Tesla is significantly accelerating its ambitions in custom chipmaking, with CEO Elon Musk confirming that the Dojo supercomputer project—now referred to as Dojo3—is back on track. The move signals Tesla’s intent to compete directly with industry leaders like NVIDIA by developing high-performance, cost-efficient AI silicon tailored for its vehicles, robotics, and data centers.
Musk revealed that Tesla plans to advance multiple generations of in-house chips, beginning with the upcoming AI5 processor. According to him, AI5 is designed to deliver performance comparable to NVIDIA’s Hopper architecture in a single-chip configuration, while a dual-die setup could rival NVIDIA’s latest Blackwell platform—at a fraction of the cost. Musk described the pricing as “peanuts,” highlighting Tesla’s goal of achieving a structural cost advantage through vertical integration.
The Dojo project had previously been paused as Tesla leaned heavily on NVIDIA hardware for AI training. However, growing compute demands—driven by Full Self-Driving (FSD), Optimus humanoid robots, and AI model training—have pushed Tesla to recommit to its own silicon roadmap. Dojo3 is expected to rely on AI5-based clusters, creating a unified compute architecture across vehicles and data centers.
Musk also outlined an aggressive development cadence, claiming Tesla aims to scale its chips through AI9 with updates roughly every nine months, mirroring NVIDIA’s rapid innovation cycle. By controlling chip design and supply, Tesla believes it can better optimize performance, cost, and scalability as its AI-driven products mature.
While Musk’s vision positions Tesla as a potential future chipmaking powerhouse, success will hinge on flawless execution across design, validation, and manufacturing—an area where even seasoned semiconductor firms face steep challenges.
See What’s Next in Tech With the Fast Forward Newsletter
Tweets From @varindiamag
Nothing to see here - yet
When they Tweet, their Tweets will show up here.
