Nvidia plans $250bn push to bolster OpenAI’s infrastructure ambitions

Published 27 July 2026 · tech

Political backlash grows as US states propose bans on new data centres, raising challenges for Nvidia and the AI industry. Semiconductor giant Nvidia is reportedly

Political backlash grows as US states propose bans on new data centres, raising challenges for Nvidia and the AI industry. Semiconductor giant Nvidia is reportedly undergoing talks with OpenAI to provide funding guarantees for a data centre in Ohio. The $250bn deal, which was first reported by The Wall Street Journal, would help the ChatGPT owner lease a 10-gigawatt project that SB Energy, a subsidiary of SoftBank, is building in Piketon, Ohio, 68 miles (109km) south of Columbus, Ohio. The project could cost more than $500bn, including chips that would go inside the data centre, and is expected to deliver as much as 800 megawatts of electricity by 2028, which is enough to power 640,000 homes. The power supply will come from a separate $33bn US government deal with Japan for a natural gas plant. The $250bn guarantee covers the lease for the centre but does not cover the chips within the centre, which are worth another $350bn. However, Jensen Huang-led Nvidia said it is also in talks for that as well. In June, The Information reported that Sam Altman-led OpenAI signed a potential 20-year lease with AI controlling both its own equipment and payment of the facilities. The Ohio centre where OpenAI will have space if the deal ultimately goes through will be a public-private partnership in which the US Department of Energy (DOE) allowed SoftBank to build the world’s largest AI data centre on leased land. Ohio with 166 centres and another 57 planned has the fourth most operational data centres in the United States after Virginia, Texas and California. The deal would open OpenAI’s infrastructure ambitions as it has otherwise been renting from other tech giants, including Amazon, Oracle and Microsoft. But analysts questioned if these multibillion-dollar commitments could be fruitful for a start-up that is valued at $852bn when it is unprofitable and its partners – firms that provide data centres, chips and processing power – held as much as $96bn of debt as of November.

Financing concerns Critics like Aleksandar Tomic, associate dean at Boston College, said the newly reported deal is the latest example of circular financing across the AI industry that blurs what is legitimate market demand. “What is happening right now with OpenAI and others is that they have the need for computing, but apparently they don’t really have the revenue or the financial capability to engage in the capital expenditures necessary to support their activities,” Tomic said. “Nvidia steps in and provides some funding so they can continue buying Nvidia chips. Where the demand for the whole chain is coming from or where the money is coming from becomes a really big issue.” In 2025 alone, OpenAI said it would buy $250bn of cloud services from Microsoft, which happens to be one of the AI giant’s backers. Also in 2025, Nvidia said it would invest $100bn into OpenAI to build out data centres that would use its chips. In early 2026, Nvidia pledged $30bn to OpenAI. Companies including Nvidia, Microsoft and Oracle have invested billions into AI developers only for them to become large purchasers of their cloud computing services, which results in the very same capital going through multiple companies, appearing as revenue, according to Tomic. This is akin to 1999, Tomic says, when companies would buy products from each other, making demand look stronger than it was, which was ultimately one of the factors that led to the dot-com bubble. “Companies would engage in circular financing deals where they would essentially buy from each other and make it look like there was greater demand for their services than there really was. At some point, the money runs out,” Tomic said. “In other words, the demand is not as big as it appears to be because the companies are buying from each other, using their own money to some degree, as opposed to OpenAI having tremendous customer demand, monetising it properly and then using customers’ money to buy Nvidia chips.