Tesla Inc. CEO Elon Musk affirmed that the automaker's early investment in custom inference computers may prove pivotal as artificial intelligence demand accelerates. Musk agreed with an internal engineer's assessment that current compute shortages represent only the initial phase of a broader constraint.

Engineer cites scaling advantage

In response to Tesla engineer Yun-Ta Tsai on X, Musk stated "Yes" regarding the strategic value of Tesla's proprietary hardware. Tsai argued that the company's years of iterating and scaling its own inference computers for each vehicle sold could be "unprecedented" before the superintelligence era. He noted, "The current compute shortage is only the tip of the iceberg. When autonomy becomes indispensable, the real shortage will follow."

Tesla's approach relies on advanced vision and planning AI supported by efficient inference hardware. The company states its custom chips are optimized for performance per watt and mass deployment in cars, reducing reliance on general-purpose processors.

Chip roadmap and Terafab strategy

Tesla's hardware strategy is advancing through its AI5 and AI6 chip generations. The company reported in January that development of both custom inference chips had progressed, with production planned for 2027 and 2028, respectively. Musk has indicated AI5 will "punch far above its weight," primarily optimizing for edge computing in Robotaxi and Optimus applications.

To address long-term supply constraints, Tesla is pursuing the Terafab project. During the second-quarter earnings call, Musk explained that the facility is necessary because Tesla otherwise "simply won't have enough AI chips" to scale Optimus. The project aims to accelerate iteration across logic, memory, packaging, and chip testing.

Reuters reported in April that Tesla, SpaceX, and xAI are pursuing Terafab partly because Musk expects external suppliers cannot satisfy their long-term chip requirements. The broader industry context involves hundreds of billions of dollars flowing into AI infrastructure as companies secure chips, cloud capacity, and power.

What the numbers show

The disclosed figures highlight a divergence between near-term product timelines and long-term infrastructure needs. While AI5 and AI6 production are scheduled for 2027 and 2028, the $3 billion Austin fab investment signals that Tesla anticipates chip scarcity extending well beyond these dates. This suggests the company views semiconductor supply, rather than software capability alone, as the primary bottleneck for scaling autonomous systems.

Tesla Inc. delivered 486,532 units in Q3FY26, a figure investor Gary Black attributes primarily to a 50% rise in gasoline prices since February rather than widespread Full Self-Driving (FSD) adoption.

Black noted that while deliveries exceeded the consensus estimate of 462,000, they remained down 2% year-on-year from the record 497,099 units sold in Q3FY25. The prior year’s peak was fueled by the expiration of the $7,500 Federal EV Credit on September 30, 2025. Black argued that only 35% of Tesla’s sales originate from the US, and unsupervised FSD is approved in a single-digit percentage of the country, limiting its direct impact on global volume.

Competitive Landscape: Tesla vs Waymo

Black highlighted a significant disparity in autonomous driving scale between Tesla and Alphabet’s Waymo. While Tesla has secured approval for unsupervised driving in six metropolitan cities, Waymo operates in 15 US metros.

The investor emphasized that Tesla’s FSD scale-up has fallen short of management guidance, with investors now focused on efforts to expand autonomy against competitors. Rivian Automotive also reported strong Q3 results, beating expectations while maintaining its FY26 guidance of 65,000-70,000 units.

Robotaxi Operational Challenges

Tesla recently adjusted its Austin Robotaxi service cutoff from 10:00 pm to 11:00 pm to address nighttime visibility issues. Elon Musk stated the company is working to improve FSD performance in poor lighting conditions, specifically noting challenges in detecting pets on roads at night. Unlike Waymo, which utilizes LiDAR alongside cameras, Tesla’s FSD relies exclusively on camera-based vision systems.

What the Numbers Show

The divergence between Tesla’s delivery beat and its autonomous scaling metrics suggests that current demand is driven by macroeconomic factors like fuel costs rather than technological superiority in autonomy. With Waymo operating nearly seven times more driverless vehicles across more than double the number of cities, Tesla’s competitive edge in the robotaxi sector appears constrained by regulatory approvals and operational scale despite its larger overall vehicle footprint.