Several years ago, Fabrizio del Maffeo and the core team of the Belgian nanotechnology laboratory Emec collaborated with Evangelos Eleftheriou and a team of researchers from the IBM Zurich Lab to develop the computer chip. Unlike conventional chips, theirs was for peripheral devices, especially those running AI workloads, because Del Maffeo and the rest of the team realized that offline computing hardware was inefficient and expensive.
After creating the startup – Axelera AI – they have secured their chip technology from blockchain company Bitfury Group, Dell Maffeo and Group Assurance from VCs including Imex’s venture arm, Imex.xpand. Innovation Industries led a $27 million Series A in Accelera AI that closed this week with participation from Imec.xpand and Belgium’s Federal Holding and Investment Company. In addition, the Netherlands Enterprise Agency has provided a $6.7 million loan to Accelera AI from the Ministry of Economic Affairs and Climate Policy.
One of the challenges we are facing is the availability and accessibility of AI to different groups and industries. Some businesses will benefit from AI, but lack the expertise or technical know-how to integrate the benefits into day-to-day operations, Del Maffeo told TechCrunch in an email interview. “We are building an AI platform to overcome this barrier to access… [by] Game-changing, user-friendly and scalable technology to accelerate computer vision and natural language processing at the edge, delivering greater performance and efficiency at a lower cost to existing players.
Axelera is working to develop AI accelerator cards and systems for applications such as security, retail and robotics, which it plans to sell through partners in the business-to-business edge computing and Internet of Things sectors. The cards and systems include Axelera’s Thetis Core chip, which enables in-memory computing for AI calculations — running “in-memory” in RAM to reduce latency introduced by memory storage devices.
Accelera is developing software to manage the chip, which Del Maffeo says will “fully integrate” with major AI frameworks — such as PyTorch and TensorFlow — when it’s available.
Axelera test chip to accelerate AI and machine learning workloads. Image Credits: For Axel
“We’re democratizing access to AI,” Del Maffeo said. “When our product is launched… our aim is. [deliver] A chip that powers the entire AI server.
Accelera has a ways to go before it gets to business. The company only produced its first test bench chip last December, and doesn’t expect Alexa to start shipping to customers until sometime in the first half of 2023.
It’s also not the first company to pursue memory for peripheral devices. Neuroblade, which raised $83 million in capital last October, is developing chips that integrate both computers and memory into a single hardware block for data processing. MemVerge, GigaSpaces, Hazelcast and H20.ai also provide memory solutions for AI, data analytics and machine learning applications.
But despite the fact that Axelera is moving forward with both pre-market and pre-revenue and venture debt rounds, Del Maffeo believes that the company is in a good position to have a foothold in the market for custom AI chips. Jonathan Ballon, former VP and general manager of Intel’s Edge AI and Internet of Things group, is joining Accelera as chairman. And Del Maffeo points out that Accelera continues to aggressively hire about 85 employees in Europe, both remotely and at the company’s offices in Eindhoven and Milan and R&D centers in Leuven and Zurich.
Following expansion in the US and Taiwan in the coming months, Del Maffeo Accelera expects to hire 130 to 140 employees by early 2023.
Del Maffeo said: “While the pandemic has caused a number of shortages in the chip industry, we are fortunate to have seen significant growth since our launch in 2021.” “While we did not disclose our burn rate, we can share that we are well positioned to raise a new round of investment in 2023, which will be significantly larger than our Series A, and we are receiving interest, including from several US investors, which will help us bring the company to the next level… We are carefully building our customer base and partner ecosystem through our portfolio of companies. Show interest in AI platform. Later this year, we will open a unique collaboration opportunity for companies to become early adopters of our AI platform.
If we think Axelera can fulfill his promise, he stands to make some serious money. The edge AI hardware market is predicted to grow from 920 million units in 2021 to 2.08 billion units in 2026 – a lucrative growth. According to one estimate, the AI chips market alone is set to reach $73.49 billion by 2025.
