Analysis
Lyte, a physical AI startup building custom sensing silicon for robots, raised a Series C led by Maverick Silicon, Crunchbase News reported. Fidelity Management and Research, Atreides Management, Key1 Capital and Ora Global (formerly Exor Ventures) also participated, alongside existing investors.
- Series C -- $165 million, led by Maverick Silicon
- New valuation -- $1.6 billion
- Total funding to date -- $272 million
From PrimeSense and Kinect to Warehouse Robots
Founded in 2021 in Sunnyvale, California by Alexander Shpunt, Arman Hajati and Yuval Gerson -- all former Apple engineers -- Lyte builds what it calls physical AI perception systems: custom silicon combined with 4D sensing, RGB imaging and motion-awareness software that lets robots understand and navigate their physical environment in real time. Shpunt's background gives the company an unusually direct technical lineage: he previously co-founded PrimeSense, the Israeli company whose 3D-depth-sensing technology powered the original Microsoft Kinect, before Apple acquired PrimeSense in 2013 and built its technology into Face ID. That's roughly fifteen years of the same underlying sensing problem -- getting a machine to understand physical space in real time -- moving from a gaming peripheral, to a phone's unlock mechanism, to a warehouse robot's navigation stack.
Lyte's current customers operate primarily in warehousing and manufacturing, the two industrial categories where robot deployment has the clearest, fastest-to-prove return on investment: repetitive, physically demanding, labor-constrained environments where a robot that can reliably sense its surroundings replaces a role that's already hard to staff. The company emerged from stealth earlier in 2026 after quietly raising $107 million across its Series A and B combined -- meaning this Series C alone is roughly 1.5 times everything the company had raised before it, a meaningful acceleration in capital intensity.
The Physical AI Hardware Field
Lyte's perception-and-sensing niche sits adjacent to, but distinct from, automotive lidar players like Luminar and Ouster, whose sensors solve a related but differently specced problem for self-driving cars rather than indoor robots. Inside the warehouse and manufacturing robotics category more directly, Lyte competes conceptually with the perception layers built into general-purpose robotics platforms from companies like Figure and Physical Intelligence, though those companies focus more on the robot itself and less on selling perception hardware as a standalone component other robot makers can integrate. Nvidia's Isaac platform is the closest thing to an incumbent default for robot perception software, which makes Lyte's custom-silicon approach a bet that purpose-built hardware beats a general-purpose software stack running on off-the-shelf sensors, at least for now.
What a 1.5x Capital Jump Signals
Going from $107 million total to a single $165 million round is the kind of acceleration that shows up when a company's unit economics start working at customer scale rather than pilot scale -- investors don't write checks that size on unproven ROI. The unresolved question, typical of hardware-heavy AI companies, is manufacturing yield and cost curve: Lyte's custom silicon has to get cheaper per unit as it scales, the same challenge that has slowed multiple photonics and sensor startups even after strong early rounds. The new capital is earmarked for scaling both hardware production and the software stack, which is an honest admission that both sides of the stack still need more investment before this becomes a category leader rather than a well-funded challenger.
