Over the weekend, Elon Musk revealed the name of Tesla’s much-anticipated supercomputing cluster at Gigafactory Texas.
In a post on X (formerly Twitter) on Saturday, Musk announced that the cluster will be called “Cortex,” following his personal inspection of the new facility.
Massive Computing Power for AI Development
This confirmation puts to rest months of speculation surrounding the building’s purpose. The Cortex supercomputer is expected to house an impressive array of approximately 100,000 Nvidia H100 and H200 chips.
These chips will be dedicated to video training of neural networks, powering the development of Tesla’s Full Self-Driving (FSD) technology and the Optimus humanoid robot.
Unprecedented Cooling Requirements
The immense computing power of the Cortex cluster comes with equally significant cooling demands. Musk has previously detailed the need for around 130 MW of power and cooling for the cluster this year, with plans to scale up to over 500 MW within the next 18 months.
The need for such massive cooling is evident in the large fans recently installed in the building, as documented by drone operator and Giga Texas observer Joe Tegtmeyer.
Advanced Cooling Infrastructure
In addition to the fans, the Cortex supercomputer will utilize four large water tanks located on the second floor of the facility to enhance cooling efforts.
These tanks will be instrumental in maintaining optimal temperatures for the high-performance computing hardware.
Significant Investment in AI
In June, Musk estimated that Tesla’s purchases of Nvidia chips would likely fall between $3 billion and $4 billion this year. This investment represents a significant portion of the $10 billion Tesla plans to spend on AI-related endeavors in 2024.
Approximately half of this budget will be dedicated to internal projects, including Tesla’s custom-designed AI inference computer, the sensors integrated into all of its vehicles, and the development of its Dojo supercomputer system.
Musk’s Dedication to AI Advancement
Musk shared a video of the Cortex supercomputing facility after the Fourth of July, indicating he had been working on the holiday. This highlights his commitment to the advancement of Tesla’s AI capabilities, even during national celebrations.