Hello, and welcome to TechScape. I\u2019m your host, Blake Montgomery. If you like reading our newsletter, forward this email to five friends with a demand they sign up like it\u2019s a chain letter warning of bad luck for five years. In this week\u2019s news, AI companies hit mind-boggling financial milestones such as a $5tn valuation, a $100bn quarter, and a string of deals worth nearly $600bn.<\/p>\n
The AI boom\u2019s incomprehensibly vast numbers make it hard to criticize<\/p>\n
Last week, the chipmaker Nvidia hit a valuation of $5tn. Just three months prior, it became the world\u2019s first $4tn company. Microsoft reached a $4tn valuation last week, as did Apple. On a smaller scale, Meta, Microsoft, Amazon and Alphabet all reported their quarterly earnings. The sums were vast. Google\u2019s parent reaped its first $100bn quarter. Amazon reported stellar growth in its cloud computing unit and added 13% to its stock price. Meta faced an unexpected tax bill of $16bn. All of the tech giants, except Apple, revised their capital expenditure figures upward, declaring to their investors that they would spend billions more on the real-world infrastructure that underpins artificial intelligence. The revisions added tens of billions to a total that\u2019s already in the hundreds of billions. Alphabet alone estimated its capital expenditure at between $91bn and $93bn for the upcoming year. That estimate is up from an original declaration of $75bn in February and a revised figure of $85bn announced in July.<\/p>\n
Not to be outdone by its publicly traded rivals, OpenAI, recently converted to a for-profit enterprise<\/a> and mulled an initial public offering at a valuation of $1tn. The world\u2019s most valuable startup has been on a dealmaking spree, signing agreements with Nvidia, which committed to investing $100bn in OpenAI in September; Microsoft, which signed an agreement in early October for OpenAI to spend $250bn on Azure cloud services; and Oracle, another cloud computing giant that OpenAI agreed to spend on in September, this time $300bn. On Monday, the ChatGPT-maker announced a deal with Amazon Web Services for $38bn. In total, OpenAI has promised to spend $588bn in the coming years.<\/p>\n Nvidia\u2019s valuation is now larger than Germany\u2019s entire annual economic output in 2025 ($4.66bn). Put another way, the market capitalization of Nvidia is two and a half times the market capitalization of every publicly traded company in Germany put together, roughly $2.04tn in 2024, per the World Bank. How can one company be worth more than the world\u2019s third largest economy, a nation of 83.5 million people, a country whose economy forms the financial backstop of an entire continent?<\/p>\n Read more: Boom or bubble? Inside the $3tn AI datacentre spending spree | Artificial intelligence (AI) | The Guardian<\/a><\/p>\n The difficulty in comprehending the financials of the AI boom makes it difficult to criticize with clarity or force. What could I say to an avalanche? Even the most insightful analysis seems like it will be bowled over and crushed beneath the weight of a billion-dollar datacenter. All of these numbers buck understanding. There is nothing in your individual human life that you could compare them with. How would I spend $91bn? How would I make hundreds of billions of choices? Thinking about them feels bizarre. Describing Meta\u2019s earnings as \u201cmixed\u201d is strange, though they were, by Wall Street experts\u2019 estimates.<\/p>\n The circular, multibillion-dollar deals between the companies atop the boom have given rise to fears of overinflated value and financial fragility. Should one company falter, it seems the others might follow suit and bring the US economy down with them. So far, they show no signs of stopping their joint rampage.<\/p>\n An Nvidia Blackwell GPU is displayed at Computex in Taipei, Taiwan, last year. Photograph: Ann Wang\/Reuters<\/p>\n On the more populist side of that criticism are reports that AI has yet to find an essential use case other than cheating on homework assignments. It can\u2019t adequately replace workers, no matter how many of them a CEO might lay off. About 95% of AI pilots conducted in businesses to date have failed, MIT researchers found<\/a> in August.<\/p>\n The financials of the AI boom are so enormous as to defy comprehension; its digital size is immense as well. Large language models such as ChatGPT, Claude Sonnet and others function in part via parameters, variables that define how the model makes predictions of what word should come next. These unseen knobs tweak the response number in the hundreds of billions, with some reports indicating GPT-5 might make use of a trillion of them.<\/p>\n AI\u2019s physical imprint on the world matches the magnitude of the technology\u2019s financial scale. Dara Kerr, one of the Guardian\u2019s technology reporters, visited the Tahoe-Reno Industrial Center, home to the largest data center in the US and several of its smaller siblings, on a reporting trip last week. Here she describes its unbelievable scale:<\/p>\n The Tahoe-Reno Industrial Center stretches from Interstate 80 up into the dry Nevada desert mountains. The complex spans tens of thousands of acres and is home to nearly 200 companies, both for fulfillment and logistics operations and tech data centers. That includes Google, Microsoft and Tesla. Some businesses have several data centers, each multiple football fields long, that snake up the desert valleys. The industrial center\u2019s landmass makes up 65% of the county\u2019s territory. It\u2019s so big, it\u2019s almost hard to comprehend.<\/p>\n Should you buy that gadget?skip past newsletter promotion<\/a><\/p>\n A weekly dive in to how technology is shaping our lives<\/p>\n Privacy Notice: Newsletters may contain information about charities, online ads, and content funded by outside parties. If you do not have an account, we will create a guest account for you on theguardian.com<\/a> to send you this newsletter. You can complete full registration at any time. For more information about how we use your data see our Privacy Policy<\/a>. We use Google reCaptcha to protect our website and the Google Privacy Policy<\/a> and Terms of Service<\/a> apply.<\/p>\n after newsletter promotion<\/p>\n You will be able to bet on US elections by giving money to the US presidentDonald Trump. Photograph: President Donald Trump via truth Social and Reuters<\/p>\n Donald Trump\u2019s Truth Social is partnering with Crypto.com on a feature for betting on election results, the parent company announced in a press release<\/a> last week. The Trump Media and Technology group will introduce its \u201cTruth Predict\u201d feature, which will allow users \u201cto trade prediction contracts related to major events and milestones, such as political elections, interest and inflation rate changes, commodity prices on gold and crude oil, events across all major sports leagues, and more\u201d using the new product, though when it will go live is unclear.<\/p>\n Devin Nunes, TMTG\u2019s head, said of the feature: \u201cFor too long, global elites have closely controlled these markets \u2013 with Truth Predict, we\u2019re democratising information and empowering everyday Americans to harness the wisdom of the crowd, turning free speech into actionable foresight.\u201d In classic Trumpian fashion, Nunes hypocritically gnashes his teeth at \u201cglobal elites\u201d from a perch high in the eyrie of US power.<\/p>\n Something feels off about this arrangement, like if the boxer in the fight was also the bookie. Trump is the president of the United States and is asking his constituents to put money on the outcome of elections that he will be campaigning in. He is in control of the policies that can determine the rates users bet on. Though he\u2019s heading towards his two-term limit in the presidency, he has flirted with an unconstitutional third term<\/a>. Would US citizens bet on whether he can make it happen?<\/p>\n