A robot’s intelligence doesn’t matter much if it can’t accurately sense the physical space around it. Lyte, a Sunnyvale-based startup building custom perception hardware and software for robots, raised $165 million in a Series C round on September 2, 2026, at a $1.6 billion valuation, putting it squarely inside the fast-growing physical-AI and robotics investment story.
Who actually built it
Lyte was founded in 2021 by Alexander Shpunt, Arman Hajati, and Yuval Gerson, all former Apple engineers who worked on the company’s advanced sensing and perception technologies. Shpunt’s background carries particular weight here: before Apple, he co-founded PrimeSense, the company whose 3D-sensing technology powered the original Microsoft Kinect before Apple acquired PrimeSense in 2013. That means Lyte’s founding team includes someone who has already taken a 3D-sensing technology from startup through acquisition by a major tech company once before, a specific, directly relevant track record rather than a general engineering background applied to a new problem.
What Lyte actually builds
Lyte develops the full perception stack for robots, from custom silicon and multimodal sensors through the spatial software that turns raw sensor data into usable information about where a robot is and what’s moving around it. The company says it owns that entire stack, from the transistor level through the final spatial data delivered to a robotic system, rather than assembling it from third-party components. That vertical integration is a meaningful technical bet: building custom silicon specifically for robotic perception, instead of adapting general-purpose sensors and chips built for other applications, is a slower and more expensive path, but one that can produce meaningfully better accuracy and reliability for the specific, demanding conditions robots actually operate in.
The funding numbers, and what’s changed since stealth
Lyte’s $165 million Series C, led by Maverick Silicon, brings the company’s total funding to $272 million. Fidelity Management and Research Co., which led Lyte’s Series B, returned to participate again in the Series C, alongside Atreides Management, Key1 Capital, Ora Global (formerly Exor Ventures), and additional new and existing investors. The company only emerged from stealth in January 2026, and in the roughly eight months since, it has moved into actual production, shipping to robotics customers across inspection, logistics, and manufacturing, a genuinely fast path from stealth to real production customers for a hardware-heavy company.
Lyte isn’t building a better camera for robots. It’s building the entire perception stack, from custom silicon through spatial software, on the bet that owning every layer produces meaningfully more reliable results than assembling one from off-the-shelf parts.
Why ‘physical AI’ is the category to watch here
Physical AI, broadly, refers to AI systems that have to perceive and act in the real physical world rather than only processing text, images, or data in a purely digital context. Robotics is the clearest current application, and investment in the category has accelerated sharply through 2026 as companies move from robotics pilots toward real deployment in logistics, manufacturing, and inspection work. Lyte’s specific position inside that category, perception infrastructure rather than a finished robot product, means its success depends less on any single robotics company’s own success and more on how broadly physical AI as a whole scales across industries, since a working perception stack is a component multiple different robot makers can adopt.
What separates this from other robotics investment stories
A meaningful share of robotics funding in 2026 has gone toward companies building complete robots for a specific application. Lyte’s bet is narrower and, in principle, more broadly applicable: rather than competing to build the best robot for one use case, it’s supplying the sensing and perception layer that any robot maker in inspection, logistics, or manufacturing could plausibly need. That positioning is closer to a picks-and-shovels strategy in a genuinely early, fast-moving category, valuable if physical AI adoption broadens across many different robot types and applications rather than consolidating around a small number of vertically integrated players who’d build their own perception stack in-house.
Frequently asked questions
Who founded Lyte?
Alexander Shpunt, Arman Hajati, and Yuval Gerson, all former Apple engineers who worked on the company’s sensing and perception technology. Shpunt previously co-founded PrimeSense, acquired by Apple in 2013.
How much has Lyte raised, and what’s it worth?
$272 million total, including a $165 million Series C in September 2026 at a $1.6 billion valuation.
What does Lyte actually build?
The full perception stack for robots, custom silicon, multimodal sensors, and spatial software, letting robots accurately sense their position and surroundings.
Is Lyte shipping real products yet?
Yes. Since emerging from stealth in January 2026, it has entered production and is shipping to robotics customers in inspection, logistics, and manufacturing.
For more on the AI and robotics funding landscape, see our Wonderful AI’s jump to a $5 billion valuation and the robots are finally leaving the factory floor. More coverage like this lives in our Robotics & Automation archive and the broader Tech & AI Evolution desk.
