Axelera Secures $68 Million Funding to Compete with Nvidia in Edge AI Chip Market

Nvidia's Pioneering Role in AI Hardware

Nvidia is at the forefront of the AI hardware revolution, with its high-performance GPUs supporting the training and inference of large-scale models by leading tech companies worldwide.

Amidst Nvidia's success, emerging startups are targeting specific niches within the AI hardware market. One standout is Axelera AI, a Netherlands-based startup that recently secured $68 million in Series B funding, marking Europe’s largest Series B round in the fabless semiconductor sector.

Innovative Edge Computing Solutions

Axelera AI specializes in AI acceleration through its AI Processing Units (AIPUs), designed for computer vision inference workloads on the edge. The funding will enable the company to expand its AIPU solutions into new markets and develop products tailored for next-generation AI workloads, including multimodal large language models (LLMs).

Axelera's Unique Value Proposition

Originating from the Bitfury Group's AI innovation lab, Axelera AI focuses on advancing AI acceleration, particularly in edge computing. Since its launch in 2021, the company has introduced Metis, a platform that combines hardware and software to optimize computer vision inference, striking a balance between performance, efficiency, and user-friendliness.

The Metis platform features two key components: a 12nm CMOS AI Processing Unit (AIPU) and a user-friendly software development kit for building computer vision applications. Each AIPU includes four autonomous AI cores capable of either collaborating to enhance throughput or simultaneously processing multiple neural networks.

“The AI core operates as a RISC-V-controlled dataflow engine, achieving up to 53.5 TOPS of AI processing power with high-throughput data paths. This design caters to the diverse needs of modern neural networks,” according to Axelera’s website.

Axelera's proprietary digital in-memory computing technology accelerates matrix operations, achieving impressive energy efficiency of 15 TOPS/W. This innovative data processing approach utilizes memory devices arranged in crossbar arrays to perform matrix-vector multiplications without intermediate data movement, enhancing operational efficiency significantly.

Growth Aspirations and Next-Gen AI Workloads

Axelera is actively shipping Metis evaluation kits to businesses like Fogsphere, XXII, and System Electronics. With the new funding, bringing its total capital to $120 million, Axelera aims to ramp up production and serve a burgeoning demand, backed by a robust sales pipeline exceeding $100 million.

Co-founder and CEO, Fabrizio Del Maffeo, emphasizes the company’s growth strategy, which includes expanding to North America, Europe, and the Middle East. Targeting sectors like automotive, digital healthcare, Industry 4.0, retail, robotics, and surveillance, Axelera plans to leverage Metis as a launchpad for further product development, including competitive data center accelerators to meet the demands of generative AI models.

“The AI industry holds transformative potential across sectors. To unlock this value, organizations require efficient, high-performance solutions at a reasonable cost,” Del Maffeo stated.

“This funding enhances our mission to democratize AI access from edge to cloud. By diversifying our product offerings, we aim to tackle industry challenges in AI inference and support both current and future AI processing needs with our proven technologies.”

In parallel, Hailo is another notable contender in the edge AI acceleration field, recently introducing the Hailo-10, an energy-efficient processor for deploying generative AI applications in edge devices like vehicles and commercial robots.