In a dramatic shift signaling the next phase of Tesla’s evolution, Elon Musk has undertaken a sweeping realignment of the company’s AI strategy—discarding its famed in-house Dojo supercomputer to embrace a new age of external partnerships and advanced robotics. This “giant AI change” fundamentally reshapes Tesla’s roadmap and may redefine the company’s place in the tech-driven future.

From Dojo to Disbandment: Tesla Rethinks Its Supercomputer Bet

Tesla’s Dojo project, once heralded as the backbone of its AI ambitions, has been officially shut down. This move follows the departure of Dojo’s leader and about 20 key engineers who formed a new venture, DensityAI. Elon Musk called the project an “evolutionary dead end” and indicated that continuing to divide resources toward multiple AI chip strategies no longer made sense.

Analysts view this as a strategic pivot rather than a setback. Tesla will now prioritize its next-generation AI5 and AI6 chips, designed to handle both training and inference tasks. The company plans to >shift reliance onto external chip suppliers—notably Nvidia, AMD, and Samsung—for compute capabilities, effectively transforming its infrastructure approach.

Biden-Scale Chip Deal: Samsung Powers Tesla’s AI Future

Central to the new strategy is a massive $16.5 billion agreement with Samsung, under which the tech giant will manufacture Tesla’s AI6 chips at a dedicated Texas plant through 2034. These chips will play a vital role in Tesla technologies spanning Full Self-Driving (FSD), Optimus robots, and the future Dojo-equivalent infrastructure

An interim AI5 chip—produced by TSMC in Arizona—will serve as a bridge until the AI6 rollout begins. Elon Musk has emphasized that the proximity of the Texas fab to his residence ensures accelerated progress

 A Robotaxi Revolution: Autonomous Ride-Hailing Moves Forward

Tesla’s robotaxi initiative, long teased as a future cornerstone, has taken its first tentative steps. In June 2025, Tesla launched a limited robotaxi service in Austin, Texas, using FSD-equipped Model Ys staffed by safety monitors—still a supervised system, but a tangible pilot nonetheless.

Elon Musk has ambitious expansion plans: by year-end, the service is expected to cover a significant portion of the U.S. population. Fully autonomous rides (Level 4/5) are projected to arrive once regulatory approval is secured.

Better Brains Behind The Wheel: A Brand‑New FSD Model

On the AI software front, Tesla is preparing to deploy a dramatically upgraded FSD model, with Musk confirming that the new version will feature 10 times more parameters and enhanced video processing. This model, if testing continues successfully, could roll out publicly in a matter of weeks.

This upgrade is pivotal—expectations are high that the improved model will bring Tesla closer to achieving unsupervised autonomy in real-world conditions.

 Robots Rising: Optimus Stepping into Reality

Tesla’s AI ambitions go beyond driving; the Optimus humanoid robot remains a key element of Musk’s “super‑tsonic tsunami” vision for AI-driven transformation. Tesla projects that it will produce several thousand Optimus units by the end of 2025—surpassing 10,000 robots remains the aspirational goal.

At a recent Tesla event, Musk even described a hypothetical future where annual global production of such robots could reach billions—each priced around $30,000.

What’s Driving This “Giant AI Change”?

Analysts contend that discontinuing Dojo may stem from both technical delays and organizational attrition—repeated leadership departures created a critical turning point. The reallocation toward external vendors may offer faster, more scalable solutions.

The Samsung deal not only secures capacity but also stabilizes Tesla’s chip roadmap—reducing risk and enabling focused advancement across AI-driven initiatives.

Furthermore, combining more powerful FSD models with an expanding robotaxi fleet aligns Tesla with its long-promised autonomy future—now approaching reality.

What’s Next: Watchpoints and Risks

Testing & Regulation: The robotaxi expansion depends heavily on regulatory green lights and public trust, especially given early driving glitches in Austin.

Execution Risks: The new FSD model and chip architecture still require seamless integration across Tesla’s ecosystem—a challenging feat amid mounting competition from Waymo, BYD, and others

Hardware Upgrades Needed: Musk has admitted that older vehicles (with HW3) may need hardware replacements for new FSD capabilities—a logistical challenge with potential customer dissatisfaction.

Robotaxi Scaling: Transitioning from supervised to unsupervised autonomy remains uncertain, and Tesla must also deliver on production of Cybercab, Robovan, and optimized updates for existing vehicle lineup.

Conclusion

Elon Musk’s sweeping AI shift at Tesla marks a monumental transformation—from in-house compute to partnerships, from delayed promises to tangible deployments. While the Dojo project may be no more, the robotaxi service, upgraded FSD model, Optimus robots, and AI6 chip infrastructure signal that Tesla’s AI revolution is very much alive.

As Musk’s vision transitions from bold presentations to real-world rollout, the question remains: can Tesla deliver on these transformative promises—and in doing so, redefine the future of mobility?