Three days ago, OpenAI announced a $110 billion funding round at a $730 billion valuation. The headlines celebrated it as a triumph. But buried in the financial analysis is a number that should terrify anyone paying attention: OpenAI is projected to lose $14 billion in 2026. Not revenue — losses. The kind of losses that, according to multiple financial analysts, put the company on a path to insolvency by mid-2027.
OpenAI generates roughly $13 billion annually from ChatGPT subscriptions and API access. That sounds massive until you look at the expense side: compute costs alone run $1.4 billion, and that's before accounting for the infrastructure expansion, the research teams, the training runs for next-generation models, and the new AWS partnership that commits them to consuming 2GW of Trainium compute. Revenue is growing. Costs are growing faster.
Sam Altman has publicly claimed OpenAI's revenue will reach $100 billion by 2027. Financial analysts are skeptical. Sebastian Mallaby, an economist at the Council on Foreign Relations, has said that even if OpenAI changes strategy and leverages its overvalued shares, it won't easily escape the financial pressure. A separate projection suggests cumulative losses could reach $40 billion by 2028.
This is the core tension of the AI bubble thesis: the most valuable private company in AI cannot generate enough revenue to cover its costs. Not because the product is bad — ChatGPT is genuinely useful — but because the cost structure of building and running frontier AI models is fundamentally incompatible with current pricing. OpenAI is charging $20/month for a product that costs significantly more than that to serve at scale.
The recent $110 billion raise doesn't solve this problem. It delays the reckoning. Every dollar from Amazon, Nvidia, and SoftBank buys OpenAI more runway, but runway without a path to profitability is just a longer fall. The company's previous round of $40 billion lasted exactly one year. At a $14 billion annual burn rate, $110 billion buys roughly 7-8 years — if costs don't increase. They will.
If the most well-funded AI company in history can't make the economics work, what does that say about the unicorn startups trying to build on top of it? Every company relying on OpenAI's API is building on a foundation that is, by the numbers, financially unstable. Every company competing with OpenAI needs to match a burn rate that would bankrupt most nations.
The survivors of this cycle won't be the companies that raised the most money. They'll be the ones that never needed it.