xAI's Appetite: How Elon Musk Burned $9.5 Billion in Nine Months

If you thought your electricity bills were high this winter, take a look at xAI's financials. According to data the company provided to its investors, which leaked through Chinese financial media including Sina Finance, Elon Musk's startup managed to spend about $9.5 billion in just nine months of last year. To put the scale in perspective: that is more than a billion dollars a month. That burn rate makes xAI one of the most expensive startups in human history, even by Silicon Valley standards.

Let's work out where the money went, because you cannot spend that much on free lunches for employees.

The biggest line item is, of course, silicon. Musk makes no secret of his obsession with compute and is actively building the Colossus supercomputer in Memphis. It required tens of thousands of Nvidia H100 chips, each of which costs as much as a decent crossover. In modern AI there is a simple rule: if you don't have your own "iron fist" of hundreds of thousands of GPUs, you are simply not in the race to build powerful artificial intelligence. Musk decided not just to take part but to take the head of the table right away, whatever it costs his investors.

Context matters here more than ever. xAI was launched as a response to OpenAI, which Musk now considers too closed and "woke." But ideological disputes are expensive. While OpenAI spent years building infrastructure with Microsoft's backing, Musk had to do it in blitzkrieg mode. That explains why xAI's spending looks so aggressive next to its competitors. The company had to do more than hire top engineers, poaching them from Google and Tesla with huge stock options. It also had to build, in the shortest possible time, infrastructure that took others years.

Investors don't seem to be protesting so far. Last year xAI raised $6 billion in a Series B round, and by all appearances that money has already largely turned into heat from servers. The market believes in the "Musk factor" and in the idea that Grok-3 can deliver results that beat GPT-4. Still, this financial strategy raises the question of how long the party of hyper-capitalism can last. We are watching a unique moment in tech history, when a product's success depends directly on how many billions you are willing to turn into computing power in a single quarter.

What does this mean for the industry as a whole? First, the barrier to entry for new players is now practically insurmountable without the backing of sovereign wealth funds or tech giants. Second, we are seeing a new economic model take shape, in which net profit takes a back seat to the prospect of dominating AGI. Musk is betting that whoever first builds a truly smart machine can recoup any billion-dollar spending within weeks. For now, his strategy looks like an attempt to break through a wall with a golden battering ram.

The bottom line: xAI is spending money faster than any other company in AI history at the same stage. Will investors have the patience if Grok-3 doesn't become an instant sensation?