This week, Openai and Oracle shocked the market with a $300 billion five-year contract, part of a surge in new businesses that spiked cloud provider stock. But perhaps the market wasn’t surprising. This agreement reminds us that despite Oracle’s legacy status, the company still plays a major role in AI infrastructure.
On Openai’s part, the agreement is more clear than the lack of details suggests. One is that startups willing to pay large amounts of money to computing to measure startup appetite.
Chirag Dekate, vice president of research firm Gartner, told TechCrunch it was clear why both parties are interested in the deal. It makes sense for Openai to work with several infrastructure providers, he pointed out. It also diversifies the company’s infrastructure, spreads risk to several cloud providers, and gives Openai the scaling advantages compared to its competitors.
“Openai seems to be bringing together one of the most comprehensive global AI supercomputing infrastructures for extreme scale. “This is very unique. This is probably a sign of what the model ecosystem looks like.”
Watchers from some industry expressed surprise that Oracle is involved, citing the company’s reduced role in the AI boom compared to cloud rivals such as Google, Microsoft Azure and AWS. But Dekate argues that observers shouldn’t be so surprised. Oracle has previously worked with HyperSchool, providing infrastructure for Tiktok’s considerable US business.
“For decades, they have actually built core infrastructure capabilities, allowing them to deliver extreme scale and performance as a core part of their cloud infrastructure,” says Dekate.
Payment and power
But even as the stock market celebrates the trade, it lacks important details and leaves behind questions about electricity and payments.
TechCrunch Events
San Francisco
|
October 27th-29th, 2025
Openai has announced a series of infrastructure investments over the past year. Openai is committed to spending around $600 billion a year on calculations from Oracle and $10 billion to develop custom AI chips using Broadcom.
Meanwhile, Openai said in June that it had reached $10 billion a year of recurring revenues, from around $5.5 billion last year. Those figures include revenue from the company’s consumer products, ChatGPT business products, and its APIs. Also, CEO Sam Altman paints a rosy picture of the future in terms of subscribers, products and revenue, but the company burns billions of dollars in cash each year.
Power is another question, or more specifically, a plan to procure the energy needed for a company to perform this level of calculation.
Industry observers are predicting a short-term boost for natural gas, but solar and batteries are probably better positioned to deliver faster and lower costs in many markets. High-tech companies are also making big bets on their nukes.
Despite the headlines of market movement, the energy impact of Openai’s expected growth is not at all unexpected. According to a report released by Rhodium Group, data centers are expected to consume 14% of all US electricity by 2040.
Calculation has always been a constraint for AI companies, so investors have purchased thousands of Nvidia chips to allow startups to access the power they need. Andreessen Horowitz reportedly bought over 20,000 GPUs, while Nat Friedman and Daniel Gross rented access to a 4,000 GPU cluster (although Meta currently owns it).
But calculations are worthless without power. To keep the data centers juiced, large tech companies are snapping solar power plants, buying nuclear power plants, and signing contracts with geothermal startups.
So far, Openai has been relatively quiet in that respect. CEO Sam Altman has made several notable bets in the energy sectors such as Oklo, Helion and Exowatt, but the company itself hasn’t thrown money into spaces like Google, Meta, Amazon and more.
4.5 Gigawatt computing trading could change soon.
The company can play an indirect role and pay Oracle extensive experience to handle the physical infrastructure just as Oracle invested in a startup that suits Openai’s future power needs. It leaves “asset light” to the company. This will undoubtedly please the investor and help match its rating with other software-centric AI startups.
