Stargate is a staggering power grab.
OpenAI CEO Sam Altman has spent the past year seeking an absurd amount of computing power to train the company’s AI models — one report says Japanese officials literally laughed at the amount of electricity he demanded. The stakes were clear: without massive computing resources, OpenAI risked losing ground to tech giants like Google and Meta, who’ve spent years building their AI infrastructure.
But last week, this impossible dream became a press release. Altman secured a mind-boggling $500 billion commitment to build OpenAI’s data center empire, called Stargate, thanks to backing from SoftBank, Oracle, and the Abu Dhabi fund MGX. The White House added its stamp of approval in a press conference, with President Donald Trump flanked by Altman, Oracle cofounder Larry Ellison, and SoftBank CEO Masayoshi Son. (The name references a 1994 sci-fi movie, where the stargate is an ancient transportation portal controlled by an all-powerful ruler.)
If it materializes, Stargate could effectively be the largest private computing infrastructure project in history. It would mean a network of massive computing complexes — each spanning hundreds of acres and consuming as much power as a small city. Each facility would draw enough electricity to power tens of thousands of homes, requiring its own electrical substations and transmission lines. It’s a bid not just for computing dominance, but for control of a significant chunk of America’s energy infrastructure.
The deal relied on a few things happening in perfect order: Microsoft loosened its hold on OpenAI after Altman was briefly fired; Trump signaled to Silicon Valley elites that he was for sale; and tech’s biggest players started salivating at the chance to stamp their names on billion-dollar AI projects. It’s also a testament to Altman’s dealmaking prowess: a progressive San Francisco tech leader walked into an administration that opposed everything he publicly stood for, and within days, he secured a crown. He’s leveraging Trump’s desire for wins, SoftBank’s appetite for hype, and the AI arms race fear to secure unprecedented computing power for OpenAI alone, all within an intricate structure that makes it hard to track the money.
Despite all this, the foundations of Stargate look strikingly shaky. Altman’s getting a fraction of the money he initially sought from investors like SoftBank that have a reputation for putting hype above results. His announcement has been upstaged by a Chinese startup, DeepSeek, that quietly created an OpenAI-caliber model with a fraction of the resources. And some of the people who are usually eager to buy what Altman is selling are openly calling his bluff.
“We’ll need much more compute,” Altman reportedly told policymakers this week after DeepSeek blew a hole in Nvidia’s market cap — ultimately raising skepticism about Stargate. “There is a very real competition in the world, and we’re very excited about the next step here.”
The numbers don’t add up
Despite being touted as a landmark Trump-era project, Stargate actually surfaced last year as a Microsoft-OpenAI plan for a massive data center. But that partnership frayed as Microsoft grew wary of its OpenAI dependence and OpenAI struggled with computing costs. So, Altman traded Microsoft’s deep pockets for a more complex web of funding partners.
OpenAI and Microsoft maintain they haven’t broken up: “absolutely not!” Altman insisted on X, adding, “very important and huge partnership, for a long time to come. we just need moar compute.” Microsoft CEO Satya Nadella, for his part, told CNBC that this was an “adjustment” that Microsoft made “in order to support [Altman’s] needs while at the same time keeping the integrity of what we wanted as the strategic value” — corporate speak for reducing Microsoft’s risk. The drama has clearly gotten under Altman’s skin. He posted a grinning selfie with Nadella this week: their partnership is “gonna be much better than anyone is ready for.”
Either way, Microsoft either couldn’t or wouldn’t build the staggering infrastructure OpenAI wanted, and Altman took his pitch elsewhere. But skeptical observers noted that the numbers don’t add up. Stargate needs $100 billion in its first year alone, and its backers’ available cash seems to fall far short. SoftBank’s sitting on $30 billion in cash. MGX has a $100 billion fund but it’s already partly committed to OpenAI and may soon include competitors like xAI, though it can supposedly tap more capital. Oracle’s got just $11 billion on hand.
Meanwhile, OpenAI — perhaps the best-funded startup in history — is burning through cash like kindling. The startup expected to torch around $5 billion last year, after paying for major expenses like employee stock, according to financial documents reviewed by The New York Times. While Altman’s otherworldly dealmaking abilities may be able to scrape that first $100 billion together, it’s totally unclear how he’ll quintuple that.
On top of that, SoftBank has a complicated history with regard to its investing acumen — just look at the Vision Fund, which resulted in the company’s disastrous backing of WeWork.
“All I know is I’m good for my $80 billion.”
OpenAI and SoftBank will each commit $19 billion to Stargate, according to The Information. While they will be the largest backers, their exact ownership percentage isn’t clear — Altman has described them as general partners (GPs), meaning they will have significant control rather than simply an equity split. OpenAI and SoftBank are expected to hold a major stake and influence — reportedly around 40 percent each — with Oracle and MGX taking smaller positions. The total initial investment from all partners is $45 billion.
The rest is supposed to come from outside investors and debt financing, which could eventually trade like bonds. SoftBank is no stranger to debt — it already carries $150 billion on its books. And Altman will likely tap the United Arab Emirates for additional capital. Even so, there’s a massive gap between $45 billion and $500 billion, and even the most aggressive capital raising efforts have limits.
For now, though, the numbers are weird — and White House “First Buddy” Elon Musk, who cofounded OpenAI but left after a power struggle, wasted no time shouting out what everyone else was whispering. “They don’t actually have the money,” Musk posted soon after the announcement. “SoftBank has well under $10B secured. I have that on good authority.”
Altman bit back, replying that his post was “wrong, as you surely know” and boasting that Stargate’s first site is already underway. But that didn’t stop doubts from rippling through tech circles. In the CNBC interview, Nadella sidestepped questions about money, claiming he’s “not in the details” of investments in Stargate. When pressed about its funding, he only confirmed Microsoft’s annual $80 billion Azure investment. “All I know is I’m good for my $80 billion,” he said with a chuckle. Arm CEO Rene Haas gave Stargate’s financial foundation a vote of confidence, calling its backing “quite solid.”
Musk is obviously not a neutral observer given his ongoing lawsuit with OpenAI. And at this point, the conversation has devolved into a mix of petty sniping and bloviation. Trump brushed off questions regarding Musk’s funding concerns: “I don’t know if they do, but you know, they’re putting up the money. The government’s not putting up anything, they’re putting up money. They’re very rich people, so I hope they do,” he told reporters, adding that Musk “doesn’t like one of those people,” a reference to Altman.
None of this feels or looks right. OpenAI is bleeding billions annually despite its rapid growth, and its compute costs keep ballooning, with no sign of a stable business model in sight — it’s losing money even by charging customers $200 per month. At this burn rate, even the deepest pockets will eventually run dry. That’s why critics are side-eyeing Son’s involvement — he’s an investor known for throwing cash at moonshot ideas only to yank the funding when reality catches up. Perhaps an investor like Son is Altman’s last resort. Just look at the latest headline: it was reported that OpenAI is looking to nab another $25 billion from SoftBank in a new funding round that would value the startup at a jaw-dropping $340 billion.
As skepticism grows, we’re seeing increasingly grandiose promises, demonstrated in the Stargate announcement: AI-powered cancer vaccines designed in 48 hours, over 100,000 new American jobs, and US dominance in artificial intelligence. The venture aims to merge AI with biotechnology to revolutionize medical research, while simultaneously boosting American national security. The project’s stated mission is sweeping — to harness AI as a tool for human advancement. AGI is just around the corner, they’ll cure cancer, transform science — if investors just hand over a few more billion dollars.
“This is the beginning of a golden age,” Son said at the press conference.
$500 billion sounds massive, but it’s actually pretty modest when you look at what Stargate is trying to do.
According to Kent Draper, who helps lead data center operator IREN, Stargate’s financial goals hold up in a best-case scenario. You can build a lot of data centers on its first-year budget of $100 billion. The problem is, there’s no guarantee that money can be spent quickly and efficiently.
Most existing data center facilities aren’t equipped for AI infrastructure, particularly its gargantuan power requirements. A normal server rack — the kind that looks like a bookshelf full of computers — uses five to 10 kilowatts: as much electricity in a few hours as the average American home in a day. A modern AI GPU rack drawing 45 to 120 kilowatts can consume in a day what several apartments or even an entire small apartment building might use in a day. Current-generation facilities require at least 100 megawatts, and Draper estimates the next generation will require 200 kilowatts per rack, consuming as much electricity per day as dozens of homes.
You can’t just add more electricity here — older data centers simply “can’t handle that sort of power density,” Draper said. They lack both the electrical infrastructure to deliver that much power and the cooling systems needed to handle the intense heat these AI chips generate.
So Stargate will likely have to retrofit existing sites or find new ones, a process that’s neither easy, fast, nor cheap. It’s hard to find enough suitable locations to power large numbers of them up in one or two years, and the costs could vary significantly. “The numbers are pretty widespread depending on how you build it,” Draper said.
A single Tier 3 (or standard) facility can cost about $10–15 million per megawatt to build, and even at the conservative end, that means $1 billion for the basic data center infrastructure. Add in the GPUs and other AI hardware and the cost balloons to “about $3.5 billion” for a single 100-megawatt facility. As he puts it, “the numbers add up pretty quickly” — and this helps explain why Stargate’s budget targets are so massive.
Stargate is breaking ground with 10 data centers in Abilene, Texas — a location chosen likely for its untapped renewable energy potential, Draper said. Trump has pledged to fast-track construction through executive orders as the venture eyes expansion beyond Texas. While traditional data centers cluster in cities to minimize lag time for business software, AI facilities can prioritize power access over location, Draper added. (When the goal is to use the compute for inference, though, spacing of the facilities is trickier.)
If Stargate doesn’t get its full $500 billion in funding, we’ll likely end up with a scaled-back version: fewer facilities, smaller builds, longer construction timeline, or a focus on specific regions rather than nationwide deployment.
This would be far from the original vision of creating enough infrastructure to democratize access to AI computing power. Instead of transforming the entire industry, they’d just be patching the most urgent holes in the current system.
It’s easy to compare Stargate to the Foxconn project.
You may remember the promised $10 billion LCD factory, which was then scaled down dramatically well after Trump got a chance to do his public victory lap. Most promised jobs never materialized, and the project largely fizzled after the publicity.
It seems entirely plausible that Stargate will follow a familiar pattern: grand announcements, media hype, then reality. The $500 billion figure will get walked back, funding will dry up, or the project will be quietly restructured into something much smaller. Some data centers will probably get built, but nowhere near the scale promised. Then again, that’s what science fiction is, right? A glimpse of the future that’s always just out of reach.
Even if Stargate secures its full $500 billion and builds out its massive AI infrastructure, there’s a more crucial question: what if raw computing power isn’t the path to AGI? “We need a fundamentally new learning paradigm,” argues Databricks AI VP Naveen Rao. “More compute alone won’t get us there.” His skepticism gained weight recently when DeepSeek, a much smaller China-based AI lab, achieved o1-level performance allegedly using far less computing power by focusing on more efficient training methods.
This raises an unsettling possibility: Stargate could succeed perfectly at building its network of power-hungry data centers, only to discover it’s betting on the wrong horse. If better algorithms and smarter training methods — not brute-force computation — turn out to be the key to next-generation AI, we’re not just looking at a bubble. We’re looking at the biggest technological miscalculation in history: hundreds of billions spent on unnecessary infrastructure, creating a financial crater that could reshape tech investing for a generation.
As Sequoia, which invested in OpenAI, pointed out last year, AI’s rapid progress makes long-term data center bets precarious. Today’s cutting-edge clusters could be outdated long before they’re even fully deployed. Yet, the AI infrastructure race isn’t about rational, long-term planning; it’s about survival. “Imagine you knew for certain that AI was going to be as transformational as the internet, and that you control the only AI company in the world. How fast would you build CapEx?” Sequoia partner David Cahn wrote. “I believe the answer is: You would take your time.”
That is, in short, the massive bet AI leaders are making right now. It isn’t totally stupid to bet on scale, though. The scaling hypothesis — the idea that if you make an AI bigger and feed it more data and computing power, it gets smarter — is largely what has given us the best models we have today. A lot of this compute will be used for inference, too, so the models can handle millions of user requests simultaneously without crashing the chatbots. When we interviewed the ChatGPT team, the infrastructure lead said that “just keeping it up and running is a very, very big feat.” Stargate would, in theory, help a lot with that.
Silicon Valley operates on faith in visionaries
It also helps with bragging rights. Mark Zuckerberg, Musk, and Altman are seemingly in a data center measuring contest, constantly one-upping each other through social media posts and press releases. Obtaining the biggest data center empire has become as stylish with the tech elite as the latest designer bag. It represents each leader’s ability to wield not just computing power, but industrial might — a flex that says they can reshape the physical world as much as the digital one.
Silicon Valley operates on faith in visionaries, and Altman has masterfully cultivated that image. In front of cameras, Stargate’s investors are selling it as a way to cure diseases and solve big problems, but practically, it’s about OpenAI securing computing power to compete with rivals.
Paul Graham once wrote that if you dropped Altman on an island of cannibals, he’d return in five years as their king. Graham wasn’t talking about charm or genius; he was talking about pure adaptability and sheer survival instincts. With Stargate, Altman has transformed again — this time into a uniquely American figure: part industrialist, part tech visionary, part techno-populist champion. He’s promising not just AI advancement, but American renewal through massive infrastructure and job creation, tapping into both Silicon Valley’s fears of falling behind and Middle America’s hunger for industrial revival.
And while the failure of a $500 billion moonshot wouldn’t exactly be a scarlet letter — tech forgives those, just look at Adam Neumann — it could puncture something bigger than Altman’s reputation. The AI sector has been running on a potent mix of technical promises and industrial ambitions. If Stargate fails, it wouldn’t just be another startup flameout. It would be a reality check for an entire industry that’s been promising to reshape the world through sheer computational force.
For now, Altman has written himself into a classic science fiction narrative: the visionary promising to transform society through technological might. But like many sci-fi stories, the most interesting question isn’t whether the technology works — it’s what happens when human ambition collides with physical reality. In six months or a year, we’ll know if Stargate was the opening chapter of America’s AI revolution or just another techno-optimist fantasy that couldn’t survive contact with the real world.
