OpenAI Is Now the Tech Economy's Atlas, Holding Valuations on its Shoulders
As OpenAI's ambitions grow, so do the ripple effects of any stumbles
Last Tuesday OpenAI brought a handful of reporters to Abilene, Texas, to gawk at its sprawling new data center and hear about the scope of its ambitions. The setting was apt: in the state that gave the world Texas Hold ’Em, the company was making clear just how high the stakes have become, and how much more money it will need to stay in the game.
The day before, Nvidia said it would sink $100 billion into OpenAI. A few weeks before that was the revelation of OpenAI’s $300 billion Oracle contract. There’s also the $40 billion SoftBank investment tied to the Stargate venture — a cascade of deals that together suggest the company is now playing at a scale where only the biggest players can afford a seat.
To make all of this work will require a nearly constant supply of new capital, mostly toward data center construction. And there are, fortunately for OpenAI, a lot of companies that want to make sure OpenAI is well supplied with cash.
The more that OpenAI has established itself at the nexus of the tech industry, the more its deal announcements and partnerships start to resemble a sort of mutual destruction pact: should the markets lose faith in OpenAI, many of its partners and rivals could see their valuations plunge. Companies with trillion dollar valuations could stand to lose more than the value of OpenAI (a reported $500 billion) if investors stop believing in the load bearing AI startup. (We’ve already seen stocks swing wildly when Altman decides to make a dour pronouncement.) OpenAI will be able to raise money because there’s a growing cadre of companies whose valuations would plummet if it fails.
We’ve got a rundown of the key companies hanging on OpenAI’s success and in many cases desperate to keep the cash hungry startup well funded to keep this AI mania running hot.
Oracle
Market cap: $803.5 billion
OpenAI connection: Contracted to receive $300 billion from OpenAI over five years
In mid-2024, Satya Nadella made the call not to go-in on OpenAI’s massive Stargate data center project (or “Mercury,” as it was known internally at Microsoft). He then gave the company a waiver to seek alternative providers for data centers. Since then, no company has benefitted more than Oracle.
At the time, the legacy database provider was thirsty to land a major client that could validate its improving but still underwhelming cloud computing business. Led by Clay Magouyrk, a hard driving Oracle executive who’s known for pushing his team to meet accelerated deadlines, the cloud unit was up for doing whatever to land AI’s golden child. Magouyrk promised they could turn around a massive data center in Abilene at record speed. It’s telling that Magouyrk was named Oracle’s co-CEO last week now that the 48-year-old business its all about the cloud.
With OpenAI set to pay the company $300 billion over five years, Oracle has already set about raising the cash to make that kind of a build-out possible. So far it’s gone pretty smoothly. The company raised $18 billion through a corporate bond sale; almost certainly there will be more to come as its build-out continues.
After skyrocketing since its earnings announcement, Oracle stock has been down more than 10% over the past week as investors try and suss out exactly how much the company is worth thanks to the OpenAI deal.
Now OpenAI needs Oracle to deliver as much as Oracle needs OpenAI to keep raising more money. If the latter starts to become a problem, no public company will be hit harder than Larry Ellison’s.
Nvidia
Market cap: $4.4 trillion
OpenAI connection: Fresh $100 billion OpenAI funding commitment
Jensen Huang’s company has already benefitted more from the ChatGPT moment than almost any other. Its shares are up more than 1,000% since the day OpenAI’s chatbot came out in November 2022.
But very little of the company’s chip sales have gone to OpenAI directly. Nvidia’s biggest customers have been tech giants like Microsoft and Meta—until recently.
While the partnership between Nvidia and OpenAI got headlines for its astonishing $100 billion commitment, the arguably bigger announcement is that OpenAI was going into the data center building game itself. This has the company dealing directly with Nvidia and taking on all the risks involved. Altman and Huang already had a tight relationship, but this brought them even closer.
Nvidia certainly has the cash, and ensuring the medium term viability of the most important company in AI is a huge win. The GPU-maker had already been one of the most prolific investors in seeding startups to essentially prop up the industry and keep demand high.
The investment is also an insurance payment to keep OpenAI an Nvidia shop. OpenAI’s chief rival, Anthropic, mostly uses Google’s tensor processing units (TPUs) as well as Amazon’s proprietary AI chips. Google’s Gemini chatbot also mostly uses TPUs. It would be worrisome to Huang if OpenAI ever went seriously away from Nvidia chips and $100 billion essentially means that the two will be connected in good times and bad.
Microsoft
Market cap: $3.8 trillion
OpenAI connection: Major OpenAI stakeholder
You could imagine Microsoft executives staring longingly at the Oracle stock bump after the OpenAI contract hit. Indeed in the next few days we got messages from a number of Microsoft insiders who were calling it a huge miss for the company that to date had been more responsible for OpenAI’s success than any other tech giant.
But in hindsight Microsoft may end up looking like the more rational actor. Nadella’s decision not to buy-in on Stargate was an acknowledgement that they have a real cloud business to run beyond supporting OpenAI.
To be clear, Microsoft’s AI cloud revenue is still fueled by OpenAI. The company stopped breaking out the percentage of its Azure revenue that comes from AI, likely because that number stopped growing, but estimates from investment bank Stifel Financial put quarterly Azure AI revenue at $4 billion. Almost all of that comes from usage of ChatGPT and OpenAI’s APIs.
But Azure itself does $75 billion a year in revenue and has customers like Walmart and Verizon. Unlike Oracle, they don’t want or need to be so dependent on a single customer.
Microsoft has taken steps to lessen its reliance on OpenAI’s models for its own software; it recently started using Anthropic’s models for some tasks within its Microsoft 365 Copilot. And there are currently 1,800 different AI models available to use on Azure.
It’s true that in a scenario where OpenAI becomes the trillion dollar transformational company Altman has been envisioning, Microsoft will not see as great an upside as they would have. But a diversified Microsoft is also, arguably, a safer place. It’s in many ways a return to the company’s comfort zone of being an old-line tech company run by financialists.