Physical artificial intelligence chip infrastructure startup Embedd Ltd. says it’s ready to fix one of the most critical bottlenecks in the multibillion-dollar robotics industry after raising $2.7 million in a pre-seed round of funding.
Today’s round was led by Seedcamp and saw the participation of Cocoa, Connect Ventures, 2100 Ventures, Vesna Capital, U.ventures, Underline Ventures, Common Magic and Roosh Ventures.
Embed says almost $19 billion has been invested in so-called physical AI this year – things like autonomous cars, connected medical devices, humanoid robots and so on. Yet almost all of these intelligent machines have had to struggle with the same problem before they can bring a product to market. That problem is that the software that powers it needs a way to communicate with dozens of different semiconductors embedded into their hardware.
Unfortunately, none of these chips share a common language, which means that engineers are forced to manually write the code that allows them to communicate with each other and the underlying system software.
Embedd’s founders have experienced this problem first hand. The startup was founded by three Ukrainian technology entrepreneurs, Michael Lazarenko, Maxim Gorinov and Valentin Gololobov. They created Embedd after their previous hardware company faced chip shortages during the COVID-19 pandemic, and later the invasion of their country by Russia.
Because of these events, they ended up sourcing dozens of new components for their hardware, and found that they had to repeatedly rewrite software for each one. It was then that they realized they’d come across a major bottleneck that would only get worse as physical AI becomes more prevalent in the world.
To get around this bottleneck, the three founders created Embedd, which automates the creation of the code chips need to talk to one another. Instead of engineers reading through thousands of pages of documentation and hand-writing the necessary code, Embed spins up a digital twin of the hardware. Its AI agents then use this digital twin to obtain the required context so they can quickly generate the code needed by the chip to communicate with the software that runs on it.
“The next wave of AI will power factories, vehicles, robots and critical infrastructure, but today, every change in hardware creates huge complexity for software teams, and that friction is already slowing innovation,” Lazarenko said. “We built Embedd to address this.”
The London-based startup says its platform enables customers to ship production-ready software for semiconductors up to six times faster than was previously possible. It has already sold the idea to a number of chipmakers, including Microchip Technology Inc., helping them attract more developers to build on their silicon.
Rodger Richey, who serves as vice president of development systems at Microchip Technology, said the key issue for chipmakers is that they’re no longer differentiated so much by how fast their chips are, but how easy they are to build on. “Our work with Embedd is about meeting developers inside the software ecosystems they’ve already committed to, rather than asking them to come to ours,” he explained.
Lazarenko said Embedd will use the funds from today’s round to expand its platform’s capabilities and help more semiconductor makers to bring their products in emerging software ecosystems.