Few people appreciate Swaffham Bulbeck’s role in changing the course of UK tech. If anything, the tiny Cambridgeshire village is better known for the antics of its amateur dramatics society and its rambunctious annual productions of Gilbert and Sullivan operettas, held each summer in a converted barn. It was in another farmyard outbuilding, however, where a dozen semiconductor engineers gathered in 1990 to build what would later become Britain’s most influential technology company.
Founded as Advanced RISC Machines Ltd and repeatedly rebranded over its 34-year history, Arm has built a winning reputation for designing mobile processors that boast low power consumption and high efficiency. These chips can be found almost everywhere, powering millions of smartphones, sensors and connected vehicles, not to mention innumerable laptops, tablets and servers. What’s more, Arm’s sales model, which sees it provide designs to tech companies in exchange for an upfront licensing fee and subsequent royalties, has enabled it to do business with anyone without fear or favour, earning the company an enviable status as the “Switzerland” of the global chip market.
This combination of ubiquity and brand neutrality attracted the attention of Masayoshi Son, the charismatic founder of Japanese conglomerate SoftBank and a man who made a fortune with a series of visionary investments in tech companies. In 2016, with Son seemingly at the height of his powers and the value of the pound plummeting following the Brexit vote, SoftBank swooped in to buy Arm, taking the company private for what many considered a bargain price of $32bn.
But SoftBank has struggled to make this investment pay off, and its acquisition of Arm has become emblematic of a series of wrong turns made by Son and his team. After an attempt to sell the chip designer to one of its customers, Nvidia, for $40bn was rebuffed by regulators, SoftBank announced plans to re-float the chip designer on the New York-based Nasdaq stock exchange – despite repeated attempts from the UK government to persuade it to list in London instead.
Arm’s second IPO is now a matter of days away. But what does the flotation mean for the company’s future – and for the tech leaders who have come to rely on devices built with its designs?
Why does Arm need an IPO, anyway?
One person who was unconvinced that the company should go down the IPO route was Simon Segars. “We contemplated an IPO but determined that the pressure to deliver short-term revenue growth and profitability would suffocate our ability to invest, expand, move fast and innovate,” the former Arm CEO wrote in a blog post on the company website in July 2021.
That was written when Arm was still actively pursuing a takeover by Nvidia. Segars, who had led Arm since 2013 and negotiated the company’s original sale to SoftBank, left his post following the collapse of that deal, to be replaced by Rene Haas. Now the company is pressing ahead with a listing which, according to a filing handed to the US financial regulator the SEC last month, will value the company at between $48-$52bn. SoftBank plans to hang on to 90% of Arm shares, meaning it can expect to raise around $5bn from the IPO.
This abrupt u-turn on the merits of an IPO illustrates the slightly uncertain position the company finds itself in, says James Ashton, author of a new book on Arm entitled The Everything Blueprint. “The joint submission made by Arm and Nvidia to [UK competition regulator] the CMA makes a pretty good case as to why you wouldn’t float Arm,” he says. “You need long-term investment, and the public markets supposedly wouldn’t let you do that because you have to spoon out dividends all the time. They made this case in public themselves, and then Rene had to spin on a sixpence six months later.”
As such, it remains debatable whether the upcoming listing is going to be more beneficial to Arm or its parent company. While the former’s revenue has grown 65% since the SoftBank takeover, costs have also soared under the Japanese company’s leadership, thanks in part to Son making good on his 2016 promise to double the company’s headcount. It has since made significant job cuts, but still employs 5,963 people globally, compared to some 4,000 at the time of the acquisition.
SoftBank has financial issues of its own thanks to a series of poor investments, not least its very public commitment to WeWork in 2017. Once valued at $47bn, WeWork said last month there are serious doubts whether it will be able to continue trading. SoftBank’s investments into the company are said to have topped $18bn.
It’s for this reason that SoftBank has been looking to increase the return on its investment in Arm, explains Alan Priestley, an analyst at Gartner. Even so, “there are only so many knobs you can twiddle” on Arm’s business model to grow profitability, says Priestly. “You can’t retrospectively increase royalty payments for existing customers.”
Such is the downside of the vaunted Arm business model. While it receives royalty payments on the number of chips manufactured using its blueprints, it does not cash in on the much higher value of the devices in which they are used. Because of this, “as a business, it’s not the same as a chip manufacturing company where you can see 50% annual growth,” Priestley says. “There’s much more limited scope.”
But Arm could benefit from the IPO if the profits are returned to the company to fund R&D into the next generation of its chips. “Developing leading edge processors costs a lot of money, and Arm is trying to move up the stack,” says Priestley. “Maybe [the IPO] will give it more resources to expand, move the architecture on and develop its IP. Arm needs to keep moving forward because it wants to compete with the x86 players in the data centre.”
Data centre dreams
Son’s initial plan for Arm was to take advantage of what he anticipated would be a boom in demand for Internet of Things (IoT) connected devices, powered by Arm chips. However, this has not proved hugely profitable, and much of Arm’s IoT business has been separated off into a different part of SoftBank.
What, then, is the big growth opportunity for Arm? Data centres, for one thing. Arm’s chips are increasingly being used in server halls around the world, taking a slice of a market that, up until now, has been dominated by the more powerful processors built on the x86 architecture developed by Intel. Recent years, however, have seen public cloud leader AWS develop its own in-house chips, the Graviton series, based on Arm designs, while other vendors such as Ampere Computing are building Arm-based data centre processors for their clients.
In personal computers, another sector where x86 traditionally rules the roost, the company is also making progress, with Apple having abandoned the aforesaid architecture in favour of Arm designs for its in-house chips, the M1 series. Qualcomm also expects to see its Arm-based Snapdragon processors become more commonly used on Windows PCs.
Both Apple and Ampere have architectural licences, meaning Arm does not benefit financially from royalties when they produce their chips. But Priestley says interest in these processors helps build Arm’s data centre credibility.
“Arm in the data centre has been a nirvana for a number of years,” he says. “The problem is performance trumps power, so Arm needed to invest in high-performance IP. It’s been gaining traction with its Neoverse cores, which are what AWS uses, and it’s reasonable to think that some of the other hyperscalers, as they move into having their own processor designs, will use Arm IP.”
Arm is also keen to stress it believes there is an opportunity in the fast-growing AI market, but some experts are sceptical.
“It's a lot more challenging for Arm to capitalise on the current trend for AI than a company like Nvidia,” says Albie Amankona, an analyst at consultancy Third Bridge. “Around 60-70% of Arm's revenues are derived from mobile, and the AI landscape is centred around cloud-based operations, rather than being heavily integrated at the device or edge level, where Arm is more prominent.”
Even as the market matures and demand for AI-enabled edge devices increases, it is unclear whether Arm will be able to cash in, Amankona says. “The growing adoption of RISC-V chip architecture is the biggest risk for Arm,” he adds, referring to the open-source blueprints which have been gaining traction in recent years.
Indeed, RISC-V is already making inroads in mid and low-end applications, such as the embedded market, Amankona explains. “It seems inevitable that Android and Microsoft will eventually embrace RISC-V,” he says. “It offers them the opportunity to lower overall costs and maintain or even improve performance.”
What does the Arm IPO mean for tech leaders?
There has also been speculation in the run-up to the IPO that Arm could change its pricing model. In March it was reported that the firm was holding discussions with several major customers about charging royalties based on the value of the final device, rather than the chip contained within it. While such a move could potentially increase revenue significantly, it may also push tech companies into considering other architectures such as RISC-V.
Gartner’s Priestley does not think such a move is realistic, at least not in the near future. “It’s a great concept but I struggle to see how it would work,” he says. “Arm has direct relationships with the customers that licence its IP – the companies that make chips – and that’s a constrained number of engagements.
“If you start charging per end-user device, the number of companies it would have to engage with and collect royalties from goes up by an order of magnitude. And is Mercedes really going to pay for every Arm chip in one of its vehicles based on the value of that vehicle?
“You might be able to do it for specific pieces of IP or specific pieces of equipment, but the risk is that you push the margins of the manufacturers of that equipment and they start to look at alternative architectures.”
IT buyers should not expect any big changes from the Arm portfolio in the immediate aftermath of the IPO, Priestley says. “If you’re in the IT industry today you may be buying certain equipment, such as routers, with Arm inside,” he adds. “Will you start buying servers based on Arm IP? I think it’s questionable because most of your software is going to be geared towards x86.
“I think we’ll start to see more AI appliances deployed which are based on Arm, but that won’t impact an IT manager at this point in time.”
Ashton says that, as a public company, Arm will be under pressure to deliver both profit and growth, and that steps taken by Haas since he took the reins will help with this.
“Arm has always been a ‘shoestring’ business, and I think what was very un-Arm in the early SoftBank days was that it had a lot of money and didn’t always know what to do with it,” he says. “That changed as soon as the Nvidia deal was called off, and Rene has done a good job of focusing that spending on areas like automotive.”
This change of dynamic in the company will eventually be reflected in its offering to IT buyers, argues Ashton. “I think they would say they have diversified their royalties in the last couple of years based on decisions and investments made six or seven years ago,” he says. As such, Ashton continues, “We shouldn’t expect to see the impact of any changes made now for some time. Arm has always been a long-term play.”
But, he says, there could be some alterations to licensing coming down the track. “I think what they might do is try and shift more partners onto broader licensing terms,” Ashton adds. Consequently, he says, “they’ll be given the opportunity to buy a whole suite of blueprints – but it might end up costing a bit more.”