The new project, xAI, by Elon Musk, the billionaire entrepreneur behind Tesla and SpaceX, has launched a powerful AI training system called Colossus, per a tweet on X yesterday, September 2. This system is so advanced that it has set a new benchmark in the artificial intelligence subsector.

Colossus is an AI training system that uses a whopping 100k Nvidia H100 graphics processing units (GPUs). To put it simply, GPUs are like the brain of a computer, and having 100,000 of them makes Colossus incredibly powerful. Elon Musk’s team at xAI built this system in just 122 days, which is a remarkable achievement.

This weekend, the @xAI team brought our Colossus 100k H100 training cluster online. From start to finish, it was done in 122 days. Colossus is the most powerful AI training system in the world. Moreover, it will double in size to 200k (50k H200s) in a few months. Excellent…

— Elon Musk (@elonmusk) September 2, 2024

Colossus is important because it is now the world’s most powerful AI training system. It surpasses other major models, including those from OpenAI, which uses 80,000 GPUs. This means that Colossus can process more data and learn faster than any other AI system out there.

Image source: X (formerly Twitter)

Nvidia’s Role and the Specs of Colossus 

Nvidia, a leading company in making computer chips, partnered with xAI to build Colossus. Nvidia’s H100 GPUs are some of the best in the market, and they play a sensitive role in making Colossus so powerful. You’ll recall that Coinfomania reported that Nvidia suffered a setback in the release of its next-generation “Blackwell” B-200 artificial intelligence chip early in August this year. Nvidia and its CEO, Jensen Huang, congratulated Musk and his team, saying, “This is a huge accomplishment in the field of AI.”

Exciting to see Colossus, the world’s largest GPU #supercomputer, come online in record time. Colossus is powered by @nvidia's #acceleratedcomputing platform, delivering breakthrough performance with exceptional gains in #energyefficiency.Congratulations to the entire team! https://t.co/UXHtPCELly

— NVIDIA Data Center (@NVIDIADC) September 2, 2024

Colossus uses 100k Nvidia H100 graphics processing units (GPUs). These GPUs are liquid-cooled and connected via a single RDMA (Remote Direct Memory Access) fabric. The system’s power consumption is estimated to range between 42 to 56 megawatts. The initial phase provides 50 megawatts, with plans to scale up to 150 megawatts. The Colossus supercomputer is located at the former Electrolux facility in Memphis, Tennessee. 

Memphis was chosen for its robust power infrastructure, supported by Memphis Light, Gas, and Water (MLGW) and the Tennessee Valley Authority (TVA). The city offered attractive economic development packages, recognizing the project’s potential to transform Memphis into a tech hub. The former Electrolux facility, acquired by Phoenix Investors, provided a ready-to-use industrial space, accelerating the project’s timeline.

The Colossus Project is a Pacesetter 

The project plans indicate a phased approach to power usage, starting with 50 megawatts available by August 2024 and ultimately aiming for 150 megawatts. The xAI facility is expected to be a multibillion-dollar investment and the largest new-to-market company investment in Memphis’s history.

Elon Musk has big plans for Colossus. He announced that the system will double in size in the next few months, reaching 200,000 GPUs by incorporating an additional 50,000 H200 GPUs. This will make Colossus even more powerful and capable of handling more complex tasks.

Impressive! Congratulations to you and the rest of the @xai team on the first of many big announcements ahead. https://t.co/AImAsrIiAN

— Cathie Wood (@CathieDWood) September 2, 2024

The AI community has been agog with excitement about Colossus. Cathie Wood, the CEO of ARK Invest, called it “impressive” and hinted at “big announcements ahead.” Many experts believe that Colossus will push the boundaries of what AI can do and lead to technological breakthroughs.

The post Elon Musk’s xAI Colossus Sets New Benchmark in AI with 100,000 GPUs As It Goes Live  appeared first on Coinfomania.