- Axelera AI, a Dutch AI semiconductor firm, emerged from stealth today with $12 million in early investment led by Bitfury.
- The firm says that their device, which will be available to select customers in 2022, would consume a fraction of the power and cost of rival gear.
- The business has already assembled an exceptional team of more than 20 senior engineers and developers from world-class AI corporations and research institutions such as Intel, Qualcomm, IBM, and imec.
- Customers of Axelera will be able to use off-the-shelf technologies like Facebook’s PyTorch or Google’s TensorFlow machine learning frameworks.
Axelera AI, a Dutch AI semiconductor firm, emerged from stealth today with $12 million in early investment led by Bitfury, a Bitcoin blockchain security and infrastructure provider. The Innovation Industries Fund took part in the round, which the business claims would help with the development of AI apps at the edge.
Edge AI is the processing of AI algorithms on devices such as drones, digital signs, cameras, and sensors. Edge computing is based on the same premise: data is processed and handled at the Internet of Things endpoints. Edge AI is useful in businesses such as retail, robotics, and manufacturing since it can provide lower latency than transmitting data to faraway computers while avoiding interruptions and expenses.
Axelera, a Bitfury-incubated startup, claims to be working on a chipset to speed AI and machine learning algorithms at the edge. The firm says that their device, which will be available to select customers in 2022, would consume a fraction of the power and cost of rival gear.
Bitfury Group incubated Axelera AI in 2019, and the firm previously operated under Bitfury AI until it became an independent entity this year. Axelera AI joined up with imec in early 2020 to develop breakthrough computing architecture for high-performance AI, building on its substantial research and development focused on combining world-class hardware and software technologies for edge and cloud-to-edge corporate solutions.
Axelera AI is headquartered at the AI Innovation Center of the High Tech Campus in Eindhoven, with R&D offices in Leuven (BE) and Zurich (CH). The business has already assembled an exceptional team of more than 20 senior engineers and developers from world-class AI corporations and research institutions such as Intel, Qualcomm, IBM, and imec.
“The market for AI at the edge is still in its early stages, with Nvidia serving as the current industry leader. New businesses such as Hailo, Mythic, and Blaize have just launched new products,” stated CEO and cofounder Fabrizio Del Maffeo, former head of AI at the Bitfury Group. Fabrizio went on to say that their aim is to create a ‘green’ (low power consumption) hardware and software platform that allows the sector to fully leverage the benefits of AI.
The edge AI hardware market is expected to expand from 920 million units in 2021 to 2.08 billion units by 2026, representing a significant increase. According to one estimate, the market for AI processors alone would be worth $73.49 billion by 2025.
According to Del Maffeo, Axelera’s technology is distinguished by a unique dataflow architecture with multicore in-memory processing. In-memory computing, also known as in-memory computation, is the technique of executing computer calculations entirely in computer memory, resulting in a great performance at minimal power consumption.
Benefits of using Axelera AI
The chip’s usability will be a priority. Customers of Axelera will be able to use off-the-shelf technologies like Facebook’s PyTorch or Google’s TensorFlow machine learning frameworks, eliminating the requirement for knowledge in emerging AI development environments.
Axelera will also deliver software tools that use performance-enhancing methods such as graph optimizations and quantization, which compresses AI models with relatively minor accuracy losses.
Axelera AI is developing the world’s most efficient and sophisticated edge AI solutions. Its game-changing hardware and software solution will concentrate a whole server’s AI computing capability into a single chip at a fraction of the power consumption and cost of AI hardware today.
The company’s ground-breaking technology will enable hundreds of edge AI applications, making AI more efficient and accessible than ever before. When Axelera AI releases to select clients and partners in early 2022, its solutions will be completely integrated with the top open-source AI frameworks.
“Recent advancements in deep learning have changed the way computer devices interpret human-centric material including pictures, video, voice, and audio,” said Del Maffeo. Del went on to say that incorporating Axelera’s technology into IoT devices may pave the way for a new generation of apps capable of sophisticated sensing and identification tasks.
While Axelera is still in its early stages, the 30-person business has collaborated with IMEC, a nanoelectronics research and development group, to create its chip design. (IMEC is also a shareholder.) According to Del Maffeo, Axelera is in talks with around 20 clients ahead of its anticipated delivery date.