-scrap-nxp-deal-as-deadline-passes-for-china-approval<\/a>; \u201cMerger Control in China Following the Termination of Qualcomm\/NXP,\u201d McDermott Will &amp; Emery, January 21, 2021, <a href=\"https:\/\/www.mwe.com\/insights\/merger-control-in-china-qualcomm-nxp\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/www.mwe.com\/insights\/merger-control-in-china-qualcomm-nxp\/<\/a>.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref195\" title=\"\" id=\"_edn195\">[195]<\/a>. Lisa E. Davenport, Jazz Diplomacy: Promoting America in the Cold War Era (Jackson: University Press of Mississippi, 2009), 141.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref199\" title=\"\" id=\"_edn199\">[199]<\/a>. Joseph S. Nye Jr., Soft Power: The Means to Success in World Politics (New York: PublicAffairs, 2004), 11\u201315.<\/p>\n<p class=\"MsoEndnoteText\"><a href=\"#_ednref204\" title=\"\" id=\"_edn204\">[204]<\/a>. Netflix, \u201cQ2 2024 Shareholder Letter,\u201d July 18, 2024, 2, <a href=\"https:\/\/s22.q4cdn.com\/959853165\/files\/doc_financials\/2024\/q2\/FINAL-Q2-24-Shareholder-Letter.pdf\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/s22.q4cdn.com\/959853165\/files\/doc_financials\/2024\/q2\/FINAL-Q2-24-Shareholder-Letter.pdf<\/a>; Cynthia Littleton, \u201cNetflix Subscribers Top 277.65 Million,\u201d Variety, July 18, 2024, <a href=\"https:\/\/variety.com\/2024\/digital\/news\/netflix-subscribers-277-million-q2-2024-earnings-1236076843\/\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/variety.com\/2024\/digital\/news\/netflix-subscribers-277-million-q2-2024-earnings-1236076843\/<\/a>; Annuario della TV (CeRTA), \u201cNetflix Makes America Great Again: Nearly 6 in 10 Hours Watched Come from U.S.-Produced Titles,\u201d March 2, 2025, <a href=\"https:\/\/annuariodellatv.it\/articoli\/netflix-makes-america-great-again-nearly-6-in-10-hours-watched-come-from-u-s-produced-title\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/annuariodellatv.it\/articoli\/netflix-makes-america-great-again-nearly-6-in-10-hours-watched-come-from-u-s-produced-title<\/a>; Netflix, \u201cWhat We Watched: The Second Half of 2024,\u201d February 26, 2025, <a href=\"https:\/\/about.netflix.com\/news\/what-we-watched-the-second-half-of-2024\" rel=\"nofollow noopener\" target=\"_blank\">https:\/\/about.netflix.com\/news\/what-we-watched-the-second-half-of-2024<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"Contents Key Takeaways 1 Introduction. 2 Societal Benefits 4 Recognize and Account for the \u201cFull Iceberg\u201d 33 Conclusion.&hellip;\n","protected":false},"author":2,"featured_media":204386,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[18],"tags":[23,3,21,19,22,20,25,24],"class_list":["post-204385","post","type-post","status-publish","format-standard","has-post-thumbnail","category-united-states","tag-america","tag-news","tag-united-states","tag-united-states-of-america","tag-unitedstates","tag-unitedstatesofamerica","tag-us","tag-usa"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/posts\/204385","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/comments?post=204385"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/posts\/204385\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/media\/204386"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/media?parent=204385"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/categories?post=204385"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/tags?post=204385"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}