Tesla confirmed that its Cortex 2 AI training cluster at Gigafactory Texas is now online, marking a new step in the company’s effort to expand in-house compute for autonomous driving, robotics, and other AI workloads. In its Q1 2026 shareholder deck, Tesla listed Cortex 2 as an “Early Ramp” system with installed annual capacity of more than 130,000 H100-equivalent GPUs, alongside Cortex 1 at more than 100,000 H100-equivalent GPUs. Tesla also said Cortex 2 has already started running training workloads and that it continues to develop its Dojo 3 custom silicon to lower training costs over time.
The update ties Tesla’s AI infrastructure buildout directly to its broader Robotaxi and Optimus plans. On the same “Supporting Infrastructure” page, Tesla framed the compute expansion as part of a larger ramp across batteries, materials, charging infrastructure, and mobility services. In the adjacent AI software section, Tesla said it launched FSD (Supervised) v14.3 in April, upgraded reinforcement learning and vision processing, and cut inference latency by up to 20%, while also positioning “Digital Optimus” as a next phase in its AI stack.
Tesla also disclosed a wider semiconductor ambition. The company said it is expanding its manufacturing scope to include semiconductor fabrication, starting with a Tesla-owned Research Fab at Gigafactory Texas, and that in April it completed the final chip design for its next-generation AI5 inference processor. Tesla further said its partnership with SpaceX aims to build what it described as the largest chip fab ever, with vertical integration spanning logic, memory, and advanced packaging. That claim is Tesla’s characterization, but it underscores how aggressively the company is now linking AI compute, chip design, and manufacturing strategy.
- Tesla said Cortex 2 is now online and running training workloads.
- Tesla lists Cortex 2 at more than 130,000 H100-equivalent GPUs in “Early Ramp,” while Cortex 1 is listed at more than 100,000 H100-equivalent GPUs in production.
- Tesla said it is continuing Dojo 3 development to reduce AI training costs over time.
- Tesla said it completed the final chip design for its AI5 inference processor in April.
- Tesla said semiconductor fabrication efforts begin with a Tesla-owned Research Fab on the Gigafactory Texas campus.
- Tesla reported Q1 2026 revenue of $22.387 billion, capital expenditures of $2.493 billion, and cash, cash equivalents, and short-term investments of $44.743 billion.
