Latest edition: 26 August 2026London — published continuously since 2026Free forever
The Founder Gazette
Startup news, held to newspaper standards
Startup news

Embedd raises €2.3m to write chip drivers with AI agents

The London startup sells semiconductor firms an automated route into software ecosystems. Seedcamp led the pre-seed; the founders started it after chip shortages wrecked their last company.

By The Gazette desk26 August 2026144

Embedd, based in London, has raised €2.3 million ($2.7 million) in pre-seed funding. Seedcamp led the round.

Cocoa, Connect Ventures, 2100 Ventures, Vesna Capital, U.ventures, Underline Ventures, Common Magic and Roosh Ventures also took part.

The money goes on the platform and on signing more semiconductor companies, which are the customers rather than the end users.

The problem Embedd sells against is unglamorous and well known to anyone who has shipped hardware. Every chip needs software underneath it, and no two chips speak the same language.

Traditionally an engineer reads the documentation — thousands of pages of it — and hand-writes the integration code. Embedd says it builds a digital twin of the hardware instead, gives AI agents that context, and generates the code that makes the chip usable by the software above.

The company reports that customers have reached production-ready software for chips up to six times faster. That figure is Embedd's own; it has not published the underlying comparison.

Commercial launch was in April 2026. Since then the company says it has signed contracts with multiple semiconductor companies.

One is named: Microchip Technology, where Embedd is enabling support for Zephyr, the open-source real-time operating system. The other customers have not been named, and contract values have not been disclosed.

The founders are Ukrainian entrepreneurs Michael Lazarenko, Maxim Gorinov and Valentin Gololobov. Their previous hardware company was hit by chip shortages during COVID and later disrupted by Russia's invasion of Ukraine.

Each time they sourced a new component, they had to rewrite software for it. That is the bottleneck they are now selling a way around, and they argue it gets worse as AI moves into machines that physically do things.

Lazarenko frames the pitch around who bears the cost. If a chipmaker's device is hard to build on, developers pick a rival's part instead, so the integration layer is a commercial problem for the silicon vendor rather than a chore for its customers.

For founders building anything with hardware in it, the useful signal is where the money went. A pre-seed round of this size, led by Seedcamp, is a bet that semiconductor companies will pay to have their own driver work done for them.

Whether that bet holds depends on renewals rather than launches, and Embedd is four months into selling. The next test is how many of those unnamed contracts turn into repeat business.

The promise of physical AI is enormous, but today's hardware fragmentation is slowing innovation. This funding enables us to expand our platform and help more semiconductor companies bring their devices into emerging software ecosystems
Michael Lazarenko, co-founder and CEO, Embedd