DMR News

Advancing Digital Conversations

Axelera secures $68M to challenge Nvidia with innovative edge AI chips

ByYasmeeta Oon

Jul 1, 2024

Axelera secures $68M to challenge Nvidia with innovative edge AI chips

Nvidia has been leading the charge in the age of AI hardware, with its high-performance GPUs being employed by some of the world’s largest technology companies to power training and inference for massive AI models. As Nvidia, led by CEO Jensen Huang, continues to dominate the market, a new wave of smaller AI hardware companies has emerged, targeting specific niches within the domain. One such startup, Axelera AI from the Netherlands, has recently announced a significant milestone in its journey.

Axelera AI, a company developing AI processing units (AIPUs) for computer vision inference workloads on the edge, has secured $68 million in Series B funding. This investment marks the largest Series B round in Europe’s fabless semiconductor category. The funding round was led by notable institutional backers, including Invest-NL Deep Tech Fund, European Innovation Council Fund, Innovation Industries Strategic Partners Fund, and Samsung Catalyst Fund.

The company plans to utilize this capital to expand its existing AIPU solutions into new markets and geographies. Additionally, Axelera AI aims to develop new products to address the computing needs of next-generation AI workloads, including multimodal large language models (LLMs).

Axelera AI, spun out of the AI innovation lab of Bitfury Group, a blockchain technology company, focuses on AI acceleration for edge computing applications. Launched in 2021, Axelera has developed a platform called Metis, which combines hardware and software to handle computer vision inference at the edge. The platform strikes a balance between performance, efficiency, and ease of use.

12nm CMOS AI Processing UnitCentral processing unit designed for AI applications, featuring four self-sufficient AI cores.
Software Development Kit (SDK)Toolset for building computer vision applications capable of running on devices using the chip.

Each AIPU includes four AI cores that can either collaborate on a workload to boost throughput or process different neural networks concurrently. The AI core is a RISC-V-controlled dataflow engine delivering up to 53.5 TOPS (trillions of operations per second) of AI processing power. This engine features multiple high-throughput data paths to ensure balanced performance across a wide range of layers, addressing the heterogeneous nature of modern neural network workloads. The total throughput of Axelera’s four-core Metis AIPU can reach 214 TOPS at a compute density of 6.65 TOPS/mm².

Axelera’s AI cores use proprietary digital in-memory computing technology to accelerate matrix operations, offering high energy efficiency at 15 TOPS/W. This technology represents a radically different approach to data processing, where crossbar arrays of memory devices store a matrix and perform matrix-vector multiplications “in place” without intermediate data movement.

  • RISC-V-Controlled Dataflow Engine: Ensures high AI processing power.
  • Digital In-Memory Computing: Enhances energy efficiency and computational throughput.
  • Versatile AI Cores: Capable of collaborative or concurrent neural network processing.

Axelera AI is currently shipping Metis evaluation kits to enterprises like Fogsphere, XXII, and System Electronics. With the latest funding, which brings the total capital raised to $120 million, Axelera plans to take its platform into full production for widespread deliveries. The company claims its product can deliver up to a five-fold increase in efficiency and performance for enterprises and is witnessing significant demand, with a strong business pipeline of over $100 million.

As part of its growth strategy, Axelera plans to expand into North America, Europe, and the Middle East, focusing on key verticals such as automotive, digital healthcare, Industry 4.0, retail, robots and drones, and surveillance. While Metis will be the starting point in these markets, Axelera also plans to develop price-competitive, data center-focused accelerators to meet the growing computing needs for generative AI models, including large multimodal models.

  • Target Markets: North America, Europe, and the Middle East
  • Key Vertical Focus: Automotive, digital healthcare, Industry 4.0, retail, robots & drones, surveillance
  • Future Products: Data center-focused accelerators

“There’s no denying that the AI industry has the potential to transform a multitude of sectors. However, to truly harness the value of AI, organizations need a solution that delivers high performance and efficiency while balancing costs,” said Fabrizio Del Maffeo, co-founder and CEO of Axelera AI, in a statement. “This funding supports our mission to democratize access to artificial intelligence, from the edge to the cloud. By expanding our product lines beyond the edge computing market, we can address industry challenges in AI inference and support current and future AI processing needs with our scalable, proven technology.”

Axelera AI is not alone in the edge AI acceleration space. Another notable player, Hailo, recently announced the Hailo-10, an energy-efficient processor designed to deploy generative AI applications across edge devices like cars and commercial robots. This competition highlights the growing demand for efficient AI processing solutions at the edge, driven by the increasing adoption of AI across various industries.

In conclusion, while Nvidia continues to lead the AI hardware market with its powerful GPUs, the emergence of startups like Axelera AI demonstrates the dynamic and evolving nature of the industry. With significant funding and innovative technology, Axelera AI is well-positioned to make a substantial impact on the edge AI acceleration market, offering scalable solutions that cater to the next generation of AI workloads.

Featured Image courtesy of IMEC

Yasmeeta Oon

Just a girl trying to break into the world of journalism, constantly on the hunt for the next big story to share.

Leave a Reply

Your email address will not be published. Required fields are marked *