OpenAI Expects $278 Billion in Losses Through 2030

A Staggering Number

A report from the Financial Times, relayed by Thurrott.com, indicates that OpenAI anticipates losing at least $278 billion through the end of 2030 1. That figure is remarkable even by the standards of the AI industry, where companies have consistently prioritized growth over profitability. If accurate, it would rank among the largest cumulative losses ever projected for a single private company, and it underscores just how expensive the frontier-AI business has become.

The number deserves some framing. These are expectations for cumulative losses over roughly a five-year horizon, not a single year's shortfall. OpenAI's costs are dominated by compute — the enormous GPU clusters required to train and run ever-larger models — along with talent, data licensing, and now the infrastructure buildouts the company is pursuing with partners. The company is betting that all of this spending will be repaid later, either through dominant market share, enterprise contracts, or eventually reaching a durable revenue engine.

Why the Losses Keep Growing

The pattern here is familiar. Every time OpenAI's models achieve a capability jump, rivals match or approach it within months, which forces more spending on the next generation. Meanwhile, inference costs — running models for hundreds of millions of users — scale with usage. A company can improve margins per query over time, but usage grows faster than efficiency improves. The FT's reporting suggests OpenAI itself sees these dynamics persisting through 2030, meaning the losses are structural rather than a temporary growth-phase phenomenon 1.

The counterweight is that revenue is also rising sharply. If OpenAI's top line keeps compounding, even massive losses in absolute terms could represent a shrinking percentage of revenue. The question investors are implicitly answering with their wallets is whether that revenue curve eventually crosses above the cost curve — and when.

The $1.2 Trillion Valuation Signal

This is where the second part of the story becomes essential. Despite the projected losses, Macquarie analysts view OpenAI's latest funding round as evidence of sustained momentum in the AI sector, and reporting suggests the company may seek new capital at a valuation of roughly $1.2 trillion — a figure that would top Anthropic and cement OpenAI's position as the most valuable private company in the world 2.

On its face, this is a striking juxtaposition: a company expected to burn hundreds of billions is simultaneously commanding a valuation above many of the world's largest public corporations. But it is not necessarily contradictory. Investors are not underwriting OpenAI's current financials — they are underwriting a claim that AI will become foundational infrastructure for the global economy, and that the company holding the most capable models and largest distribution network will capture an outsized share of that value.

Where the Sources Converge and Diverge

The two pieces of reporting overlap on the central fact that capital continues to flow into OpenAI at extraordinary scale, but they emphasize different implications. The loss projection, as reported by the FT, highlights the enormous cost side of the ledger and the open question of when — or whether — the model becomes self-sustaining 1. Macquarie's analysis, by contrast, reads the funding environment optimistically, treating investor appetite as confirmation that the AI investment cycle still has strong momentum 2.

Both can be true at once. Deep-pocketed investors may be willing to absorb years of losses in exchange for a chance at the eventual winner-take-most economics of AI, while simultaneously the absolute size of the burn raises real fragility questions. OpenAI's strategy depends on continued access to enormous capital, favorable terms from cloud and chip partners, and a macro environment that keeps funding flowing.

The Bottom Line

My reading is that the $278 billion figure should be taken seriously as a statement of intent rather than a mere warning. OpenAI is telling investors that the arms-race phase of AI is not ending, and that the company plans to outspend everyone through 2030. The trillion-dollar valuation suggests investors believe the spend will be worth it — for now. The real risk is not that the losses exist, but that the strategy leaves OpenAI structurally dependent on a small circle of backers, several of whom are also its competitors and suppliers. If capital enthusiasm cools even modestly, the gap between projected losses and projected revenue becomes the most important number in tech.