CEO Interview: XMOS on the UK semiconductor strategy
Cette publication existe aussi en Français
XMOS is exactly the kind of company that you would expect the UK’s semiconductor strategy to be aiming to support. A UK-headquartered fabless chip company with world-leading AI technology, making microcontrollers for the consumer market in volume with customers around the world. However that is not the case.
“XMOS is at a cross point of being a scaleup semiconductor company rather than a startup and also fabless,” says Mark Lippett, the CEO.
He has been lobbying for support for the semiconductor industry in the UK for many years to boost the prospects of fabless companies like his, which haven’t been mentioned in the long awaited strategy. Late last year, the company added Jalal Bagherli as a special advisor, and Bagherli is heading up the Semiconductor Advisory Panel launched today.
“The compound semiconductor side is great and I agree in principle that we need to be fostering university spinouts but there has to be somewhere for them to go,” Lippett tells eeNews Europe.
“The problem with the ‘strategy’ in the past is the lack of a rising tide rising the industry and clusters of talent. Without supporting the later stage companies to remain part of the UK ecosystem and remain in the UK, there is nowhere for them to go. Ownership is not necessarily the key, it’s the critical mass,” he said.
XMOS, founded by the inventor of the Transputer, David May, and one of his students in 2005, now has about 70 people worldwide and serves about 200 customers in volume.
“We did $24m last year, that’s public information, and are profitable, but most of our revenue is in consumer so 2023 is challenging but we are looking for a good 2024. The thesis of the technology is to build a profitable semiconductor business around a fragmenting IoT market. It’s a dynamic sum of niches and the only way to address that is with a programmable platform and that’s what XMOS has got. We are still private, but looking for late stage scale up investment and that’s a very small community of investors.”
“At the moment I feel the company has moved on and is an inflexion point in many different areas and now is the time for that scaleup investment and I have realised quite how challenging it is.”
“We have been beavering away in the background,” he said. “The fourth generation architecture is a RISC-V variant and we have retained the flexibility of our proprietary instruction set, and we benefit from the RISC-V ecosystem already for more mature tools but there’s no way we can deliver the same horsepower in tools as the RISC-V ecosystems.”
The move to RISC-V has opened up access to a wide range of new customers, he says.
“Now that we can continue the conversation beyond proprietary instruction sets with customers we can have wider conversations on using RISC-V. From a tools perspective is absolutely the right thing to do it in an open source capacity and we are already pushing changes back and our intention is to get more involved,” he said. “There are elements of the architecture we are looking at opening up so we are serious about open source and being a significant contributor.”
Fourth generation microcontroller sampling with RISC-V
“We are still on track for samples of the new chip by the end of the year. Xcore is a very clever uncommitted platform and the special sauce is in software.”
“Cost is the most important and the second factor is performance as we emulate hardware in software. Third is power and the next generation will be significantly lower on power, along with architectural changes and we are starting to see the results.
“We are seeing positive signs for 2024. We are mostly in consumer and professional electronics which has had a rough ride but we are seeing the light at the end of the tunnel as well as opportunities in other industries. We are known for audio and now audio even detection with AI , noise reduction, our platforms are being used in those applications with different AI models and we are also doing image AI.”
He is positive but cautious about the semiconductor strategy.
“We do a lot of thinks right in the UK. We have great engineering talent, probably not capturing enough of the talent but I am constantly impressed by the teams we can put together in the UK,” he said.
“There’s a couple of things that feed of each other. We have to take more risks and celebrate success and be prepared for failures along the way and that might involve government money, that’s just par for the course. And we have to put some money behind it from the capital markets and the government to place some of those bets. In the end, the economy will win, the ecosystem grows and the tide will rise and positive feedback occurs.
“We have a starting point and that’s good. The sum of money is small but the UK government is having to prime the pump in a way that other governments are not. I hope $1bn is the tip of the iceberg and that will catalyse a more detailed funding.”
