Over the past two weeks, four separate headlines may look unrelated, but my bet is they’re adding up to something\u00a0much\u00a0larger. <\/p>\n
First, Elon Musk committed SpaceX exclusively to NVIDIA GPUs during the company’s first conference call on August 4th. Second, Musk set a target of 10 gigawatts of AI compute by the end of 2027. Third, on August 10, NVIDIA announced financing partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to mobilize over $500 billion of third-party capital. And fourth, NVIDIA disclosed roughly 122.8 million SpaceX shares worth about $21 billion in its Q2 13F filed August 14, 2026. <\/p>\n Together, they tell one story: the buildout Jensen Huang calls “the largest infrastructure expansion in human history” now has a defining partnership. <\/p>\n Putting Together the Partnership That Could Define the AI Factory Buildout <\/p>\n NVIDIA (NASDAQ:NVDA<\/a>) sells the GPUs, networking, and full-stack software behind essentially every frontier AI model. Data Center revenue reached $75 billion last quarter, up 92% year over year. <\/p>\n SpaceX (NASDAQ:SPCX<\/a>) is the newest hyperscale AI buyer. Beyond rockets and Starlink, its AI segment (Grok, Colossus II, cloud services) delivered $2.56 billion in Q2 revenue, up 247% year over year, with $15.83 billion of a $18.37 billion single-quarter capex bill directed to AI infrastructure. Here’s why those numbers will soon look tiny.\u00a0 <\/p>\n On August 4th, Musk spoke of driving SpaceX to 10 gigawatts of computing capacity by the end of 2027, which is up from 1.4 gigawatts today. The cost of getting SpaceX to 10 gigawatts by the end of 2027? Somewhere around $500 billion. <\/p>\n From Our Partners <\/p>\n Written by Insiders. Answering to No One. <\/p>\n Before Doomberg published a word, its team spent long careers in heavy industry, private equity, and the hard sciences. They take no advertisers and serve no institution \u2014 which is why their lateral-thinking coverage of energy, finance, and geopolitics reads nothing like consensus financial media. Doomberg has set aside a discounted rate exclusively for 24\/7 Wall St. readers \u2014 it isn’t available on their main page. <\/p>\n Story Continues <\/p>\n Claim the 24\/7 Wall St. Rate \u2192<\/a> <\/p>\n The day after earnings, the market didn’t ‘buy’ Musk’s vision. SpaceX shares fell. Several Wall Street firms questioned how Musk could possibly receive the financing to build out the scale of compute he was discussing. However, a couple of days later (August 7th), SpaceX shares began a massive rally that soon brought the company back above its IPO price. <\/p>\n The catalyst for the rally? Research from\u00a0SemiAnalysis stated that SpaceX was actually in position to achieve its goals and reach up to 10 gigawatts by the end of 2027. If SpaceX hit that target,\u00a0SemiAnalysis\u00a0predicted they would exit 2027 with an ARR of around $305 billion. <\/p>\n Wall Street is currently modeling $98 billion in 2027 revenue and $157 billion in 2028 revenue. <\/p>\n NVIDIA Announces $500 Billion in Funding <\/p>\n Then, on August 10th, NVIDIA announced a brand new ‘Compute Infrastructure Financing Platform’ alongside Wall Street’s largest banks. The platform would mobilize over $500 billion in third-party capital, with NVIDIA agreeing to backstop up to 25% of the project’s cost. <\/p>\n And finally, NVIDIA’s 13F released on August 14th revealed the company had established a $21 billion position in SpaceX. <\/p>\n First, we had SpaceX announcing its intention to build a massive amount of data centers (exclusively using NVIDIA’s chips), and then NVIDIA announcing a $500 billion financing initiative and large investment into SpaceX’s stock. <\/p>\n I trust you’re following along with the sequence of events. NVIDIA’s CEO Jensen Huang has repeatedly expressed his admiration for Elon Musk’s drive and ability to stand up data centers in record time. Musk is now targeting a data center buildout that is extremely\u00a0ambitious, but one which faces challenges in financing\u00a0Musk’s ambitions. <\/p>\n It’s a partnership that makes sense. NVIDIA has been looking for more ways to use its balance sheet to develop the AI ecosystem, and Musk gives them a ‘backdoor’ to the creation of a new hyperscaler. The company’s $500 billion platform becomes a way for not just SpaceX, but other neoclouds, to receive more capital and compete with the largest companies like Amazon, Alphabet, Microsoft, and Meta Platforms. <\/p>\n And by creating a new funding source for this group, NVIDIA continues to diversify its customer list. The fact that NVIDIA’s revenue remains so concentrated amongst a small group of customers remains one of the largest reasons the company trades near a market-average forward P\/E despite its incredible growth rates. <\/p>\n The Future of AI Hinges on Elon Musk and Jensen Huang’s Partnership <\/p>\n Jensen Huang has said he expected AI infrastructure spending to reach $3 trillion to $4 trillion annually by the end of the decade. When he started using that number on conference calls, many analysts mistakenly believed it was an amount of\u00a0total\u00a0spending across the next five years. <\/p>\n AI infrastructure estimates vary, but they’re generally around $800 billion in 2026. Consensus continues to grow for 2027, but many estimates place spending next year at closer to $1.2 trillion. That number is eye-watering\u00a0for those who have been following AI, but it’s still just 1\/3 of where Huang sees the market going by the end of the decade. <\/p>\n So, it’s clear that Huang envisions a future for AI that’s much larger and happens much sooner than just about anyone else imagines. Well, anyone else aside from Elon Musk. <\/p>\n My bet: the budding partnership between NVIDIA and SpaceX will become a key driver across the next generation of the AI race. What we’ve seen the past two weeks is just the beginning. <\/p>\n What $28 a Month Buys a Serious Investor <\/p>\n Through this reader link<\/a>, 24\/7 Wall St. readers get their first year of Doomberg for $332 \u2014 17% off the standard $400, which works out to about $28 a month. That buys six to eight deeply researched articles a month on energy, finance, and geopolitics, the full archive, and one of the most resource-rich comment sections anywhere, where engineers, executives, and fund managers add context under every piece. After the first year it renews at the standard rate, and you can cancel anytime. If today’s article touched energy or the macro picture, odds are the green chicken has already gone three layers deeper \u2014 see for yourself at the discounted rate<\/a>. <\/p>\n<\/a> Shutterstock <\/p>\n