• Post author:
  • Post category:AI World
  • Post last modified:November 29, 2025
  • Reading time:4 mins read

Why Tesla’s AI Engineers Are Flocking to Stealth Sunday Robotics

What Changed and Why It Matters

Tesla just lost more AI and robotics talent. Several senior engineers have left to join a stealth startup called Sunday Robotics.

This isn’t an isolated blip. It points to a broader shift: elite autonomy and embodied AI talent is moving from mega-platforms to focused robotics teams. The implication is simple—breakthroughs in humanoids may come from small, fast groups, not only from incumbents with scale.

“Perry Jia, who worked on Tesla’s Autopilot and Optimus programs for nearly six years, announced last week that he’d left the electric-car maker.”

Here’s the part most people miss: talent flow is the leading indicator. Where the best builders go usually predicts where the next compounding edge will form.

The Actual Move

  • Business Insider reports that senior Tesla AI engineer Perry Jia left after nearly six years across Autopilot and Optimus. Reprints from AOL and MSN underscore the same move.
  • Multiple posts and community threads point to Sunday Robotics—a stealth startup—as the landing zone for several ex-Tesla AI/robotics engineers.
  • Context matters: Forbes previously chronicled a broader engineering exodus at Tesla in 2025, spanning sales, battery, and robotics leadership, as the company reoriented beyond its core EV mission.
  • Earlier in 2025, Ashish Kumar, who led AI for Tesla’s Optimus humanoid robot, also departed.

“Tesla’s AI and robotics divisions are facing a significant ‘brain drain’ as a stealth startup called Sunday Robotics emerges with a roster of [ex-Tesla talent].”

Community chatter: “I know several engineers/engineering managers that left the nightmare called Tesla.” (Reddit anecdote; directional, not definitive.)

What’s concrete: a stealth robotics company is attracting senior autonomy and humanoid engineers from Tesla. The exits include leaders with deep Optimus and Autopilot experience.

The Why Behind the Move

Zoom out and the pattern becomes obvious.

• Model

A focused, humanoid-first robotics startup can align every decision around embodied AI. Smaller teams mean tighter learning loops, faster iteration, and fewer distractions from legacy product lines.

• Traction

Stealth doesn’t mean small impact. Early hires from tier-1 autonomy programs are an indicator of velocity. Talent is the traction for deep-tech before public demos.

• Valuation / Funding

No public funding yet, but the market context matters: Figure AI, Agility Robotics, and others have raised sizable rounds. Capital is chasing humanoids. That gives new entrants room to staff up and build quietly.

• Distribution

Humanoid use cases lean enterprise-first—logistics, manufacturing, and high-mix automation. Distribution will hinge on pilot-to-fleet expansion with a few anchor customers rather than mass consumer channels.

• Partnerships & Ecosystem Fit

Startups win by stitching commodity hardware, custom actuation, and frontier foundation models into a system-level product. Expect deep supplier and integrator relationships early.

• Timing

We’re in the post-foundation-model phase: the next edge is embodiment. As models gain world understanding, the bottleneck shifts to data from real-world interaction. Teams that can harvest on-robot data loops will compound fastest.

• Competitive Dynamics

Tesla’s Optimus aims for scale via vertical integration. Startups counter with focus and top-decile talent density. In California, non-competes are largely unenforceable—so recruitment velocity becomes a real competitive lever.

• Strategic Risks

Robotics is capital- and cycle-time intensive. Stealth buys time but not physics. The risks: supply chain, safety certification, unit economics, and the gap between lab demos and reliable, low-latency manipulation in messy environments.

What Builders Should Notice

  • Talent flow is the earliest KPI. Follow where senior builders choose to spend their next 5–7 years.
  • Focus beats scope. One platform problem, end-to-end ownership, tight loops.
  • Data flywheels shift to embodiment. Real-world interaction data becomes the new moat.
  • Distribution is earned via operations. Nail one high-friction workflow before scaling SKUs.
  • Culture is a weapon. Low-bureaucracy, high-agency teams out-iterate even the best-funded incumbents.

Buildloop reflection

“The next AI edge isn’t bigger models—it’s tighter loops between code and the real world.”

Sources

Tags: , , , , , , ,