OpenAI's Bet to Survive the Bubble: Spend Nearly $300 Billion More
OpenAI has announced a $38 billion deal with Amazon Web Services, securing immediate access to AWS EC2 UltraServers, hundreds of thousands of Nvidia chips, and the ability to scale to tens of millions of CPUs. This seven-year commitment comes on the heels of a $250 billion agreement with Microsoft Azure that also granted OpenAI permission to diversify beyond its exclusive Azure relationship. Combined with Microsoft's $135 billion stake (approximately 27% equity) in the company, OpenAI has now committed nearly $300 billion in infrastructure spending while simultaneously securing critical backing from the two cloud giants most likely to survive any industry shakeout.
The timing carries a certain irony—both AWS and Azure experienced service disruptions last month, and OpenAI just launched its own web browser to compete with established players. While these massive deals likely took months to negotiate, the convergence of events underscores OpenAI's clear signal: they want options to expand, and expand quickly, without depending on a single "basket." Undoubtedly, several companies caught in last month's outages wished they'd made similar diversification moves.
The Bubble Everyone Admits Exists
But now let's address the elephant in the data center: AI is in a bubble—very much like the dot-com bubble of decades past. We're constantly told that AI will change our lives, that we need AI, that AI can solve problems X, Y, and Z. Companies are rushing either to build the next big AI or to shoehorn AI into their existing products, whether it makes sense or not.
Don't get me wrong—AI can be a very useful tool in the right use case. But it's not the answer to every problem, despite the marketing hype suggesting otherwise. This has led to far too many AI companies for the industry to actually support. Even Sam Altman himself admits it's a bubble, comparing it directly to the dot-com era. Bill Gates has made similar warnings. Goldman Sachs reports that bubble concerns are "arguably more intense than ever."
The economics tell the story: AI revenue doesn't even cover the **depreciation costs** of the infrastructure required to run it. **62% of employees believe AI is overhyped.** A U.S. Census Bureau survey shows **declining AI adoption rates** despite the massive spending. It's eventually going to pop, and many companies will be lucky to survive.
The $288 Billion Question
So if that's the case, why on Earth is OpenAI spending almost $300 billion?
The answer reveals a calculated survival strategy with several layers:
First, they're guaranteeing their own survival by becoming the biggest player in the game. By locking in seven years of capacity, they ensure access to compute resources even when scarcity hits after the bubble pops.
Second, by throwing massive money at both Amazon and Microsoft—including that $135 billion stake—they're ensuring neither company can easily afford to let OpenAI fail. Whether this ends up looking like a pivot or OpenAI emerges as the sole survivor doesn't matter. They'll be around in some form, and regardless, the infrastructure will be useful for whatever they do next.
Third, they're making themselves indispensable. AWS now offers OpenAI's open-weight models, giving Amazon something they previously lacked in AI offerings. Microsoft maintains exclusive access to OpenAI's best models while holding 27% equity. Both tech giants are now deeply invested in OpenAI's success.
In short, AI companies are playing a high-stakes (nearly Squid Games level) game of musical chairs, and OpenAI is positioning itself with an RSVP'd seat—while, fun fact, looking to fire up enough data center energy draw to power whole nations.
The Environmental Cost Nobody Wants to Discuss
Data centers already account for **4% of total U.S. electricity use**, with demand expected to more than double by 2030. Global consumption could reach **620-1,050 terawatt-hours by 2026**—equivalent to France's entire electricity consumption.
Large data centers consume **up to 5 million gallons of water per day**, equivalent to towns of 10,000-50,000 people. A single Iowa data center consumed **1 billion gallons in 2024**—enough to supply all Iowa residences for five days.
OpenAI's $288 billion commitment translates to infrastructure that will consume these resources 24/7/365 for the next seven years, all for technology that most employees think is overhyped.
Forcing Everyone's Hand
By putting in this hard "we're in it for keeps" stance, OpenAI is forcing other AI companies to raise, call, or fold—thus pushing us closer to the inevitable collapse. Why? Because there are only so many instances that can be fired up and only so much physical space to run the machines that power this technology. Either get a seat or get left behind.
Economics dictates that as resources become scarcer, prices will rise. At some point, juggling all this infrastructure alongside the rest of the web, something will have to give and the whole thing will crash down.
OpenAI is positioning itself to be the guy retelling the Tom Toro cartoon from The New Yorker (November 26, 2012)—the one showing businessmen in suits around a post-apocalyptic campfire, with the caption:
"Yes, the planet got destroyed. But for a beautiful moment in time we created a lot of value for shareholders."
Except this time, it's not satire. It's a $288 billion business plan.
The Bottom Line: The question isn't whether the AI bubble will pop—even the people inflating it admit that. The question is: when does it burst, who's left standing, and was the environmental and economic cost worth creating "shareholder value" for that beautiful moment in time?
What do you think about OpenAI gamble? I'd love to hear your thoughts—find me on Mastodon at @ppb1701@ppb.social and let's talk about it!
Part of the ongoing TheranasAI series, a sub-series of Big Tech's War on Users.
Read the terms. They're more honest than the marketing.