This is a partner publication, brought to you by AWS.
By 2030, the digital landscape will be swarming with over 125 billion Internet of Things (IoT) devices, from smartphones to cameras and smart home gadgets. But here’s the deal: these devices will generate a mind-boggling amount of data, with 80% of it coming from images and videos. The problem is that we don’t have the tools or hardware to make sense of it all. That’s where artificial intelligence (AI) comes in – but to harness its full power, we need a serious new generation of computational hardware.
As privacy, security, and bandwidth issues skyrocket, we’re moving towards processing data closer to its source, the so-called network “edge”. Integrating AI at this level is a monumental challenge, mainly because AI technology is built for cloud computing operations, which don’t have the same cost, power, and scalability constraints as edge devices. But there’s a catch: these data centres are often miles away from the action. This distance results Into extra latency and a solid punch of risk to your applications. Not to forget, they use huge amounts of power.
Enter Axelera AI. The Eindhoven-based startup, only founded in 2021, is seizing the opportunity to create a hardware and software platform that accelerates AI in edge devices. Their Metis AI Platform, powered by its disruptive in-memory computing and RISC-V controlled dataflow technology, offers industry-leading performance and usability, all while using a fraction of the cost and power consumption of today’s solutions.
Axelera AI has integrated their technology into AI acceleration cards, boards, and vision-ready systems, opening the door for small and medium-sized businesses to speed up AI adoption. Their Voyager Software Development Kit allows developers to effortlessly incorporate AI into their devices, with user-friendly neural networks for computer vision and future natural language processing applications.
Ioannis Koltsidas, VP of AI Software at Axelera AI, elaborates on their work with AWS. “At Axelera AI, we’re using AWS as our go-to development infrastructure for building our software stack and managing Machine Learning Operations (MLOps), such as training and deploying cutting-edge computer vision neural networks,” Koltsidas said. “We’re also crafting a customer-centric Cloud Platform that serves as a one-stop-shop for AI at the Edge, letting customers grab AI models from partners and the community, develop end-to-end computer vision applications, and deploy them to Axelera AI-enabled edge devices. We employ a range of AWS services, including EC2, S3, Lambda, Cognito, Amplify, RDS, EKS, and Sagemaker which has enabled us to increase the speed at which we roll out new features by more than 60%.”
Koltsidas went on to share how AWS has been their trusted partner from the start, helping them develop a state-of-the-art software stack for computer vision at the edge. “With AWS we can focus on what is important for our customers without worrying about the development infrastructure. As we build our Cloud Platform, AWS will play a key role in realizing our vision to bring the simplicity and User Experience of the Cloud to Edge AI and to our customers.”
“Scaling compute infrastructure to keep up with our growing needs was a significant challenge when we first started,” Koltsidas continued. “But AWS let us scale fast and adjust our infrastructure on the fly. The elasticity and flexibility offered by AWS services have let us run large-scale machine learning experiments for tons of neural networks and datasets, continuously optimizing the accuracy and performance of ML workloads on our AI platform. Our Machine Learning engineers can get from an idea to code that is tested across tens of neural networks and datasets in a matter of hours as opposed to days with alternative approaches.”
Axelera AI is revolutionizing the world of AI by tackling the challenges associated with IoT devices and edge computing head-on. Team Axelera’s rapid growth, now tracking at 125 colleagues in less than two years, boasting over 45 PhD holders and a presence in 15 countries, demonstrates their commitment to innovation and excellence. An early access program is available for a select group of customers, delivering evaluation kits in July 2023, making their cutting-edge products and technology more accessible.
As we approach a future with an ever-increasing number of connected devices, the work of trailblazers like Axelera AI will play a vital role in unlocking the true potential of artificial intelligence and transforming the way we live, work, and interact with technology.
About the author
Tara Advaney, AWS Startup Team, helps AI & health-care life science startups minimise their cost and time to market. Before this, Tara worked in the health-care team at Meta growing early-stage startups, and has a background in Science and Technology from Georgetown University.