Few people provoke as strong a reaction as Elon Musk. The outspoken Tesla The founder is fond of making bold claims and big promises – promises he doesn’t always keep, much to the chagrin of Tesla investors. Add to that his social media activities, and he is often a divisive figure.
One thing that cannot be denied is that Musk has vision. That – and a huge net worth – allow him to amass vast resources to accomplish incredible tasks. Look no further than the recent successful ‘capture’ of rocket boosters carried out by his company SpaceX. And while that’s truly an incredible feat, Musk recently accomplished something else Nvidia (NASDAQ: NVDA) CEO, Jensen Huang, was deeply impressed.
While X, Tesla and SpaceX dominate the headlines, Musk’s own artificial intelligence (AI) startup.xAI, recently built what could be the most powerful AI training supercomputer available. The supercomputer, called “Colossus,” will be used to train Grok, a large language model (LLM) and the company’s answer to ChatGPT. The LLM will initially be available to paying X customers, but many believe Grok will eventually power Tesla’s humanoid robots. Sounds like science fiction, right? That’s not it.
The thing is, xAI built the facility in a few months, but installed 100,000 Nvidia H100s in just 19 days. The Nvidia chief was recently interviewed about the installation and it was clear that it was something he had never seen before. He called it “superhuman” and “incredible.” Why? According to Huang, it is a process that normally takes years.
However, Musk is not done yet. He plans to install another 50,000 H200s in the coming months, roughly doubling current output. While nothing has been officially announced yet, I wouldn’t be surprised if it’s first in line for Nvidia’s upcoming B200s. xAI has spent a few billion on the project so far, and at a cost of $25,000 each, an estimated $2.5 billion of that was spent on the H100 chips themselves alone.
While Musk and xAI’s performance is undeniably impressive, I would take Huang’s statements with a grain of salt. CEOs of companies that collectively do billions of dollars in business are often quite complimentary to each other, and like Musk, Huang doesn’t shy away from making bold claims.
The entire project shows the incredible lengths companies will go to to win the AI arms race: companies with billions to burn. During the latest quarterly earnings call, CEOs of the tech sector’s largest companies reiterated that losing is not an option, and that they will continue to spend whatever it takes to win. Alphabet CEO Sundar Pichai put it bluntly: “The risk of underinvestment for us is dramatically greater than the risk of overinvestment.” As a result, the company expects to spend roughly $50 billion on capital expenditures this year, up from $31 billion the year before, much of which will go to AI infrastructure and Nvidia chips.
