OpenAI’s revenue is reportedly $20 billion less than previously projected

We have reached the part of the hype cycle where the spreadsheets finally meet reality. For months, the industry whispered about OpenAI’s astronomical growth, with some reports suggesting an annualized revenue run rate hitting $70 billion. It was a number so large it felt like a gravity well, pulling every venture dollar in Silicon Valley toward Sam Altman’s desk. But new data suggests those projections were off by roughly $20 billion. That isn't a rounding error; it is the size of a Fortune 500 company disappearing from the ledger.

The Valuation Gap

As a founder, you learn early on that there are two types of numbers: the ones you tell investors to get a check, and the ones you see in your bank account at 2:00 AM. OpenAI has been operating in a stratosphere where the usual rules of fiscal gravity don't seem to apply. When your valuation is tied to the promise of AGI—a concept that is still more philosophical than technical—your revenue projections become a Rorschach test for optimism.

A $20 billion discrepancy suggests that the enterprise adoption of LLMs isn't happening at the breakneck speed the marketing decks promised. While every C-suite executive is talking about AI, the actual implementation of paid, high-scale API usage is lagging. Companies are experimenting, but they aren't necessarily shipping at the volume required to sustain a $70 billion run rate. They are worried about data privacy, hallucination rates, and the simple fact that running these models is incredibly expensive.

The Cost of Being First

OpenAI’s burn rate is the stuff of legends. Building frontier models requires a level of capital intensity that few companies in history have ever faced. We are talking about billions spent on H100s, electricity, and the rare talent capable of tuning these massive neural networks. When you miss your revenue target by this much, the burn becomes a bonfire.

For builders, this is a cautionary tale about the "subsidized user" trap. For a long time, we’ve been building on top of models that are effectively priced below their cost of production. OpenAI and its competitors have been willing to take losses to gain market share and collect data. But if the revenue isn't catching up to the infrastructure spend, the subsidies will eventually end. We are already seeing this with the introduction of tiered pricing and more aggressive token management.

Why the Numbers Swell

Why was the $70 billion figure even circulating? In the crypto world, we call this wash trading; in the AI world, it’s often just creative accounting regarding credits and partnership deals. When you have massive investments from companies like Microsoft, the line between "investment" and "revenue" can get blurry. If a partner gives you $10 billion in cloud credits and you use those credits to provide services back to them, is that organic growth? The market is starting to realize that much of the reported AI revenue might just be money moving in a very expensive circle.

What This Means for Founders

If you are building an AI startup right now, you need to stop looking at OpenAI as a blueprint for business health. They are a research lab masquerading as a platform, funded by the largest entities on earth. You don't have that luxury. A $20 billion revenue miss at the top of the food chain means that venture capital is going to start asking harder questions about unit economics at the bottom.

• Focus on efficiency: If the giants can't make the math work on massive models, you should be looking at smaller, fine-tuned models that do one thing well for a fraction of the cost.
• Own your distribution: Relying solely on the OpenAI API is becoming a risky bet. If they need to close that $20 billion gap, your API costs are the first place they will look for margin.
• Solve real problems: The era of "GPT-4 but for [X]" is dying. Revenue misses happen when the product is a luxury, not a necessity. You need to build something that saves a company more money than the tokens cost to generate.

The Reality Check

I’ve seen this movie before in the early days of the blockchain boom. We saw massive "treasury" valuations that didn't translate to liquid cash flow. OpenAI is facing a similar reckoning. They are still the leader, and they still have the best tech in the room, but the myth of infinite, frictionless growth is showing cracks. The industry needs to move past the hype of "intelligence too cheap to meter" and start building sustainable businesses.

The gap between projection and reality is where most startups die. For OpenAI, it’s a pivot point; for the rest of us, it’s a warning.

We should expect a cooling period. This isn't necessarily a bad thing. When the hype dies down, the builders who are actually solving problems—rather than just burning VC cash to generate tokens—will be the ones left standing. The $20 billion miss is a signal that the market is finally starting to value actual utility over theoretical potential.

Moving Forward

Don't be distracted by the massive numbers. Whether OpenAI makes $50 billion or $70 billion doesn't change the fact that the underlying technology is transformative. However, it should change how you structure your own company’s finances. If the king of the hill is struggling to hit its marks, you better make sure your own house is in order. Stop chasing the scale of a research lab and start chasing the profitability of a software company.

Read the original at TechCrunch AI →