SoftBank acquires UK AI chipmaker Graphcore | TechCrunch

U.K. chip company Graphcore has been formally acquired by Japan’s SoftBank.

Rumors of the deal have abounded for some time, but protracted negotiations and regulatory approvals have meant neither company has confirmed anything until now. Even today, the company wouldn’t confirm the one thing most people will be wondering: How much does Japanese multinational SoftBank value a startup touted as a potential rival to the mighty Nvidia in the AI chip space?

While the figure of $500 million has been bandied around in various reports for months, in a press briefing early Thursday morning, Graphcore co-founder and CEO Nigel Toon remained coy on the details. “We have agreed with SoftBank that we’re not going into the details of the deal; whether anything comes out in the future, we’ll see,” Toon said.

Toon did say, however, that the $500 million figure was inaccurate. Make of that what you will.

When the chips are down

Founded out of Bristol in 2016, Graphcore has devised a new kind of processor dubbed an “intelligence processing unit” (IPU), distinct to the kinds of graphics processing units (GPU) developed by the likes of Nvidia. While both accelerate computation, IPUs have a different architecture designed from the ground-up for AI workloads. Graphcore pitches its chips as a more efficient alternative to GPUs, with a focus on supporting large-scale parallel processing and executing complex machine learning models, where the model and data are tightly coupled.

Graphcore had raised around $700 million since its inception, reaching a valuation just shy of $3 billion in late 2020. With big-name corporate and institutional investors like Microsoft and Sequoia, and angels such as DeepMind’s Demis Hassabis and OpenAI co-founder Greg Brockman, hopes were high that Graphcore could be an AI beacon in the U.K. or even all of Europe. But AI hardware is a resource-intensive business, and Graphcore ultimately couldn’t hit the giddy heights many had hoped it could reach. It lost out on potential lucrative cloud deals with Microsoft, while the U.K.’s own government ignored Graphcore (despite a public plea from Toon himself) for its new “exascale” computer plans last year.

Graphcore has not had the best of times of late, compounded last year by its forced China exit due to U.S. export rules.

With losses widening and Graphcore approaching four years since its last capital injection, it was increasingly clear that something had to happen somewhere — and an acquisition always seemed the most likely outcome, particularly at a time when demand for AI hardware is at fever pitch.

SoftBank, for its part, is no stranger to U.K. semiconductor companies, having previously acquired Arm for £24 billion ($31 billion) and then retained a stake as it spun Arm out as a $55 billion publicly traded company last year. Arm is now worth close to $200 billion — a sign, perhaps, that SoftBank might not be the worst bedfellow for Graphcore, as the well-financed Japanese powerhouse seeks to bolster its AI aspirations with everything from data centers and robotics, to the semiconductors needed to power the AI revolution.

Certainly, this is how the folks at Graphcore view matters. While those on the outside might view the sale to SoftBank as a missed opportunity for a U.K. or European company to create an independent AI hardware giant, Toon’s tone in Thursday’s briefing was one more of upbeat positivity.

For starters, Toon confirmed that he doesn’t expect any layoffs as a result of this acquisition, across its U.K., Polish and Taiwanese hubs, adding that, if anything, it will likely be adding “quite significantly” to its headcount in the U.K.

And importantly, both he and CTO co-cofounder Simon Knowles will be staying put both in their executive and directorship roles.

However, in most people’s eyes, Graphcore hasn’t really fulfilled its early promise. So what happened?

In a nutshell, the expenditure required in the space that Graphcore operates is an order of magnitude higher than what Graphcore was able to access as an independent company.

“Simon and I sat in a pub back in 2012 talking about this: AI, and the hardware that will be required for AI,” Toon told TechCrunch. “We’ve been thinking about this and what’s required for a long period of time, and we’ve probably been one of the early thinkers in the whole space. I think the piece that has surprised us [most] is the speed at which this has all taken off, and the scale that is involved.”

This “scale,” Toon says, involves systems of 100,000 interconnected AI processors, networking, liquid cooling and all the rest; it’s not exactly child’s play, and it’s far from cheap.

“This is a level of investment which is utterly massive, and the piece which is really interesting here is Graphcore is a relatively modest-sized company — large in terms of U.K. investments, but still modest in terms of the companies we’re competing with — and we’ve managed to go toe-to-toe and build world-class technology.”

Graphcore was always relatively demure in terms of headcount. For comparison, Nvidia has in the region of 30,000 employees, whereas Graphcore has around 500. And while Nvidia has grown organically over some three decades, Graphcore was trying to scale at a time when the post-pandemic capital markets weren’t friendly to startups of Graphcore’s disposition.

“The right outcome for the company here is to work closely with a partner who’s willing to make these levels of investments that are required to succeed in what is going to be probably the most important market in technology over the coming years,” Toon said.

Time will tell whether the acquisition proves a prudent move for the companies concerned, but Toon did confirm reports this week that some former employees’ stock was wiped out in this deal. This suggests that the acquisition price was below (or around) the figure it had raised, as investors and senior executives would be more likely to hold preferred stock options over ex-employees. Indeed, Toon confirmed that he did make some money from the transaction, without disclosing how much.

Toon also stressed that for all current employees and investors, the deal was a relatively positive outcome — for those willing to hang around, at least.

“There are a number of ways you can structure M&A deals. Sometimes that means that former employees don’t participate in what happens going forward, and unfortunately that’s the case here,” Toon said. “We’re sorry about that, but what I can say is, for all of Graphcore current employees, and the people who will be working with the company going forward, this is a great outcome for all of those. [And] it’s a good outcome for our investors. They’re all very happy.”

Graphcore gets the green light from regulators

Often when acquisitions of this magnitude are announced, there are months or even years of prolonged regulatory tussles. But here, SoftBank and Graphcore have gone through all the necessary antitrust and security approvals already; as a major infrastructure company, a deal such as this was always going to draw scrutiny under the U.K.’s National Security and Investment Act, which came into force two years ago.

“We have been through a very rigorous process of receiving all of the regulatory approvals for this deal, which is maybe why things have been rumored for some period of time,” Toon said. “All the approvals are in place, in the U.S. and other places as well.”

So that’s it: Graphcore is now officially owned by SoftBank, operating as a wholly owned subsidiary under its existing Graphcore name. The company’s headquarters will remain in Bristol, with additional hubs in London and Cambridge in the U.K., as well as offices in Gdansk (Poland) and Hsinchu (Taiwan).

What happens next for Graphcore as a SoftBank subsidiary is up in the air, but Vikas J. Parekh, managing partner at SoftBank Investment Advisers, stresses that Graphcore will now play a big part in its pursuit of AI riches.

“Society is embracing the opportunities offered by foundation models, generative AI applications and new approaches to scientific discovery,” Parekh said in a press release issued to TechCrunch. “Next-generation semiconductors and compute systems are essential in the AGI [artificial general intelligence] journey; we’re pleased to collaborate with Graphcore in this mission.”