Oracle’s $300 Billion OpenAI Bet Is Valued at Minus $74 Billion
Oracle’s planned $300 billion computing agreement with OpenAI is now being assessed at an implied value of minus $74 billion, highlighting the financial risks of debt-funded AI infrastructure expansion. The estimate reflects the gap between the contract’s headline revenue potential and the capital, financing and execution burden Oracle may carry to deliver it.\n\nOracle and OpenAI agreed to develop up to 4.5 gigawatts of additional US data-center capacity for OpenAI’s Stargate infrastructure program. OpenAI said the partnership, announced in July 2025, would bring Stargate capacity under development to more than 5 gigawatts and support more than 2 million chips. A later OpenAI announcement described the arrangement as a partnership exceeding $300 billion over five years (OpenAI’s July announcement; OpenAI’s September update).\n\nThe scale of the commitment made Oracle a major public-market proxy for the expansion of OpenAI and the wider AI infrastructure cycle. It also shifted attention from signed demand to the economics of supplying that demand: data-center construction, GPUs, power, leases and debt must be funded before the associated revenue is fully realized. The Financial Times’ analysis (the reported valuation assessment) therefore treats the contract’s value as an economic calculation rather than as a claim that Oracle has literally paid $74 billion for an asset worth nothing.\n\nOracle’s own filings show why investors are scrutinizing the financing. For fiscal 2026, the company reported $55.7 billion in capital expenditures and negative free cash flow of $23.7 billion while expanding its cloud-infrastructure business (Oracle’s fiscal 2026 results). The same release reported $638 billion in remaining performance obligations, with $75 billion of large AI-contract commitments backed by customer prepayments or customer-supplied hardware. Those arrangements can reduce Oracle’s funding requirement, but they do not eliminate construction, concentration or delivery risk.\n\nThe reception has consequently been divided. Supporters view the OpenAI relationship and Oracle’s growing backlog as evidence that the company has secured a central role in AI computing. Critics point to the dependence on one exceptionally large customer, the need to raise capital during a period of heavy spending, and the possibility that AI demand, pricing or deployment schedules may not develop as planned. The central limitation of the headline figure is that a multiyear contract is not the same as near-term cash generation.\n\nThe broader Stargate plan underscores the ambition: OpenAI originally announced a goal of investing $500 billion in 10 gigawatts of US AI infrastructure over four years, and said the Oracle partnership put it on track to exceed that target. The next test will be execution—bringing power and data centers online, converting contracted capacity into billable usage, and demonstrating returns that justify the financing burden.\n\nOracle’s case is therefore an early stress test for the AI infrastructure boom. The technology opportunity remains substantial, but the financial outcome depends less on the size of the announced contract than on who ultimately funds construction, who bears delays and whether end-user demand supports the promised scale.