The race for semiconductor leadership is on, and European semiconductor start-ups are rising to the challenge.
Europe is already home to well-known semiconductor companies such as ASML, NXP, Arm and Infineon. But it lags behind in manufacturing capacity. The EU currently produces around 10% of the world’s semiconductors. The UK was projected to account for 0.5% of global semiconductor sales in 2023.
In the ongoing battle against semiconductor giants from Taiwan, China and the US, the EU and UK are focusing on leveraging their respective strengths in research and development and chip design to gain a competitive advantage in the next generation of semiconductors, an area where the mainland’s start-ups can play a vital role.
New Chip Designs for AI Inference
Last week, London-based Fractile emerged from stealth with a £15 million seed round of funding for a “game-changing” chip design that promises to improve AI inference, where a trained AI model uses learned knowledge to analyze live incoming data and generate results.
Fractile argues that existing chip hardware has significant limitations, including reduced model performance, increased costs, and reduced adoption.
To address this, the startup has developed a new computing architecture with circuitry that it says can perform 99.99% of the operations needed to perform inference.
A key aspect of this technology is in-memory computing, which means that instead of constantly moving data between memory and processing units, calculations are performed within the memory itself.
Fractile says its chips will make AI inference 100 times faster, 10 times cheaper and use less power than current solutions. If successful, the company aims to offer an alternative to U.S.-based Nvidia, the leading supplier of advanced AI chips.
The funding round also included support from the NATO Innovation Fund.
Gaining an Edge AI Advantage
Another potential challenger to Nvidia is Axelera AI. Founded in 2021, the Netherlands-based company develops chips for computer vision and edge AI, which refers to the deployment of AI algorithms directly on local edge devices such as sensors.
Axelera specializes in proprietary digital in-memory computing and AI inference solutions based on RISC-V technology, an open-source instruction set architecture for designing computer processors.
The startup claims that its Metis platform, which combines both hardware and software, will improve efficiency and performance by three to five times.
In June, Axelera raised $68 million in new funding, bringing its total to $120 million, including investment from the European Innovation Council Fund.
The company plans to use the new funding to expand its product portfolio and target applications in verticals such as data centers, robotics, healthcare, and generative AI applications. Axelera is also looking to expand into North America and the Middle East.
Graphene-based chips
Meanwhile, in Germany, Brac Semiconductor is developing a new type of chip connection technology that uses graphene, the “miracle material” discovered in 2010 by Andre Geim and Konstantin Novoselov.
The startup claims that its graphene-based chip technology will address current shortcomings of traditional silicon chips, such as low processing power and high energy consumption.
Black Semiconductor’s technology and hardware enable optical connections between chips, speeding up data communication between them and allowing them to interact as a near-integrated system. The company also expects to reduce manufacturing costs by requiring 60% fewer steps.
In particular, such chips could be transformative for technologies such as GenAI, data centers, and autonomous vehicles.
Brac Semiconductor plans to set up a pilot line in Aachen to produce 300 million chips by 2026. If successful, it will be the first time in the semiconductor industry that graphene will be integrated into chips on a commercial scale.
Not surprisingly, the company recently raised €254.4 million, the largest amount of funding ever raised by a European semiconductor startup, with the German government providing the bulk of that amount through the EU’s state aid scheme, aimed at supporting the EU’s digital transformation.
