Lyte Raises $107 Million to Establish Foundation for Physical AI

In 2021, while most of the market was busy arguing about software eating the world, a different conversation was happening quietly in Mountain View. Alexander Shpunt, Arman Hajati, and Yuval Gerson were staring at robots that could move but could not really see. Not metaphorically. Literally. These machines were surrounded by cameras, depth sensors, and inertial units that refused to agree with each other. Perception was fragmented, brittle, and expensive, and Physical AI was paying the tax every single day.

This is the context behind Lyte stepping out of stealth with $107 million in total funding, announced January 5, 2026. Not a victory lap, more like a statement of intent. Backed by Fidelity Management & Research Company, Atreides Management led by Gavin Baker, Exor Ventures, Key1 Capital, VentureTech Alliance, and a private investor group led by Founding Investor and Chairman Avigdor Willenz, this is capital aligned around infrastructure, not theater.

Alexander Shpunt has lived this movie before. PrimeSense put depth sensing into the Microsoft Kinect, then Apple took it home and turned it into Face ID, shipping perception at consumer scale. Arman Hajati and Yuval Gerson were right there in the engine room, building the depth sensing stack inside Apple. After shipping perception to hundreds of millions of devices, they saw the same mistake repeating in robotics. Everyone was duct-taping vision together and calling it innovation.

Lyte exists to kill that tax. LyteVision fuses 4D sensing, RGB imaging, and inertial motion into a single spatial model, one interface, one perception brain. Hardware and software designed together, not negotiated after the fact. This is what robots need to operate in warehouses, factories, streets, and anywhere else reality refuses to behave. Autonomous mobile robots, robotic arms, quadrupeds, robotaxis, humanoids. Different bodies, same nervous system.

The market backdrop is loud for a reason. AI robotics is projected toward $125 billion by 2030, yet most industrial companies still lack the internal muscle to build safe, reliable perception stacks. Lyte is not selling a robot. Lyte is selling time, safety, and sanity to teams who would rather ship than calibrate sensors for months.

Avigdor Willenz does not fund science projects. Galileo, Annapurna Labs, Habana Labs followed the same pattern. Foundational tech, brutal execution, zero patience for fluff. That signal matters here, especially in a Physical AI cycle that will reward discipline long before it rewards headlines.

This funding is being used to deepen the platform, expand the team across sensing, silicon, and AI, and build relationships with robotics OEMs who understand that perception is destiny. When robots stop bumping into the world and start understanding it, the conversation gets quieter, because the people who saw it coming are already building.