Tesla AI staff brace for a challenging year

Staff at Tesla AI prepare for a big year

In Q3 2025, Tesla reported that its AI division is focused on three priorities: advancing Full Self‑Driving (FSD), scaling the Dojo supercomputer, and developing the Optimus humanoid robot. These initiatives are tied directly to Tesla’s 2026 goal of launching a Cybercab robotaxi.

The main external challenge is regulatory. The National Highway Traffic Safety Administration (NHTSA) has not yet granted blanket approval for driverless cars.

Instead, it continues to issue exemptions and guidance via NHTSA’s Part 555 exemption process and related safety rules; Tesla must demonstrate safety to obtain exemptions for vehicles without traditional controls.

The FSD system is the hardest challenge

In October 2025, Tesla began rolling out FSD (Supervised) v14, which still requires driver attention. The system expanded features with Hardware 4, but still requires human oversight.

Tesla’s 2026 target is to make FSD fully autonomous for robotaxi use. Achieving this requires solving edge cases and rare driving events, such as unusual road debris or unpredictable pedestrian behavior, without requiring human intervention.

The Dojo supercomputer must scale up

Tesla built Dojo to process fleet video for FSD training. In Aug. 2025, major outlets reported Tesla disbanded most of the Dojo team; Tesla’s Q3’25 deck instead highlights its Cortex training cluster (81k H100-equivalents) and a chip-manufacturing deal with Samsung.

This shift reflects the high cost of AI computing, with billions of dollars spent on specialized chips. Tesla’s challenge for 2026 is ensuring its new hardware strategy can deliver the scale needed to train autonomous driving models.

The Robotaxi must be approved by law

Tesla’s planned Cybercab robotaxi, designed without a steering wheel or pedals, is being prepared at Giga Texas. In 2025, Elon Musk projected production could begin in April 2026, though this remains a target rather than a regulatory approval.

U.S. federal rules still require driver controls unless exemptions are granted. Tesla must demonstrate safety under NHTSA’s AV Framework before the Cybercab can operate commercially.

The Optimus robot is another big challenge

Tesla showcased an Optimus R&D line in 2025 and discussed future scale-up; a dedicated factory timeline has not been formally disclosed in filings. Executives stated ambitions to scale to millions of units annually by the end of the decade.

Optimus is intended to perform repetitive tasks on assembly lines, such as carrying parts or using tools. The challenge is making the robot reliable enough to improve car production efficiency.

Intense competition is growing in 2025

In 2025, Waymo expanded fully driverless ride‑hailing services in Phoenix and other U.S. cities, while Cruise faced setbacks after suspensions in 2023-24. These competitors rely on LiDAR and radar, contrasting Tesla’s camera‑only vision approach.

Tesla’s AI team is under pressure to prove its vision‑based system can match or surpass rivals. The race for robotaxi dominance is intensifying as regulators and consumers compare safety records and the scale of deployment.

Tesla’s vision‑only AI is a high risk

Tesla continues to pursue a camera‑only vision approach for Full Self‑Driving (FSD), rejecting LiDAR and radar. Tesla’s support materials emphasize a camera-centric Tesla Vision approach; FSD (Supervised) requires active driver supervision.

Competitors like Waymo expanded their robotaxi services to five new U.S. cities in 2025, utilizing LiDAR and radar. Tesla’s challenge is to prove that cameras alone can perform reliably in adverse weather conditions, such as rain, fog, and snow.

The team must manage huge amounts of data

Tesla reported in Q3 2025 that FSD (Supervised) logged 1.3 billion miles in one quarter, averaging 14.1 million miles per day, up from 11 million in Q2. This data comes from Tesla’s fleet of over 4 million vehicles worldwide.

Managing petabytes of video requires identifying and addressing rare edge cases, then feeding them into AI training pipelines to optimize performance. This scale of data is central to Tesla’s effort to validate safety and improve autonomy.

AI training still needs many human labelers

Despite the prevalence of automation patents, Tesla continues to utilize large human labeling teams alongside automated labeling; the current headcount for 2025 has not been publicly specified. Job postings confirm full‑time teams labeling pedestrians, traffic signs, and lane markings.

This human‑in‑the‑loop process remains critical to FSD training. While Tesla is developing automated labeling systems, human oversight ensures accuracy in complex edge cases.

The Dojo computer uses immense power

Tesla’s Dojo supercomputer, launched in 2023, was designed to process millions of terabytes of fleet video. By 2025, Tesla operated clusters with 5,760 Nvidia A100 GPUs, consuming a massive amount of electricity.

Bloomberg reported in Aug 2025 that Tesla disbanded the Dojo team, citing it as an evolutionary dead end. Energy costs for AI compute remain in the hundreds of millions annually, forcing Tesla to pivot to new AI6 chips.

Tesla is in a talent war for AI staff

The AI talent war intensified in 2025, with reports indicating that 74% of employers struggled to hire skilled AI engineers, and salaries for AI roles averaged 25% higher than those for non-AI tech jobs.

OpenAI and xAI poached senior Tesla engineers in mid-2025, and with rivals offering multimillion-dollar packages, talent retention has become a critical challenge for Tesla.

The AI team must prove FSD is safe

In Oct 2025, NHTSA opened a probe into 2.88 million Tesla vehicles equipped with FSD after 44 reported collisions. Regulators require Tesla to demonstrate safety beyond human drivers.

Tesla’s own Safety Report claims accident rates per mile with Autopilot engaged are lower than human averages, but independent studies (RAND 2025) emphasize the need for statistically significant validation.

Want to see how AI could make driving safer? Dive into the latest insights from Tesla and Rivian.

A high‑pressure year for Tesla AI

Tesla’s Q3 2025 Update confirmed record deliveries and $41.6 billion in cash reserves, but also highlighted $9 billion in CapEx for AI6 chip development and Optimus factory construction.

In 2026, Tesla AI must deliver on FSD autonomy, scale Optimus production, and prepare Cybercab robotaxi rollout, all while facing regulatory probes and competition from Waymo’s expanding LiDAR‑based fleet.

Curious about when Tesla’s Cybercab might hit the streets? Check out the latest on its early rollout plans.

How do you feel about Tesla balancing record deliveries with massive investments in AI and robotaxis? Share your thoughts in the comments.

Read More From This Brand:

• Tesla plans Robotaxi testing in five more U.S. cities
• Features people love about EVs
• Understanding battery lifespan and replacement costs in EVs

Don’t forget to follow us for more exclusive content right here on MSN.

If you liked this article, you’ll LOVE our free email newsletter.

This slideshow was created with the assistance of AI and human input.