Economics of AI: why market leaders aren't even trying to make money

If you thought Uber or WeWork in their worst years were masters of burning money, the artificial intelligence industry is asking you to hold its bottomless wallet. Today, the largest market players—from OpenAI to Anthropic—don't even pretend they're aiming for operating profit in any foreseeable future. We're witnessing a unique moment in the history of technology, when companies valued at hundreds of billions of dollars openly acknowledge: our spending on model training will always outpace our subscription revenue. This is not merely temporary hardship, but a fundamental feature of the current arms race, where the winner won't be the one who sells better, but the one who can ignore an emptying bank account the longest.

Context matters more than any numbers. To understand the scale of what's happening, we need to recall how Silicon Valley has worked over the last ten years. Typically, a startup burns money to capture the market, then begins monetizing its loyal audience.

But with AI, that pattern has broken down. The cost of training each subsequent version of GPT or Claude grows in geometric progression. If training GPT-4 cost around a hundred million dollars, the next iterations already require billions.

The lion's share of these funds doesn't go to programmer salaries, but into Nvidia's pocket for chips and energy companies for electricity. We've found ourselves in a situation where the product becomes more expensive to produce faster than the market can get used to its current price.

Why do investors keep writing checks for billions of dollars while looking at these financial ruins? The answer lies in blind faith in the concept of AGI—Artificial General Intelligence. In the corridors of OpenAI and Google DeepMind, there's a conviction that once they create a system capable of replacing humans in most cognitive tasks, the question of money will resolve itself.

Such a system supposedly will figure out how to make trillions, optimize the economy, and close all debts. This is a kind of technological messianism: we're building a digital god, and gods don't need quarterly reports. But the problem is that we still need to survive until that moment, and the cost of entry into this club of the chosen continues to rise every day.

Meanwhile, the business models being offered now look more like an attempt to ease shareholders' consciences than an actual plan. Twenty-dollar-a-month subscriptions don't even cover server depreciation for the machines running these models. Corporate implementations are moving slowly due to security concerns and AI hallucinations. As a result, we have a paradoxical situation: a technology that's supposed to automate and cheapen everything is itself the most expensive and inefficient business in the world. Companies are forced to constantly fundraise just to keep the lights on in server rooms, creating a dependence on venture capital that resembles a financial pyramid built on very smart algorithms.

The connection to previous technology bubbles is obvious, but there's one significant difference. During the dotcom era, companies spent money on marketing and capturing attention. In the AI age, money is spent on physical infrastructure and raw computing power. This makes any collapse, if it happens, much more painful for the entire economy. If investors tomorrow decide the path to AGI is too long, we won't just be left with bankrupt websites, but mountains of expensive hardware and giant data centers that consume the energy of entire cities. The irony is that market leaders don't even try to hide this situation, openly declaring the need for hundreds of billions of investment to continue operations.

What does this mean for us? We're using incredibly expensive tools practically for free, while venture funds and Microsoft foot the bill for this feast. This is a golden age for users, but a warning sign for the industry. The long-term sustainability of AI companies now depends not on code quality, but on geopolitics, lithium supplies, and the patience of the planet's largest investors. As long as they believe in the miracle, the party continues. But the moment someone first asks 'where's the money?', the rules of the game will change instantly, and many will have to remember what traditional economics looks like with its boring concepts of profit and loss.

The point: today's AI industry is not a business in the classical sense, but a grand scientific experiment funded by other people's money. Will models become smart enough to pay for their own existence before investors run out of patience?