The news that Japan’s Softbank is spending £24bn to buy ARM – the UK company which designs (but does not actually make) the chips that power pretty much every smartphone and tablet in existence, plus almost every internet-of-things product in existence – has got people very confused.
The price is a 41% premium over its closing price on Friday – though the share price leapt on the news by 45%, creating a puzzle for those who hold the shares. A few points first:
• it isn’t specifically caused by the fall of the pound after the Brexit referendum. True, Softbank is buying in yen, and that has appreciated against the pound – from about 160 yen/pound just ahead of the referendum to 140 yen/pound on the day of the bid. That’s a 15% change, so Softbank got some benefit on the side. Possibly the market thinks it could go higher.
• it isn’t going to affect “British companies” and your smartphone won’t get more expensive. ARM designs chips, but doesn’t make them; there are no British companies making chips in British foundries from ARM designs. There would be no advantage to Softbank or ARM to increase the royalty on its new licences because companies would simply stick with the older versions.
• Softbank says the purpose of the takeover is to benefit from all the “internet of things” applications that are coming into view. Given that this is a market which will be multiple times bigger than the smartphone/mobile phone market, which was already multiple times bigger than the PC market, you can see that Softbank thinks it’s on to a good thing.
Design for life
But take the valuation on its face, and consider this: ARM doesn’t actually make anything physical. All it does is produce the designs for chips, which others then have to take away and turn into something physical. They pay ARM a royalty for each chip they make using its designs.
It might not sound like a promising way to rake in money, but ARM has the advantage that it can shift its focus as quickly as the market demands, and can hire and deploy people wherever it needs to. There’s barely any capital investment to consider, apart from a few offices and testing facilities. Its value comes directly from its staff and their capabilities. As is often said of IP-related businesses, the company’s future value literally walks out of the door every night. It truly is a weightless corporation.
ARM lets other companies, such as TSMC, Samsung and Apple do things with its designs, in return for royalties. Apple goes one stage further, tweaking the designs and then getting someone else to build them.
But through this all, ARM is weightless. It has a comparatively tiny staff, but precisely because of that it is able to pick the path to focus on. When it was set up in November 1990, the idea seemed a bit fanciful: everyone knew you had to own your own chip foundry to be successful at making chips, like Intel and AMD. Letting other people make your chips wasn’t going to be profitable. Being just a design company seemed even less sensible.
However, ARM turned out to have made all the correct calls. It focussed on low power consumption. It focussed on functionality. It focussed on the RISC (reduced instruction set computing) model – unsurprising, since its name is a contraction of “Advanced RISC Machines” – which has benefits both for low power and for reliability. (To understand the difference between RISC, as found in phones/tablets, and Complex Instruction Set Computing, CISC, as found in Intel-style PCs: think of RISC as a short-order cook who can do a few simple meals but quickly; think of CISC as the chef who can do amazing a la carte, but needs lots of time and care to prepare it.)
RISC turned out to be where the world wanted to go – though not for PCs, where RISC turned out to be a bust (Motorola’s PowerPC model, used by Apple and IBM, simply couldn’t keep up for sheer processing power with the Intel CISC model; Apple abandoned it in 2005, having always hedged its bets by keeping an Intel-compatible version of Mac OSX in the laboratories ready to be deployed.
However mobile phones and smartphones and tablets and chips in your lightbulb and thermostat and monitoring camera all benefit from being RISC-based. There’s a big, big future there.
Compare the £24bn bid value to Intel’s valuation, which at the moment stands at $165.60bn (£114.84bn at current exchange rates). Intel has a business model very unlike ARM’s: it designs the chips and it then builds them. This has allowed it to make gigantic profits when times are good and when the world wants the chips that Intel is making.
But you could see Intel as two companies: an Intel-ARM (call it Intel-A) which designs chips, and a “sub-Intel” which then makes them. Intel-A only licences to sub-Intel, and sub-Intel generally only makes chips designed by Intel-A.
Put that way, it sounds a bit inefficient: wouldn’t Intel-A make more money by licensing its chips to anyone who wanted to make them? And couldn’t sub-Intel use its foundries more efficiently by making chips from anyone?
Certainly a lot of (combined) Intel’s value is tied up in its capital assets. Its Q1 2016 balance sheet shows “property, plant and equipment” having a value of $32.64bn, and inventories of $5.75bn – that’s $38.39bn tied up in capital, or nearly a quarter of its value tied up in capital. It’s hard to believe though that an orderly selloff of Intel’s foundries would recoup all that. Much of the valuation (and by extension the company’s market cap) lies in the expected utility of those foundries to make chips that people will want to buy.
Compare that with ARM’s 1Q results. Its direct revenues are tiddly by comparison – £276m, but the operating margin is 48.6% and it had operating profits of £137.5m. Its “plant and equipment” is a grand £58.9m, and inventories £1.6m; that’s
just 2% of its value. ARM could change direction to focus on any sort of chip design it wanted at a moment’s notice. If Intel wanted to start making ARM-design chips, it might be challenged because those foundries aren’t optimised for it.
The trouble with the future
That’s all great while the world is buying a growing number of PCs and Intel-design x86 chips. But it isn’t. The number of PCs sold continues to fall (it’s now down to 2007 levels, ie pre-financial crash) and there’s no sign of Intel making great inroads into either the smartphone/tablet space or the internet-of-things space. Intel recently reorganised itself to roll its “mobile” element into its “client computing” group, thus removing a lot of embarrassing red ink that the mobile side had pulled together.
But Intel also knows that only the paranoid survive; and so it has an “Internet of Things” group which is strongly profitable ($123m on sales of $571m in 1Q 2016), though the sums involved aren’t big by its standards. But they are growing.
The trouble with the future for Intel is all that capital tied up in its foundries, which represent a legacy that is now fading. They’ve done great, and will surely continue to make money. But the growing number of delays to new Intel chips, and its move away from “tick-tock” to a three-year schedule, suggest that it’s struggling to cope with the demands both of its past investment, and the demands of the future.
Contrast that with ARM’s stated intention:
ARM technology now reaches around 80% of people in the world, with chips based on our technology driving billions of products every day. To date more than 86 billion ARM-based chips have been shipped, and our Partners are shipping over 4 billion every quarter. Our strategy is to develop and deploy energy-efficient technology; to enable innovation through a broad ecosystem of Partners, building on our shared success; and to create superior returns for our shareholders by investing in long-term growth.
You’d have to say that ARM looks like a good bet for the future – better, for the moment, than Intel.