Graphcore gets big backing for machine learning

October 31, 2016 // By Peter Clarke
Graphcore Ltd. (Bristol, England), a startup developing a processor for machine learning, has completed a $30 million Series-A round of funding led by Robert Bosch Ventures and Samsung Catalyst Fund.

The unusually large Series A has been put together by a notable team of investor that also included Amadeus Capital Partners, C4 Ventures, Draper Esprit plc, Foundation Capital and Pitango Venture Capital. Other technology companies besides Samsung were involved in the funding Graphcore said, without revealing names.

Nigel Toon, co-founder and CEO, said the money will allow Graphcore to "execute at scale" and that the investors being contacts and market insights that will help Graphcore understand and solve real-world challenges.

Graphcore has been able to hit the ground running because it was incubated for two years within XMOS Ltd. while Toon was CEO there. The two years were spent building an experienced hardware and software team. The technology is based on a neural network accelerator called the intelligent processor unit (IPU) which Simon Knowles, previously CTO of XMOS, had been working on for some time, the source said. Knowles has now taken up the CTO position at Graphcore.

The processor and modules based are being design from the ground up to accelerate both current and next-generation machine intelligence applications such as natural language dialogue, autonomous vehicles and personalized medicines.

Graphcore plans to bring the IPU to market in 2017 initially to lower the cost of accelerating artificial intelligence applications in cloud and enterprise data centers. The company claims the IPU-Appliance can increase the performance of both training and inference by between a factor of 10 and 100 compared to the fastest systems in use today.  The IPU will also be made available for embedded applications such as autonomous cars, collaborative robots and intelligent mobile devices.

As well as computation software tools and libraries that will enable machine learning across a broader front than the current focus on feed-forward neural networks, Graphcore said.

"Machine intelligence will have a bigger impact on our lives over the next 10 years than mobile technology has had in the last two decades," said Toon, in