First Round Capital led an $11 million seed round for Proception, a San Francisco-based startup building robotic hands with human-level dexterity, just weeks after founder Jay Li settled a trade secret lawsuit Tesla filed against him in early 2025. The settlement, finalized in May 2026, ends a dispute that began when Li left Tesla's Optimus humanoid robot division to launch his own company. Neither party disclosed terms, though court filings show Tesla withdrew all claims and Li agreed to cooperate with a forensic review of his devices. The funding round closed in June 2026, with participation from Lux Capital and several angel investors with robotics backgrounds at Boston Dynamics and DeepMind.
Li spent two years at Tesla working on end-effector design for Optimus, the humanoid robot Elon Musk first demonstrated in prototype form in 2022. He left in January 2025, six weeks before Tesla filed suit alleging he copied proprietary CAD files and sensor integration schematics. The complaint cited forensic evidence showing Li accessed internal design repositories days before his departure, though it never specified which files he allegedly took. Li's attorneys countered that he worked exclusively on tactile sensing algorithms—software, not hardware—and that Tesla was conflating his personal research notes with company IP. The case dragged through discovery for fourteen months before both sides agreed to dismiss with prejudice in late spring 2026. Legal observers say the swift resolution suggests Tesla's claims were thin, or that Li's backers were willing to pay to make the problem disappear. Either way, the settlement clears the path for Proception to operate without the shadow of ongoing litigation.
Proception's technical approach differs sharply from Tesla's published work on Optimus hands. Where Tesla emphasized actuator density and grip strength in its 2024 hardware updates, Proception is building hands optimized for fine motor control and real-time force feedback. Li has said publicly that his team uses a distributed sensor array embedded across each finger segment, allowing the system to detect slip, texture, and contact geometry with millisecond latency. The company has not released detailed specs, but early prototypes shown to investors feature eleven degrees of freedom per hand and a claimed force resolution below 0.1 newtons. That puts Proception in the same performance tier as Shadow Robot's Dexterous Hand, a research platform that costs north of $100,000 per unit and requires external control hardware. Proception's pitch is that it can hit similar benchmarks at a fraction of the cost by using off-the-shelf motors and custom-designed sensor substrates that integrate into standard robotic arm interfaces. The company is targeting warehouse automation and light assembly applications first, where delicate part handling justifies premium pricing. A pilot deployment with an unnamed logistics customer is scheduled for late 2026, according to a source familiar with the partnership.
Tesla, meanwhile, has yet to show Optimus performing tasks that require genuine dexterity. Demonstrations at the company's 2025 shareholder meeting featured the robot folding a shirt and sorting blocks, but both tasks used preprogrammed motions rather than reactive manipulation. Industry engineers who have examined publicly available footage say the Optimus hand lacks sufficient tactile sensing to adjust grip force dynamically, a prerequisite for handling variable or fragile objects. Tesla has published no peer-reviewed research on its end-effector design, and the company's AI Day presentations in 2023 and 2024 focused almost entirely on locomotion and vision systems. The gap between Tesla's marketing narrative and its demonstrated capabilities creates an opening for startups like Proception, especially as humanoid robots move from research labs into commercial pilots where reliability and task flexibility matter more than headline-grabbing demos. First Round's investment thesis, according to a partner at the firm, centers on the belief that robotic manipulation remains the bottleneck for humanoid deployment, and that solving it requires purpose-built hardware rather than adapting industrial grippers. The firm previously backed Vicarious in 2013, a robotics AI company Alphabet acquired in 2018, giving it a decade of pattern recognition in the manipulation space.
What to Watch: Proception plans to release pilot program results from its unnamed logistics partner in Q4 2026, which will offer the first real-world validation of its tactile sensing claims. Tesla's next Optimus hardware revision is expected at the company's annual shareholder meeting in mid-2027, where engineers will scrutinize any improvements to hand dexterity. Shadow Robot, the UK-based leader in research-grade robotic hands, has signaled it will announce a lower-cost platform targeting commercial users sometime in 2027, potentially intensifying competition in a segment that has remained stubbornly expensive for decades.
