Why OpenAI’s $1 trillion bet could remake — or break — the AI industry

OpenAI

OpenAI has launched one of the boldest spending sprees in the history of the tech sector, announcing deals and commitments that could total $1 trillion in new infrastructure. The company’s leadership, under CEO Sam Altman, has cast these moves as a demonstration of self-confidence, intended to reassure partners, rivals, and markets that it remains the leading force in artificial intelligence. At a moment when rivals are catching up and OpenAI’s corporate structure is in flux, the strategy is as much about signalling dominance as it is about future revenue, the Financial Times reported.

The Stargate project and mega-partners

At the centre of OpenAI’s plan is the Stargate project, a collaboration with Oracle and SoftBank to build massive new data centres. These facilities, five of which are scheduled to come online within the next three years, could cost as much as $400 billion. Oracle, already heavily indebted, is set to take on a central role in construction. Meanwhile, Nvidia has pledged up to $100 billion in equity support tied to data centre spending that could reach $500 billion. Together, these projects represent a giant bet that AI demand will not just grow, but explode.

The financing puzzle

The financing behind the expansion remains murky. Nvidia’s equity is structured in tranches, with only the first $10 billion arriving once the initial data centre is live, likely in 2026. OpenAI also faces a commitment to pay Oracle roughly $300 billion for cloud services by 2030. Much of this debt-fuelled expansion depends on sustained revenue that the company has not yet demonstrated. For now, OpenAI’s business model relies heavily on subscription products such as ChatGPT and licensing fees, which generated about $13 billion in revenue last year — far short of what would be needed to justify trillion-dollar outlays.

Echoes of a bubble

The aggressive scale of the investment has fuelled comparisons to past technology bubbles. Analysts recall the dot-com era, when telecom firms blanketed the US with fibre optic cables that sat unused for years. Critics worry that OpenAI may be overbuilding capacity, risking an oversupply of infrastructure that customers may not fully use. Each new AI model costs significantly more to train than its predecessor, while improvements appear to be incremental rather than transformative. The risk is that hardware depreciates faster than OpenAI’s revenue can grow, leaving partners burdened with debt and shareholders facing disappointment.

Market reaction

Investors have responded with mixed signals. Oracle’s shares surged on the announcement of its role in Stargate, buoyed by hopes that it would anchor the next wave of AI growth. Nvidia initially saw gains from the confirmation it would remain a key chip supplier, but those quickly faded as analysts questioned whether the projected $350 to $400 billion in new chip sales would materialize. Markets appear uncertain whether to treat OpenAI’s strategy as visionary or reckless, with valuations swinging between enthusiasm and doubt.

The road ahead

OpenAI insists it will lay out a detailed financing plan later this year, while partners like Oracle are expected to reveal their own strategies to reassure investors. The company’s wager is that by the late 2020s, demand for AI capacity will outstrip supply, allowing it to capture revenue streams large enough to cover debt and justify spending. For now, its message is clear: it is racing to outspend rivals and define the terms of the AI future. But the questions remain the same as those posed during every boom — how much of this capacity will actually be used, who bears the risks, and whether the returns will justify the gamble.