The AI Giants Are Spending $50 Billion Each. Can They Ever Make Money?
The AI leaders are spending like almost no companies ever have. OpenAI and Anthropic are each on track to spend roughly 50 billion dollars on computing power this year alone, even as neither is clearly profitable. As both race toward near-trillion-dollar IPOs, the awkward question gets louder: does this business actually make money?
The costs are extraordinary. Training and running frontier models requires vast amounts of expensive computing power, and the bills are enormous, Anthropic alone has committed about 15 billion dollars a year to a single compute provider. Revenue is growing fast, but so is spending, and much of the cash raised is going straight back out to pay for chips and data centers. The faster they grow, the more they spend.
The bull case is that this is normal for a new platform. Transformative technologies, from railroads to the early internet, often lost money for years while they built out infrastructure and captured markets, before the spending turned into profit. If AI keeps growing and prices hold, today's losses are an investment in dominating a massive future market. Scale first, profit later.
The bear case is that the economics may not work. Competition is fierce, model quality is nearly matched across rivals, and prices are falling as everyone races to be cheaper, which squeezes the revenue side while compute costs stay high. If AI becomes a commodity that customers get cheaply from several providers, no one may earn enough to justify the spending. Racing to the bottom on price is bad for profit.
The honest answer is that no one knows yet. These companies are betting that demand and revenue will eventually dwarf the costs, and they might be right, but they are spending real money now against a payoff that is still a forecast. Investors buying into the IPOs are making the same bet. The valuations assume the profits arrive, and that is not guaranteed.
So the AI boom is running on staggering spending and an unproven path to profit, and the IPO wave is about to test how much investors believe. Fifty billion dollars each in compute, fierce competition, and prices under pressure. The technology is real, the business model is the open question. Watch whether revenue growth finally outpaces the spending.
The AI Giants Are Spending $50 Billion Each. Can They Ever Make Money?
The AI leaders are spending like almost no companies ever have. OpenAI and Anthropic are each on track to spend roughly 50 billion dollars on computing power this year alone, even as neither is clearly profitable. As both race toward near-trillion-dollar IPOs, the awkward question gets louder: does this business actually make money?
