A startup in Bristol has raised $30m from investors around the world for its machine learning and artificial intelligence chip and system development.

“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”

 

GraphCore already employs 40 engineers in Bristol and Silicon Valley and is looking to expand its design team in Bristol.

“We are doing this is stages – we have been here before with other companies and the right thing to do is have a manageable crack team of less than 100 while developing the technology and then scale the company behind the revenue growth,” said Nigel Toon, CEO. “We have enough experience in the company and there’s a huge depth of talent here, many of whom have worked together before.”

Toon and chief technology officer Simon Knowles co-founded Icera Semiconductor in Bristol developing the next generation of cellular modems, with Knowles going onto NVIDA and Toon to XMOS Semiconductor where he is still chairman.

“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.

The funding was led by Robert Bosch Venture Capital with Samsung Catalyst Fund and other major technology firms, alongside leading venture capital funds from London, Silicon Valley and Israel: Amadeus Capital Partners, C4 Ventures, Draper Esprit plc, Foundation Capital and Pitango Venture Capital.

Bristol to be engineering base

“We could have raised lots more, but there’s a point at which you need to optimise value of the company and we can go back and raise more later at a higher valuation,” said Toon.  “We already have 40 people in the company and quite a large team working on this for the last two years. We are focussed on an engineering team here in Bristol and activities in Silicon Valley will be focussed on business development and customer support.”

“Graphcore has a unique technology that has massive potential in the fast emerging market for deep learning”

 

The first stage is a box – the IPU-Appliance – that boosts machine learning in data centres so that apps can learn by themselves, based around GraphCore’s Intelligent Processing unit (IPU). “We will have product in the market by Q3 next year as a system product,” said Toon. “We are building our own customer microprocessor and packaging that into a PCI-Express system and putting that into a server system to create an appliance that will go into cloud infrastructure,”.

The next stage is to develop low power chips for the edge of the internet in consumer devices and driverless cars, which is where companies such as Bosch and Samsung come in.

“Graphcore has a unique technology that has massive potential in the fast emerging market for deep learning,” said Dr Hongquan Jiang, Partner at Robert Bosch Venture Capital. “A new processor technology is needed for intelligent systems and Graphcore has the first technology that we have seen which really delivers the performance and efficiency needed for this style of compute. We are excited to have led this very significant investment round.”

Investors from all over the world

“We have gone all over the world to collect our cash,” said Toon. “Bosch is very interested in the machine learning space and the next generation of automotive and transportation and there are some projects that we are working on that we can’t talk about. Longer term, our technology can go into embedded projects and driverless cars so having Bosch as an investor makes a lot of sense.”

“With the technology we are developing we are opening new areas of machine learning so having a system that supports training is a useful initial step before moving to the edge of network and embedded designs,” he said. “We’ve been working on this for nearly two years so we are a long way down the path and have a lead [on other companies].”

There is more detail on the company at https://www.graphcore.ai/