Two-Year-Old Chatbots Now Worth More Than Burger Empires

Anthropic and OpenAI have reportedly crossed revenue thresholds that place them ahead of some of the world's most durable consumer franchises—McDonald's and Starbucks among them. Two companies that barely existed a quarter-decade ago are now generating more top-line sales than institutions that have spent decades optimizing supply chains, real estate, and unit economics. The premise sounds triumphant. It is not. It is, instead, a masterclass in mistaking hyperinflation for health.

Let us be precise about what these companies actually do: they operate large language models and sell API access and subscription services to enterprises and consumers willing to pay for probabilistic text generation. OpenAI offers ChatGPT Plus subscriptions and enterprise licensing. Anthropic does much the same with Claude. The revenue figures being cited reflect genuine bookings, yes—but they reflect a market still in the grip of what one might charitably call 'infrastructure spending euphoria.' Trillions in forecasted AI capex spending have created a gravitational pull toward any company remotely credible enough to monetize the trend. McDonald's, by contrast, sells hamburgers. Actual hamburgers. To actual humans. Who actually eat them. The comparison is not apples to oranges; it is apples to stock certificates.

This is not unprecedented behavior in venture capital. The pattern is familiar: a new technology arrives, the venture ecosystem becomes convinced it will remake civilization, and companies in that space experience valuation multiples that would make a 1999 dot-com founder blush. Pets.com had runway too. So did Theranos, until it didn't. Anthropic and OpenAI are obviously superior to both in terms of actual product utility, but superiority in execution does not immunize a company from the gravitational pull of speculative excess.

The investment thesis, if you squint, is defensible: if AI infrastructure spending truly reaches the trillions—if enterprises genuinely adopt these tools at scale—then early-mover advantage matters enormously. But notice the conditional language: 'if,' 'truly,' 'genuinely.' These are wagers, not facts. And they are being priced as certainties. The press releases will call this 'accelerating revenue growth' and 'market validation.' What they mean is: 'people are buying this at any price because they believe it will be indispensable tomorrow.'

What could go wrong? Commoditization, for one. If large language models become sufficiently standardized and widely available, margin compression follows as night follows day. Competition from better-capitalized incumbents—Microsoft, Google, Meta—poses a non-trivial threat. And there is always the possibility that the market's appetite for AI-powered services peaks at a lower ceiling than current forecasts suggest. Or that regulatory friction cuts into the addressable market. Or that the computational economics simply do not work at scale. Any of these scenarios would render current valuations absurd.

What this moment reveals about American venture capital is neither surprising nor reassuring: we have learned nothing from the cycle of 2000, 2008, or 2020. We are simply better at generating narrative scaffolding around the latest technology. Starbucks took twenty years to prove its model. McDonald's spent decades. Anthropic and OpenAI have two years. That they have generated impressive revenue in that window is real. That the markets are pricing them as if they have already solved for unit economics, retention, and defensibility is a very different proposition entirely.

Come back when the hamburgers are actually cheaper because of AI. Until then, this is just a very expensive race to see which chatbot can burn capital the fastest while calling it growth.