Zurich-based startup Synthara has today raised 11m, to develop its technology to make embedded chips — the type of smaller processors that power compact devices like wearables and sensors — more capable for AI-powered applications.
The round was led by Vsquared Ventures, with other new backers OTB Ventures, Onsight Ventures, and Deep Tech Labs participating alongside existing investors. $5.9m of the total investment also came in the form of European and Swiss grant funding.
Cofounder and CEO Manu Nair says he and his team started Synthara to address a growing need for more sophisticated capabilities in smaller devices, as the chips that power them have struggled to keep pace with new, AI-powered use cases.
“These guys [embedded chips] are the workhorse of pretty much any portable device,” he tells Sifted. “With AI that got massively disrupted — the embedded chips can’t do everything anymore.”
Nair gives the example of a hearing aid application that could amplify the voice of the person you’re looking at.
“These things are really AI workloads, a traditional hearing device that uses conventional embedded chips could never do that,” he says.
Nair says this has led the device makers — companies that build anything from factory sensors to AirPods and hearing aids — to experiment with costly and complicated new hardware and software solutions. Synthara says its product, dubbed ComputeRAM, allows these device makers to use the same style of embedded chips they’ve become used to while allowing them to perform a broader range of applications in the new era of rapid AI adoption.
“We enable an embedded chip that used to basically be only applicable for a very narrow set of use cases to be broadly usable,” he says.
5x a week
Sifted Daily
Stay one step ahead with news and experts analysis on what’s happening across startup Europe.
By Sifted journalists
How does it work
Rather than getting into making these small chips for portable devices itself, Synthara helps device makers and chip makers by designing the architectures for processors and then licensing out that intellectual property.
It says it can get better performance and more use cases out of embedded chips by enabling something called “in-memory computing”. Traditional chips have separated computer memory and computer processing which means they have to send data from the memory (eg. a hard drive) to the processor.
Synthara’s software designs allow embedded chips to carry out computing within the memory unit, which it says can make these processors “100x better, both in terms of energy efficiency and latency”.
The first major device maker that it’s working with and can talk publicly about is Bosch. Nair says Synthara is working with the German multinational engineering company on a variety of use cases, ranging from accelerometers in sensors to microphones for voice-controlled computing.
What’s next?
Synthara, a small startup of 20 people spun out of ETH Zurich and the University of Zurich’s Institute of Neuroinformatics in 2019, is very much not the only company working on improving chips for portable and compact devices.
Other startups in the space include Netherlands-based Innatera and UK-based Red Semiconductor, which are both developing their own chips for similar applications. Nair says Synthara is in a strong position because, by designing chip architectures rather than the chips themselves, all of these companies are potential clients.
“This is a very important problem and there are many people attacking it from all directions,” he says. “The strength we have is we are most of the time complimentary to all of these providers.”
With the new fundraise, Synthara plans to grow its team to 25 people. It will be focusing on a go-to-market strategy that involves getting its product working smoothly with a small handful of clients, before it plans to raise again in 2025.
Read the orginal article: https://sifted.eu/articles/synthara-raise-ai-chip-news/
