Seedcamp leads $2.7M round for Ukrainian-founded physical AI startup Embedd
Tech Funding News Abhinaya Prabhu ● Covered by 2 sources
Seedcamp led a $2.7M round for Embedd, a Ukrainian-founded startup that helps robots and chips talk faster. It says setup work that used to take months can now take about three weeks.
Based on reporting by Tech Funding News, Abhinaya Prabhu — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Embedd has raised $2.7 million in pre-seed funding, with Seedcamp leading the round and Cocoa, Connect Ventures, 2100 Ventures, Vesna Capital, U.ventures, Underline Ventures, Common Magic and Roosh Ventures joining in. The London-based company was founded by Ukrainian entrepreneurs and is building software infrastructure for robots, cars, drones and medical devices.
The pitch is simple enough to understand even if the underlying mess is not: today, engineers can spend four to six months writing the glue code that lets dozens of chips inside a machine work together. Embedd says its platform turns that slog into about three weeks. It also says it can produce production-ready software for a chip up to six times faster than doing it by hand.
That matters because the money pouring into physical AI has mostly gone into the visible stuff — robots, foundation models, the hardware headline. Nearly $19 billion has gone into robotics and physical AI startups this year, including $16.3 billion in the first quarter alone. Embedd is betting on the less glamorous layer underneath, the one that decides whether all that hardware is actually usable.
The company’s founders — Michael Lazarenko, Maxim Gorinov and Valentin Gololobov — had already run a hardware business together. Then came pandemic chip shortages, repeated component swaps and firmware rewrites, and later the loss of their production facility after Russia’s invasion of Ukraine. That cycle convinced them the integration problem was only getting worse as AI moved into physical machines.
Embedd works by turning chip documentation into a digital twin instead of forcing engineers to read datasheets line by line. It launched commercially in April 2026 and has already signed several semiconductor customers, including Microchip Technology, where it is helping support the Zephyr real-time operating system. Microchip’s Rodger Richey says the competition in embedded is shifting from fast silicon to silicon that is easier to build on.
It is not the only company chasing this layer. Bengaluru-based H2LooP raised $2 million earlier this year for a related approach. Embedd says it will use the new money to expand its platform and deepen semiconductor partnerships, which is basically the polite way of saying the real fight is just starting.
My take — AI-written commentary, not fact-checked reporting
This is the unglamorous part of AI that actually matters: less demo theatre, more wiring and software plumbing. The market loves robots until someone has to make three chips speak the same dialect. Seedcamp backing this says a lot — the next embedded war may be won by whoever makes integration boring.
Read more about this at: Tech Funding News