Nvidia Ditches $100bn OpenAI Deal for $30bn Bet — What Changed?

Nvidia is close to finalising a $30 billion equity investment in OpenAI, replacing the sprawling $100 billion multiyear partnership the two companies agreed last September, in what amounts to a significant recalibration of the relationship at the centre of the AI boom.

The deal, which could be concluded as early as this weekend according to people with knowledge of the negotiations, forms part of a larger funding round that is on track to raise more than $100 billion and will value the ChatGPT maker at $730 billion before the new capital is included. OpenAI will reinvest much of the proceeds into Nvidia hardware — preserving the commercial relationship between the two companies — but the structured, decade-long commitment announced with great fanfare six months ago will not proceed.

The retreat matters. Last year’s agreement, framed as a letter of intent, would have seen Nvidia invest ten increments of $10 billion as OpenAI’s computing requirements grew, in return for a significant stake in the AI start-up. It was designed to lock the two companies together at the apex of the AI supply chain — Nvidia as the dominant supplier of the chips that power large language models, OpenAI as the dominant builder and deployer of those models. The announcement helped propel Nvidia above $5 trillion in market value and accelerated a period of frenzied dealmaking for Sam Altman’s company, which subsequently struck complex partnerships with AMD, Broadcom, Oracle and other major players in the AI infrastructure stack.

Now, that architecture is being quietly dismantled — not because the commercial logic has changed, but because the market environment has.

Why the Deal Changed

US technology stocks have fallen 17 per cent since the start of 2026. The sell-off has been concentrated in the AI-adjacent names that led the market higher over the previous two years, driven by a combination of factors: earnings that failed to match elevated expectations, growing scepticism about the near-term revenue potential of generative AI, and mounting concern about the circular structure of AI sector dealmaking.

That circularity was always the vulnerability in the original Nvidia–OpenAI arrangement. Nvidia would invest in OpenAI. OpenAI would spend the money on Nvidia chips. Nvidia’s revenue growth would justify its valuation. Its valuation would justify the investment in OpenAI. Analysts flagged the feedback loop at the time, warning that it resembled the kind of self-reinforcing capital structures that have historically preceded market corrections.

A $30 billion lump-sum investment is a cleaner, more conventional transaction. It gives Nvidia a significant equity position in the most valuable private AI company on earth without the open-ended financial commitment of a decade-long partnership that was, in effect, a bet on perpetually accelerating demand for AI compute. For Nvidia, it reduces exposure to a scenario in which OpenAI’s growth slows, compute requirements plateau, or alternative chip architectures erode its dominance. For OpenAI, it secures a massive injection of capital from its most important supplier while freeing it from a structure that bound it tightly to a single hardware provider at a time when diversification — through deals with AMD, Broadcom and custom chip programmes — is becoming strategically important.

The Valuation Question

The $730 billion pre-money valuation is extraordinary by any measure. It would make OpenAI the most valuable private company in history — more valuable than most publicly listed European corporations and roughly equivalent to the market capitalisation of companies like LVMH or Samsung. The broader funding round, raising over $100 billion, dwarfs anything previously seen in private markets and reflects the unique position OpenAI occupies: too large and too strategically important to operate as a conventional start-up, but still privately held and not yet subject to the scrutiny of public market investors.

The question is whether the valuation is sustainable. OpenAI’s revenues have grown rapidly — the company reportedly exceeded $10 billion in annualised revenue last year — but its costs are growing faster. Training and running frontier models requires enormous and expanding compute infrastructure, and OpenAI has not demonstrated a clear path to profitability at scale. The bulk of its revenue comes from subscriptions and API access, but competition from open-source models, including those built by Mistral, Meta and a growing number of Chinese developers, is intensifying pressure on pricing.

Nvidia’s willingness to put $30 billion behind OpenAI at this valuation is a powerful signal of confidence in the company’s long-term position. But it is a smaller signal than $100 billion would have been — and the restructuring itself suggests that even the most committed participants in the AI boom are beginning to recalibrate their assumptions about how fast, how far and how profitably this technology cycle will unfold.

What It Means for the AI Ecosystem

The revised deal has implications beyond the two companies directly involved. The original $100 billion partnership was part of a broader web of interconnected agreements that defined the AI sector’s capital structure in 2025. OpenAI’s deals with Oracle for data centre capacity, with AMD and Broadcom for alternative chip supply, and with Microsoft for cloud infrastructure and distribution created a tightly coupled ecosystem in which capital flowed in loops — from investors to AI companies to infrastructure providers and back again.

The unwinding of the Nvidia commitment does not break that ecosystem, but it loosens it. It suggests that the era of maximalist, long-term AI partnerships may be giving way to something more transactional and more cautious. That is consistent with the broader correction in US tech markets and with the rotation into European and other non-US equities that has accelerated in recent weeks.

For investors, the message is nuanced. The AI sector is not collapsing — a $100 billion funding round at a $730 billion valuation is not what collapse looks like. But the terms on which capital is being deployed are shifting. The unbounded optimism that characterised 2024 and early 2025, when every AI deal was bigger than the last and every commitment was longer-term than the one before, is being replaced by something more disciplined.

Nvidia remains the dominant force in AI hardware. OpenAI remains the dominant force in frontier AI models. The commercial relationship between them — OpenAI buying Nvidia chips, Nvidia profiting from OpenAI’s growth — is intact. What has changed is the financial structure around that relationship. The $100 billion commitment was a bet on an AI future that would grow without interruption. The $30 billion investment is a bet on an AI future that looks more uncertain than it did six months ago — but still worth billions.

That distinction may prove to be the most important signal the AI sector has sent all year.