Why did ARM need Softbank to buy it rather than the need coming from the other side? SemiAccurate thinks that while Softbank will do very well investing in ARM, the real benefit is to the other side.
First let me say that I think this deal is a good thing, and Softbank should profit handsomely in the long run by it. It will also be good for the ARM investors and more importantly their partners, the latter being the key to the deal succeeding. Softbank has the size and clout to do things that ARM just couldn’t muster, ubiquitous in computing as they are. This added financial heft will benefit ARM in the short and mid-term.
There have been a lot of rumors surrounding ARM and acquisitions over the years, and all seem to be based on fantasy or more likely short positions rather than a shred of sanity. The main reason is that they all take one of the following forms, “chip company X is going to buy ARM”, or “device company Y is going to buy ARM”.
These both can’t happen for one simple reason, ARM’s strength and value lies in its ubiquity. If you buy an ARM based product, software designed for that core will run on anyone else’s ARM core. This does not mean Apple apps will run on Android devices or even Android apps will run on other Android devices, but the part that runs on the ARM cores will be compatible with the ARM cores on different devices.
This is due to ARM’s very deliberate and rather militant compatibility rules enforced over the years. If you are going to use one of their designs in your product, it will be compatible with ARM’s very exacting and extensive software test suite or it won’t be given the green light to come to market. Everyone grumbles and occasionally complains publicly about this but in the end it benefits the designers and they know it.
It is also what has allowed the current wave of mobile computing to happen, no compatibility, no ecosystem. It is the same with the dying industry of desktop computing, WIntel allowed PCs to flourish and then smothered it. ARM seems to have learned from their predecessor’s failings and is keeping competition and cooperation alive among their licensees. Some of those licensees may likewise grumble but they also benefit from a massive TAM.
ARM’s militant compatibility diktat means every SoC vendor out there doesn’t need to reinvent the software wheel with every new chip, a process that can cost more than silicon development in some cases. On top of this they have an extensive tool, compiler, and middleware library to choose from, not to mention the millions of existing apps. It is a virtuous cycle and it is partly why we raised the question of compatibility for SiFive, but in that case it was more about how compatibility will be enforced, not a red flag that it won’t happen.
So this is why SemiAccurate laughed at the semi/device takeover rumors. If any company in either category bought ARM, the other 99+% of their licensees would immediately bail out. Why? Would you want your competitor to have control of your silicon flow? Would you want to have them check out your plans, know your finances, and every relevant date in the process? Probably not. From the device side, would you want your competitor to control your silicon supply knowing that any vendor who displeased them could have unforeseen compatibility suite problems like Microsoft does with Windows drivers for companies not being compliant enough to their business plans?
Neither category of businesses could take over ARM without the entire ecosystem falling apart in very short order. A takeover would mean that every licensee would move to a different standard ASAP. It may not currently exist, may take time to herd the proverbial cats, and may have several failures before the real path is discovered, but it will happen. If you think this is impossible, just look at the lock on computing that Microsoft had the day before they announced that they entered the hardware business with Surface. Look at the number of their ‘valued partners‘ who were daring to mention Android or Chromebooks that day. The very next day WIntel ended.
As SemiAccurate has chronicled, it took ~18 months for the Taiwanese and Asian industry that builds most compute devices to snap out of their shellshock and find a direction. They did. It isn’t Windows and largely isn’t Intel. Microsoft and Intel’s response has been panic, desperate folly, And a precipitous decline in sales with no end in sight and no real plan to right the ship.
That again brings us back to Softbank, they have effectively no conflicts of interest with ARM partners on the SoC design/manufacturing side or the end-user device side. They do have investments in infrastructure and bandwidth providers, both wired and wireless, but neither of those should scare current ARM licensees or their customers. In short if Softbank can resist mucking around with the fundamentals of how ARM does business, and it looks like they don’t do that with their other investments, all should be smooth sailing.
From the hour long press conference with Masayoshi Son of Softbank’s, it was pretty clear he doesn’t intend to change much and sees ARM as a long-term bet, not a short-term cash out. Again looking over the company’s past investments, it is clear that Softbank takes this strategy seriously. While they definitely intend to make money, it doesn’t need to happen overnight. That is why Softbank wanted ARM, but why does ARM need Softbank?
Note: The following is analysis for professional level subscribers only.
Disclosures: Charlie Demerjian and Stone Arch Networking Services, Inc. have no consulting relationships, investment relationships, or hold any investment positions with any of the companies mentioned in this report.
Latest posts by Charlie Demerjian (see all)
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018
- Intel can’t supply 14nm Xeons, HPE directly recommends AMD Epyc - Sep 7, 2018
- AMD reintroduces the Athlon name with two CPUs - Sep 6, 2018