Graphcore, the UK-based chipmaker founded only two years ago in 2016 has achieved a $1.5 billion (£1.18 billion) valuation with its latest $200 million (£157.67 million) funding round.
With investors now including the likes of Microsoft and German automaker titan BMW, Graphcore will have ambitions to emulate the success of ARM, the UK chipmaker last year acquired by Japan’s Soft Bank for $31 billion (£24.44 billion). Graphcore designs and manufactures the kind of next generation microchip whose processing power contemporary and future AI applications need to run their complex data-crunching algorithms at high speed.
The recent funding round, which included Microsoft and IBM, was led by Atomico and Sofina, both major European investment funds. Graphcore, who only began shipping its first consignments of chips to customers earlier this year, will use the funds to ramp up production. The company was founded in Bristol two years ago by Simon Knowles and Nigel Toon, both semiconductor industry veterans.
Graphcore makes the ambitious claim that its microchips represent ‘the world’s most complex processor’. Mr Toon added that it has been “designed from the ground up to do machine intelligence applications”. Industry scions such as DeepMind co-founder Demis Hassabis and Uber chief scientist Zoubin Ghahramani endorse the claim to the extent that both have become investors in Graphcore.
Tobias Jahn, principal of BMW I Ventures, the venture capital unit of the automaker that is investing heavily in becoming a leader in autonomous vehicle technology explained his company’s decision to invest in Graphcore:
“The versatility of Graphcore’s intelligence processing unit — which supports multiple machine learning techniques with high efficiency — is well-suited for a wide variety of applications from intelligent voice assistants to self-driving vehicles.”
Other investors in the company’s most recent round included Samsung and Dell from the technology industry and high profile VCs Sequoia Capital and Amadeus Capital. The specialised nature of the kind of microchips needed for AI and machine learning algorithms has seen a new dedicated microchip industry spring up. The titans of tech such as Google, Apple, Facebook, Nvidia, Huawei and Intel have all invested in developing their own AI-specialised processors and several start-ups in the field have raised hundreds of millions of dollars.
AI has made leaps and bounds forward over the past few years thanks to superfast GPU chips and cloud computing offering access to the huge volumes of data needed to train machine learning algorithms. However, the kind of chips being manufactured by Graphcore take things a significant step further than Nvidia’s GPUs. They are able to work on multiple tasks in parallel. Previous chip generations have worked on the basis of sequential processing. This is of great value to the complexity of machine learning tasks.