Stargate was to be the world’s largest AI investment: a $500 billion infrastructure project to “ensure American leadership in AI.” Not one to shy away from hyperbole, its key backer, OpenAI, creator of ChatGPT, promised “massive economic benefits for the entire world” with facilities to help people “use AI to uplift humanity.”
Now, OpenAI appears to be exiting one part of the deal: the expansion of a landmark data center stretching across a swath of land in Abilene, Texas, that has become one of the most visible manifestations of an investment frenzy in the chips and power plants needed to build and operate AI. There has been a breakdown in negotiations over funding for the project, as well as a timeline for when the expanded capacity could come online.
This may be fine for OpenAI; presumably you can find other data centers. Not so good for OpenAI’s partner on the project, Oracle, which has already spent billions on hardware for the site. It’s one of many cracks appearing in the capital side of the AI economy that are making investors quite nervous.
Both companies have said the development will not derail their AI plans. They also said that a month ago, when a different $100 billion deal fell through between OpenAI and Nvidia, the world’s largest maker of chips that train AI models and answer the billions of questions people ask them every day.
The fate of such agreements for the global economy only increases in importance. Future data center leases agreed upon by the largest cloud computing companies (including Amazon, Oracle and Microsoft) have increased nearly 340% in two years and now exceed $700 billion, according to Bloomberg. It’s a lot of money if technology doesn’t begin to deliver on its promise of boosting economic productivity. On Friday, more than three years since the launch of ChatGPT sparked the AI hype, the United Kingdom reported zero GDP growth in January.
On Monday, The Guardian exposed another crack in the AI edifice. An investigation has found that the UK’s major AI deals, many of them announced with great fanfare during Donald Trump’s state visit last September, are not as described in government and corporate press releases. Key projects are delayed or unlikely; The crucial “investments” are in fact vague deals between mostly American tech companies that ministers are desperately using as an engine of economic growth.
If the cracks in this data center boom widen, the consequences range from Britain ending up without the AI infrastructure it needs to keep up with the global economy to the more serious risk of the entire AI bubble bursting in a repeat of the 2001 dotcom crash that could bring down the global economy.
“There has been a lot of blind optimism around building AI infrastructure,” said Andy Lawrence, executive director of research at the Uptime Institute, which inspects and rates data centers. “While there is an incredible boom underway, with construction on a scale never seen before, it has also been evident for quite some time that many projects would not go ahead or would take much longer to build and begin operating than many of the claims suggested. Because of the high risks and rewards of AI, it has attracted speculators who promise investments but have little experience in the sector.”
Most emblematic, The Guardian investigation highlighted a site in Loughton, Essex, which the government said would host “the UK’s largest sovereign AI data centre” by the end of 2026. The then technology secretary, Peter Kyle, called it “a new beginning for our economy and for workers”. A year later it was still being used as a scaffolding yard with almost no chance of being open when billed. Following The Guardian’s investigation, Nscale confirmed that he had purchased the land on which the computer would be built, eight months after saying he had done so in January 2025. He does not yet have a building permit, but said on Friday that he planned to start construction before July and would power on the data center between April and July 2027.
The rickety AI deals came amid a tightening between American tech corporations and senior politicians in the United States and the United Kingdom. Donald Trump’s top AI advisors include David Sacks and Sriram Krishnan, both with recent records as technology investors. In London, OpenAI hired former chancellor George Osborne; Anthropic and Microsoft employed former Prime Minister Rishi Sunak; Peter Mandelson owned a consulting firm that lobbied for Palantir; and the Tony Blair Institute has received funding from Oracle’s billionaire owner Larry Ellison’s foundation.
These figures have helped create an AI policy in which the UK has essentially agreed to be a staging ground for US-designed hardware that is mainly rented to US tech companies. The UK government says it is creating a “sovereign AI infrastructure”, which has a controversial definition ranging from UK-owned hardware and data to retain control of a piece of critical national infrastructure in a world of unstable international alliances, to AI minister Kanishka Narayan’s more flexible definition as “strategic leverage” so the UK “can ensure continued access to critical inputs”.
In the United Kingdom that means dependence on the United States. As Nvidia CEO Jensen Huang said during Trump’s state visit last September: “The United States must lead all AI technology.”
Former deputy prime minister Nick Clegg put it more clearly that week, calling the UK a “technologically vassal state”. This week, Clegg became a board director at Nscale, the British company involved in the AI deal with Loughton, whose client is Microsoft, part of the American tech hegemony whose power he lamented six months ago.
On BBC Radio 4’s Today program this week, Nscale senior vice president Imran Shafi was asked whether its Essex data center would be operational by “Q4 2026” as promised. He responded: “The time you will be available will be the time we have approved with our client.”
Narayan, meanwhile, defended the broader pace of progress. “What we are saying is that we are making concerted progress,” the minister told CityAM. “We already have active data centers in Lanarkshire. We have shovels in the ground in parts of the North East.”
Narayan might consider the example of the current crisis in Texas. Billions were pledged, construction began, billions of dollars worth of equipment was purchased, and then OpenAI folded, leaving its partners in the unenviable position of having to find another giant AI company to work with.
OpenAI reportedly wanted a newer chip model: and when construction in Texas is finished, the hardware Oracle purchased may no longer be cutting-edge. It was like buying a bunch of iPhones right before a much more powerful model was released. The pace at which chips are becoming obsolete casts a further shadow over the UK government’s claims of massive investment in AI. He describes in terms of cash “investments” which are mostly computer chips. Chips aren’t money: They depreciate, possibly even faster than most tech companies say they will.
This means it matters when a data center in Essex or an AI center in Lanarkshire needs to come online. By the time they are ready and the additional electricity has been sourced, will advances in the design of AI systems mean that using 2025 chips is like owning a propeller plane in the jet age? Or if the announced offers relate to future chips, will they be available? Iranian drone attacks have already affected Qatar’s helium supplies, which chipmakers need. What happens if China cuts off supplies from Taiwan?
“Data centers, especially large, high-density AI ones, are very complex engineering projects,” said Lawrence of the Uptime Institute. “Few are launched in less than two years and usually take much longer. It is not uncommon for some projects to be delayed for years or postponed indefinitely.”
The last component here is the banks. Nscale’s chips and those of other data center companies are taken advantage of. These operators have taken out billions of dollars in loans thanks to their graphics processing units (GPUs). At least in the case of Nscale, this debt will go towards funding its expansion into the UK, but when does that debt come due? If it can’t be paid, what happens to Nscale or the financial institutions left looking for a buyer for potentially obsolete chips?
An Nscale spokesperson said it “works with established financial counterparties and maintains disciplined governance around financial decisions. We take a conservative approach to our financing, aligned with long-term infrastructure construction.”
Alvin Nguyen, an analyst at Forrester, said: “The people who lend the money, the financial institutions, are taking on a lot more risk because the chips have a shelf life.”
The boom in data center investment represents one of the biggest infrastructure bets of this or any era. Whether that scaffolding yard in Loughton ends up becoming a real artificial intelligence factory could tell us a lot about who will win and who will lose.
