AMD is buying Finnish AI company Silo AI for $665 million. The chip manufacturer aims to strengthen its position in artificial intelligence and offer customized AI solutions for businesses.
Helsinki-based Silo AI specializes in developing tailored AI models, platforms, and solutions for companies in cloud, embedded, and endpoint computing. Silo AI claims to be Europe's largest private AI lab, employing over 300 AI experts across six countries.
The company also builds open-source multilingual large language models like Poro and Viking on AMD platforms. Dr. Pekka Manninen, director of science and technology at CSC-IT Center for Science in Finland, says Silo AI has pioneered training large language models on LUMI, Europe's fastest supercomputer powered by over 12,000 AMD Instinct MI250X GPUs.
Its clients include Allianz, Philips, Rolls-Royce, and Unilever. German AI company Aleph Alpha and French AI start-up Mistral are strategic partners of Silo AI. The deal is expected to close in the second half of 2024.
Silo AI will keep its headquarters in Finland but take a global approach. The company will continue tackling complex R&D problems in AI and developing multilingual open-source models, especially for European languages.
AMD Vice President Vamsi Boppana says the acquisition will accelerate the company's AI strategy and drive the development and rapid deployment of AI solutions for global customers. Silo AI CEO and co-founder Peter Sarlin will continue to lead the Silo AI team as part of the AMD Artificial Intelligence Group, reporting to Boppana.
According to AMD, the addition of Silo AI is part of a series of acquisitions and investments supporting its AI strategy. Over the past 12 months, AMD has invested more than $125 million in a dozen AI companies, such as the acquisitions of Mipsology and Nod.ai.
With this acquisition, AMD may also be trying to capitalize more on the AI boom that has made rival Nvidia the world's most valuable company over the past year and a half. The main reason AMD is lagging behind is its inferior machine learning software infrastructure.
