Nvidia Denies Upfront Payment Requirement for H200 AI Chips Amid China Export Scrutiny 
Trump Pushes Tech Giants to Absorb AI Data Center Power Costs, Citing Microsoft Changes 
TikTok Expands AI Age-Detection Technology Across Europe Amid Rising Regulatory Pressure 
Anthropic Appoints Former Microsoft Executive Irina Ghose to Lead India Expansion 
Federal Judge Clears Way for Jury Trial in Elon Musk’s Fraud Lawsuit Against OpenAI and Microsoft 
TSMC Set to Post Record Q4 Profit as AI Chip Demand Accelerates 
xAI Restricts Grok Image Editing After Sexualized AI Images Trigger Global Scrutiny 
Micron to Buy Powerchip Fab for $1.8 Billion, Shares Surge Nearly 10% 
Trump Administration Approves Nvidia H200 AI Chip Sales to China Under New Export Rules 
South Korea Seeks Favorable U.S. Tariff Terms on Memory Chip Imports 
Proposed Rio Tinto–Glencore Merger Faces China Regulatory Hurdles and Asset Sale Pressure 
China’s AI Models Narrow the Gap With the West, Says Google DeepMind CEO 
One Percent Rule Checklist For Safer Forex Trading Risk 
California Attorney General Orders xAI to Halt Illegal Grok Deepfake Imagery 
Google Seeks Delay on Data-Sharing Order as It Appeals Landmark Antitrust Ruling 
White House Pressures PJM to Act as Data Center Energy Demand Threatens Grid Reliability 
China Halts Shipments of Nvidia H200 AI Chips, Forcing Suppliers to Pause Production 
Tesla Inc (NASDAQ: TSLA) is set to restart development of its next-generation supercomputer, Dojo3, marking a renewed push into artificial intelligence infrastructure. Tesla CEO Elon Musk confirmed on Sunday that work on Dojo3 will resume now that the company’s AI5 chip design has reached a stable stage. Musk shared the update via social media, adding that Tesla is actively recruiting new talent to support the project.
The announcement comes just months after Tesla shut down the Dojo program and disbanded the team behind it. At the time, Musk described Dojo2 as an “evolutionary dead end,” signaling that the previous architecture failed to meet the company’s long-term AI and self-driving ambitions. The renewed effort suggests Tesla has reassessed its strategy and sees Dojo3 as a more viable path forward.
Dojo was first deployed in 2023 as a custom-built supercomputer designed to train Tesla’s artificial intelligence systems, including its Full Self-Driving (FSD) and driver assistance software. The system plays a critical role in processing massive volumes of video data collected from Tesla vehicles worldwide, enabling rapid improvements in machine learning models.
Musk’s latest comments indicate that Dojo3 may be built around Tesla’s in-house AI5 chip, which he has previously described as a significant leap over the company’s current hardware. While Tesla had earlier floated the idea of Dojo2 using Tesla-designed chips, the AI5 appears to be the cornerstone for the third-generation system.
Tesla currently operates several large-scale computing clusters and relies on a hybrid approach, using both its own chips and those supplied by NVIDIA Corporation (NASDAQ: NVDA). Musk has emphasized that Tesla does not plan to replace Nvidia’s data center GPUs entirely. Instead, the company aims to complement them with custom silicon optimized for its specific AI workloads.
The AI5 chip is expected to be manufactured in collaboration with leading semiconductor players, including TSMC (NYSE: TSM) and Samsung Electronics (KS: 005930). If successful, Dojo3 could strengthen Tesla’s vertical integration in AI, reduce long-term computing costs, and accelerate progress in autonomous driving technology, reinforcing the company’s competitive position in the rapidly evolving AI and electric vehicle landscape.
