Altman faces $1.4 trillion in skepticism: As long as computing power remains scarce, OpenAI must keep burning money.

OpenAI CEO Sam Altman emphasized that as long as computing power remains scarce, OpenAI must continue to burn money.

On December 19th, OpenAI CEO Sam Altman explained in a Big Technology Podcast interview that the company’s current losses are due to aggressively expanding model training scale. As revenue increases and inference accounts for a larger portion of the computing cluster, it will eventually cover the training costs.

Faced with the huge gap between $1.4 trillion in spending commitments and $20 billion in revenue, Altman admitted that the company's training cost growth still outpaces revenue growth. However, he stressed that OpenAI has always been in a “compute deficit”—which precisely proves that demand is strong.

Altman stated, Only when the company has a large amount of idle computing power that cannot be monetized will external concerns be reasonable. Currently, the severe shortage of computing power greatly limits the company's revenue growth potential, which is the core reason for continued large-scale investment.

Altman’s path to profitability depends on a simple bet: OpenAI can find buyers as quickly as it can develop. Ultimately, this bet will either keep winning, or exhaust all resources.

Training Costs Drag Down Current Profits

In the interview, when asked when the company might break even, Altman did not shy away from the still-expanding expenditures.

Altman explained that as revenue grows, inference business will account for an increasingly larger share of the fleet, and will eventually swallow up training expenses. This is OpenAI’s plan: to spend huge sums on training, but earn more returns through inference.

Altman bluntly stated:

If we didn’t continue to greatly increase training costs, we would have already been profitable. But what we are doing now is heavily investing in training these large models.

The current financial situation is based on a clear bet: that it is necessary to continue investing heavily in model training to maintain a technological edge and expand the market.

Data shows that before OpenAI is projected to become profitable in 2028 or 2029, it may face a loss of about $120 billion. Regarding this, Altman affirmed the company’s strategic focus: using revenue growth to support compute expansion, not shrinking back due to short-term losses.

Compute Is the Lifeline

When questioned about the huge gap between $1.4 trillion in spending and $20 billion in revenue, Altman’s initial response seemed a bit confused.

Altman spoke about how humans find it hard to intuitively grasp exponential growth, saying "Evolution has made us good at a lot of mental math, but building quick mental models for exponential growth doesn't seem to be one of them."

Altman described computing power as the “lifeline” that makes everything possible. He stated that the company believes it can maintain a steep revenue growth curve, and all signs at present point to the fact that, without enough computing power, this goal cannot be achieved.

He emphasized that OpenAI is "extremely limited by compute," which directly impacts its revenue performance.

Altman admitted "we’ve always been in a compute deficit; it’s always limited what we can do," and expects this situation to persist, though he hopes it may ease over time.

Altman revealed that the company has planned for this from multiple angles, and expects that as technology advances, compute efficiency per dollar will improve.

Risk Disclosure and DisclaimerThe market carries risk; investments require caution. This article does not constitute personal investment advice, nor does it consider the specific investment goals, financial situations, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article fit their particular circumstances. Investing on this basis is at your own risk.