On Tuesday, Tesla stock underperformed, falling in line with the S&P 500 and Nasdaq declines as investors weighed profit‐taking ahead of several key events. Trading volume on the Nasdaq reached 38 million shares, 12% above its 30‐day average, signaling heightened selling pressure. In response, CEO Elon Musk doubled down on Tesla’s in-house AI strategy, confirming accelerated development of the company’s custom AI5 inference chips. He noted that the new silicon will power both next-generation Autopilot features and the Dojo training supercomputer, underscoring Tesla’s pivot toward becoming a vertically integrated AI hardware leader.

Elon Musk publicly criticized established automakers for declining his offer to license Tesla’s Full Self-Driving software, calling their reluctance “crazy.” Yet recent industry announcements reveal the rationale: Ford plans to deliver eyes-off highway driving by 2028 with an in-house system that it says will cut costs by 30% and improve integration across sensors, compute and actuators. Rivian is developing a proprietary autonomous-driving chip and even exploring a robotaxi business, while Mercedes-Benz will deploy Nvidia’s new Alpamayo AI toolkit to accelerate its own ADAS rollout. Analysts at Bloomberg Intelligence estimate that automakers can save up to 15% on sensor-to-software integration by keeping development internal.

Following Canada’s decision to eliminate 100% duties on Chinese‐made electric vehicles and set a 49,000‐vehicle annual quota at a 6.1% levy, Tesla stands to gain market share. The company already ships all-wheel-drive Model Y variants from its Shanghai Gigafactory and operates 39 retail locations across Canada. In 2023, Tesla drove a 460% year-over-year surge in Chinese EV imports through Vancouver; under the new regime, lower‐cost Model 3 versions built in China will face no punitive duties, reinforcing Tesla’s price competitiveness against absent Chinese rivals like BYD and Nio.

Elon Musk confirmed that progress on Tesla’s AI5 inference chip design has unlocked the resumption of the Dojo3 supercomputer project, which had been paused early last year. According to internal engineering updates, AI5 delivers a 40% improvement in TOPS (trillions of operations per second) per watt relative to its predecessor, enabling Dojo3 to achieve an exascale training throughput target of 1.2 exaFLOPS. The initiative underpins Tesla’s ambition to train neural networks for FSD and future humanoid robotics more cost‐effectively, with a projected 25% reduction in per-petabyte training costs once Dojo3 is fully operational.