×
Tesla AI chief Milan Kovac leaves after 9 years leading FSD development
Written by
Published on
Join our daily newsletter for breaking news, product launches and deals, research breakdowns, and other industry-leading AI coverage
Join Now

Tesla’s VP of AI and Optimus Engineering Milan Kovac announced his departure from the company this week after nine years leading the development of Full Self-Driving (FSD) technology and the Optimus humanoid robot program. The resignation of the Belgian engineer, who was instrumental in Tesla’s autonomous driving evolution since 2016, represents another high-profile executive exit as the company consolidates its AI leadership under Ashok Elluswamy, Tesla’s director of Autopilot software.

What you should know: Kovac cited family obligations as his primary reason for leaving, emphasizing his need to spend more time with family abroad after years away from home.

  • He joined Tesla in 2016 during a pivotal moment when the company launched its second-generation Autopilot system built on Nvidia’s Drive PX2 platform.
  • Kovac personally handled the complex migration of code and functionality from Tesla’s original Autopilot 1.0 system to the new architecture.
  • With his departure, Ashok Elluswamy will assume responsibility for both the FSD program and Optimus robot initiatives, consolidating Tesla’s AI development under single oversight.

Key technical contributions: Kovac earned recognition as one of Tesla’s Autopilot “Three Musketeers” alongside Andrej Karpathy, the OpenAI co-founder, and Ashok Elluswamy for his role in developing the company’s first in-house FSD chip.

  • By 2018, he had become central to Tesla’s HW 3.0 chip development, leading engineering efforts from initial installation to mass production across Tesla’s entire vehicle lineup.
  • The transition required coordinating hardware integration with software optimization across multiple vehicle platforms simultaneously.
  • His technical expertise extended from core Autopilot engineering to leading the Optimus robotics division, which he described as building “an entire platform: hardware & software all at once” from minimal initial resources.

Executive tensions revealed: Kovac’s relationship with Elon Musk wasn’t always smooth, as documented in Walter Isaacson’s biography of the Tesla CEO.

  • During preparations for Tesla’s 2021 AI Day presentation, tensions reached a breaking point when Musk harshly criticized Kovac’s technical presentation slides, calling them “So boring!”
  • Tesla executives Lars Moravy and Pete Bannon intervened to prevent Kovac’s resignation, helping restructure the presentation after sharing drinks.
  • The incident illustrates the intense pressure and expectations that define Tesla’s executive environment, where technical excellence must align with Musk’s vision for public communication and talent acquisition.

The bigger picture: Executive departures aren’t unusual at Tesla, with former leaders including J.B. Straubel, Jerome Guillen, and Doug Field having moved on without causing major operational disruptions.

  • However, Kovac’s technical contributions span nearly a decade of Tesla’s most ambitious AI projects, making his departure particularly significant for the company’s autonomous driving and robotics ambitions.
  • The consolidation of AI leadership under Elluswamy suggests Tesla is streamlining its technical organization as it advances both FSD and Optimus programs.

What they’re saying: In his departure announcement, Kovac maintained strong support for Musk and the Tesla team while explaining his personal decision.

  • “I’ve been far away from home for too long, and will need to spend more time with family abroad,” he explained.
  • Musk’s response acknowledged Kovac’s decade-long contributions while expressing gratitude for his technical leadership.
Tesla AI VP Milan Kovac Resigns After 9 Years Leading FSD and Optimus Projects

Recent News

AI startups reach $100M revenue in year one—rewriting growth rules

Consumer AI apps are generating substantial revenue from launch, ditching the freemium playbook.

Anthropic launches Claude Gov for US classified intelligence operations

The models "refuse less" when handling sensitive material, removing safety restrictions that block consumer versions.