Softbank Buying ARM Is a Bad Idea for Everyone

Project Marmaris_infographic_v1

Japanese tech conglomerate and Sprint owner Softbank today proposed to buy ARM for $32 billion, a 50 percent premium on what ARM would otherwise be worth (according to Recode) and a terrible, terrible idea. It's bad for ARM. It's bad for phone makers and wireless carriers. It's even bad for Sprint.

ARM holds a benign monopoly on the instruction set used by almost all smartphone processors, and increasingly on the chips used by the "Internet of Things," the chips which will go into everything from smart roads to medical equipment. Because nearly everyone agrees to use ARM's instruction set, which is like a language, OS and software developers don't have to sweat out an entire level of compatibility that they would otherwise have to worry about. Intel, AMD, and others agreeing to use Intel's x86 instruction set helped jumpstart the home PC revolution. ARM did the same for mobile tech.

Apple and Qualcomm take ARM's instructions and build their own processor families. But Samsung, Mediatek, and many lower-cost chip designers, especially in the Internet of Things realm, buy ARM's second product: entire chip blueprints. ARM's perfect neutrality, never preferring one customer over another, means these vicious competitors can all trust they can get a fair deal from ARM. If they doubted that for a moment, the system would break.

Softbank isn't neutral. Quite the contrary. It owns two major wireless carriers, in the US and Japan. Even the perception that Sprint engineers might get early looks at ARM chipset designs, or that ARM's designers would have a special financial interest in advancing Sprint's and Softbank's TD-LTE network technology over others, will create suspicion in the industry. This is one of those "appearance of impropriety" things; it won't take actual evidence of synergy for AT&T to start looking for other vendors for its very IoT-focused business strategy, just the possibility that synergy may be happening in secret.

Even more than synergy, though, Sprint could really use that $32 billion, and the focus Softbank will give to ARM. Sprint is doing better than it was doing before Softbank invested, but it has sunk to fourth place in the surprisingly competitive US wireless market, largely because T-Mobile has been such a juggernaut. Sprint needs all the help it can get, and a parent company spending its time and money on ARM won't give it that help.

Softbank CEO Masayoshi Son isn't an idiot, so what's really going on here? I suspect he's shopping for a buyer for Sprint. The US wireless market is growing slowly, but there's huge potential growth in the future connected world for ARM. Remember that Son's previous game plan was to merge Sprint with T-Mobile, putting T-Mobile CEO John Legere in charge and possibly spinning it all off for a profit? The US government put the kibosh on that, as the whole point of the merger was to reduce competition. Sprint hasn't been a stellar performer in Softbank's portfolio—and ARM has a lot more potential for the future. The question there is, who wants a slightly damaged US wireless carrier?

Recode says Apple and Intel are other potential bidders. If ARM isn't going to stay independent, Intel is by far the best choice of that trio. Apple buying ARM would be like a nuke dropped on the rest of the mobile industry, as Apple has never been shy about leveraging its purchases for maximum competitive advantage. Apple would automatically get the newest ARM designs in advance of everyone else, ultimately driving Samsung and others towards Intel, MIPS, or more exotic competitors. This hurts consumers because a huge amount of effort would then have to go into making multiple versions of operating systems and apps for different processor designs; getting Android to be completely compatible with Intel x86 took years, and even now, popular games like Pokemon Go don't run on Intel designs.

Intel buying ARM, on the other hand, would be the end of a process I've long hoped was inevitable. Long ago, Intel made the best ARM-compatible early smartphone chips, then decided to sell off the business to Marvell so the company could bash its head against a wall for 10 years trying to get smartphone firms to convert to its competing x86 architecture. This year, mobile x86 lost Asus, its last major client.

Intel has been talking a lot about the Internet of Things, but it's playing from a position of weakness there, too. So Intel isn't a strong competitor in the mobile architecture world; it knows chip building like nobody's business; it has a strong position in servers, where ARM has been trying to grow; and it's used to being neutral in a competitive PC industry. If ARM has to sell to one of these players, let's hope Intel finds the cash.

This article originally appeared on PCMag.com.