Elon Musk Claims He Just Activated the World's Most Powerful AI Supercomputer

Hello!

Behold Colossus: Elon Musk’s new supercomputer, allegedly powered by a staggering 100,000 Nvidia AI chips—more than any single AI system on the planet.

Built in Tennessee for his artificial intelligence startup xAI, Musk announced that the formidable data center was finally brought online over the Labor Day weekend. The entire assembly took just 122 days—a record, according to Nvidia.

“Colossus is the most powerful AI training system in the world,” Musk said in a post.

Powering the Next Generation of AI

The supercomputer runs on Nvidia H100 graphics processing units, the industry’s most sought-after hardware for training and running generative AI systems such as chatbots and image generators.

xAI’s current cluster of 100,000 chips is only the starting point. Musk stated that within a few months Colossus will double in size to 200,000 AI chips. This expansion will include 50,000 H200 GPUs—the newer generation that Nvidia says offers nearly twice the memory capacity and 40 percent more bandwidth than the H100.

Fast Learners

Musk founded xAI in summer 2025. Its flagship product, Grok, is a conversational AI chatbot integrated into X (formerly Twitter). That xAI has quickly reached hardware parity with established players such as OpenAI and Microsoft is notable—provided Musk’s well-known tendency toward bold claims is not overstated.

As Fortune notes, Nvidia already counts Musk among its top customers: he had previously purchased tens of thousands of GPUs for Tesla, worth an estimated $3–4 billion, before launching xAI. Some of those chips, originally earmarked for Tesla’s Full Self-Driving system, were later used to train an early version of Grok.

Securing the additional 100,000 H100 GPUs for Colossus likely required several billion dollars more, given each chip’s approximate $40,000 price tag. xAI raised around $6 billion in a May funding round backed by prominent venture firms including Andreessen Horowitz.

Claim to the Throne

The launch was not without controversy. Last week, residents near the Memphis, Tennessee facility complained of “untenable levels of smog,” hinting at possible future environmental disputes.

Colossus’s status as the world’s most powerful AI training system will also face competition. Other major players—OpenAI, Microsoft, Google, and Meta—already operate hundreds of thousands of GPUs and show no sign of slowing down.

Microsoft reportedly aims to amass 1.8 million AI chips by the end of 2026. In January, Mark Zuckerberg indicated that Meta plans to purchase an additional 350,000 Nvidia H100 GPUs by the same deadline.

For now, Colossus stands as a singular demonstration of raw computing power. According to Fortune, it will be used to train Grok-3, which Musk targets for release in December 2026.

Thank you!
Join us on social media!
See you!