OpenAI on the Brink: Why Microsoft's Billions May Not Be Enough

Imagine building the most powerful engine in human history, only to find that every kilometer it travels burns more gold than the useful work it delivers. That, roughly, is what OpenAI's financial model looks like today. While the whole world discusses ChatGPT's new capabilities, people who are used to counting money are starting to glance nervously at the exit. Asset managers who have spent decades watching Silicon Valley's booms and busts are sounding the alarm: OpenAI is showing every sign of a classic corporate collapse. This is not mere skepticism but sober arithmetic based on the company's spending on compute and staff.

The problem is not that the technology is bad. The problem is the math of survival. To train its next model, OpenAI needs tens of thousands of NVIDIA chips, each costing as much as a decent car. Add electricity bills that could rival the budget of a small country, and engineer salaries running to hundreds of thousands of dollars. Sam Altman is playing a very dangerous game in which the stake is endless growth and the only way not to fall is to keep finding new investors. Valuation is growing, but losses are growing even faster, creating a dangerous scissors effect that cannot be ignored forever.

Until now, Microsoft has remained the main lifeline. But even the software giant's patience is not infinite. Investors are starting to ask uncomfortable questions: when will these investments begin to bring real profit rather than just pretty demos on social media? History offers plenty of examples of companies with supposedly limitless potential that turned to dust once the flow of cheap money dried up. Think of WeWork or the early years of Uber: the scale was enormous, but the economics did not add up for years, which ultimately led to a painful reassessment of values.

The gap between media hype and financial reality is becoming critical. If OpenAI cannot prove that its models can generate net profit, not just losses, a hard landing awaits. It would hit the entire AI market and set off a chain reaction. After all, if even the industry leader cannot get into the black, what about the hundreds of small startups trying to copy its path? We are watching a classic situation in which technological progress runs well ahead of the creation of a sustainable business model able to pay for that progress.

The situation is complicated by the fact that competitors are not asleep and have far deeper pockets. Google and Meta* can afford to burn money on research for years, backed by their advertising empires. OpenAI has no such backstop. Every model update is an all-in bet. Either the next iteration becomes so useful to businesses that they start paying many times more for it, or we will witness the most expensive bankruptcy in tech history. Market expectations are overheated, and any hiccup in shipping new products could trigger a mass exodus of investors who are now held in place only by promises of AGI arriving soon.

Bottom line: OpenAI has turned into a financial kamikaze whose success depends solely on investors' faith in a miracle. Will that faith last until profits arrive, or will the bubble burst first?

*Meta has been designated an extremist organization and is banned in Russia.