Advanced Computing in the Age of AI | Friday, April 19, 2024

AI Chip Startup Graphcore Rings Up $200M in Funding 

via Shutterstock

Graphcore, the U.K. AI chip developer, has attracted more investors as it rolls out its intelligent processing unit to enterprise datacenters and cloud services providers.

The GPU rival reported this week it has closed a $200 million funding round led by the London-based venture firm Atomico and other existing investors. New investors included automaker BMW and Microsoft, who join Robert Bosch Venture Capital, Dell Technologies Capital and Samsung Electronics as auto and cloud investors.

Graphcore, Bristol, U.K., said Tuesday (Dec. 18) its latest funding round brings its venture capital total to more than $300 million. The two-year AI chip startup claims a market capitalization of $1.7 billion.

Automakers said they are intrigued by the possibility of using Graphcore’s IPU in their datacenters as well self-driving vehicles. “Graphcore’s IPU also presents the possibility of reduction in development times and complexity,” added Tobias Jahn of BMW i Ventures.

Graphcore said the new funding would be used to ramp chip production as it expands its international operations to Silicon Valley, the booming Chinese market and elsewhere. “Our goal is to build the global market leader in machine intelligence hardware,” said Nigel Toon, Graphcore’s co-founder and CEO.

The startup has been shipping AI chips to “early access” customers. Its IPU has entered high-volume production as demand increases for its AI chip running on a PCI Express processor card.

Graphcore is positioning its IPU as a GPU competitor and as an AI chip customized for machine intelligence training and inference applications. The startup claims up to a 100-fold performance boost over GPUs and other AI chips.

Graphcore’s first chip, codenamed Colossus, is a based on 16-nm FinFET process technology. The massively parallel, mixed-precision floating point processor began shipping earlier this year. Two IPUs run on a single PCIe card, and cards can be configured into a cluster, the chip maker said.

About the author: George Leopold

George Leopold has written about science and technology for more than 30 years, focusing on electronics and aerospace technology. He previously served as executive editor of Electronic Engineering Times. Leopold is the author of "Calculated Risk: The Supersonic Life and Times of Gus Grissom" (Purdue University Press, 2016).

EnterpriseAI