SoftBank: Oi, we paid $32bn for you, when are you going to strong-Arm some more money out of your customers?

Arm was pressured by its owner SoftBank to jack up its processor core licensing fees for some of its customers in an effort to squeeze more money out of its $32bn buy of the British chip designer, sources say. According to those who spoke to Reuters, the licensing price rise was steep: four times the usual cost in some cases …

[elkster88]

SoftBank bought a goose that lays golden eggs...

And the first thing the proud new owners do is decide to practically squeeze the life out of it, to extract more eggs, then they wonder why it's suddenly stopped laying.

Penguin icon for obvious reasons.

[big_D]

Re: SoftBank bought a goose that lays golden eggs...

Yes, groups like SoftBank aren't interested in sustainable business models, just as much return as they can get in the short term, regardless of whether it kills the host in the long term.

[Steve K]

Re: SoftBank bought a goose that lays golden eggs...

Not sure if that is entirely fair for SoftBank as they have a stated 300-year mission (or is that Star Trek)

Having said that, what was Mr. Son smoking when he invested so heavily in WeWork?

[BillG]

Re: SoftBank bought a goose that lays golden eggs...

When investment firms that don't understand semiconductor companies buy a semiconductor company, it never works out well for the semiconductor company.

I'm thinking of multiple examples here, including a company that had a batwings symbol.

[FatGerman]

Re: SoftBank bought a goose that lays golden eggs...

There's a genuine definition of an entity that behaves like that - parasite.

The drive for short-term profits at the expense of everything else has been the death of every company I've ever worked for. And still nobody learns. I don't understand why the money men are so stupid.

[John Brown (no body)]

Re: SoftBank bought a goose that lays golden eggs...

"I don't understand why the money men are so stupid."

From their own perspective, they are not stupid. They got the cash from the cow and then moved onto the next cash cow. They don't lose out when the previous cash cow dies.

[Gonzo wizard]

Re: SoftBank bought a goose that lays golden eggs...

Next they'll be wondering why they can't achieve a good price if they do indeed try to divest. The effect of price rises won't be immediately apparent while current customers investigate and adopt alternatives (if suitable ones are available), but in the medium term this could have a noticeable impact on ARM's revenue stream.

Heck, if enough customers jump to the same alternative, that could fund further improvements and give ARM a proper fight...

There's always something to look forward to ;-)

[J. R. Hartley]

Re: SoftBank bought a goose that lays golden eggs...

I just still can't believe the Tory scum allowed the sale in the first place. It should have been blocked. It was one of the very few British success stories. I know the original Acorn crew were very upset about the sale.

[Tom 7]

Re: SoftBank bought a goose that lays golden eggs...

I was too - despite what Softbank think now I swear it was worth at least twice what they paid for it. If you care for investments over 5 years that is.

[Timto]

Re: Meh

LOL - Do you honestly think Labour would have done any different?

I doubt more than 10 MPs have even heard of ARM

[J. R. Hartley]

Re: Meh

Who mentioned Labour you nonce?

[Anonymous Coward]

Re: Meh

Americas "tribal politics" is starting to get more and more popular over here.

You can't criticise any politician these days without automatically being assumed to be praising their opposition.

It's always been here to some extent, but with Trump, and the increasing tribal rhetoric from the tabloids, it's getting a lot worse.

Still, it exposes the morons!

[Anonymous Coward]

Re: Meh

@ J.R. Hartley

You have just called someone a "nonce".

https://dictionary.cambridge.org/dictionary/english/nonce

You should be ashamed.

Cheers… Ishy

[J. R. Hartley]

Re: Meh

Why would I be ashamed? Everyone knows Tories are nonces.

[Anonymous Coward]

Re: Meh

Don't say that... Linus will have to remove all kernel mention of cryptographic nonces!

[Anonymous Coward]

Re: SoftBank bought a goose that lays golden eggs...

Perhaps the current customers should all get together and make Softbank the only offer it's going to get.

[Zippy´s Sausage Factory]

Re: SoftBank bought a goose that lays golden eggs...

Why do I think it's either going to be Apple or Google that ends up the winner here?

[Anonymous Coward]

Re: SoftBank bought a goose that lays golden eggs...

@Zippy Established practice, I suppose?

[Mark Exclamation]

Re: SoftBank bought a goose that lays golden eggs...

Or China?

[aki009]

Re: SoftBank bought a goose that lays golden eggs...

There are quite a few core options that have been left by the wayside by ARM and its reasonable IP pricing. If ARM kicks up the fees, those guys will be dusted off faster than you can say buhbye-Softbank.

If Softbank keeps it up, then I'd say the cycle will be inflation in prices, drop in customers/volume/revenues, the emergence of a serious alternative for their bread&butter IP, and finally Softbank dumping a massively discounted ARM to someone else who hopefully knows what to do with it.

In the meantime we'll have to deal with a plethora of instruction sets and an uneven set of development tools. But thankfully the tools are much better suited for that scenario today than they were in the past.

[c1ue]

Re: SoftBank bought a goose that lays golden eggs...

Multiple somebodies hosed up that transaction. On the sell-side, it is understandable: the ubiquity of ARM in mobile conveys the impression of monopoly market presence. However, the value of ARM is primarily that it is cheaper and easier to buy than to build your own.

The "market dominance" is a function of realistic pricing; higher pricing tilts the equation to build your own or use another.

[Doctor Syntax]

I suppose the inflexibility of WeWork and Uber revenue is obvious even to beancounters, Arm less so. I wonder if they've considered the possibility that having had a lot of money to splash around might have made them a tad too careless and that if they want to find out who's responsible for poor ROI they should look for a mirror.

On an entirely disconnected topic do we have any idea when the HP verdict is due?

[Jellied Eel]

6.4bn chips, $1.8bn in licence fees & royalties.. $32bn seems reasonable..

Or not. WeWork & Uber seemed more reasonable as they were more of a punt into disruptive models, with potential for big earn-outs. Hence all the hype, until that started generating tricky questions like 'when will you be profitable?', and 'how are you burning so much cash?'. And I guess some interesting cultural exchanges, ie traditional Japanese conservativism vs Silicon Valley exuberance. Never mind the P&L, look at the PPT showing spectacular growth!

Arm should have been a bit more obvious given market maturity, competition and revenues being relatively inelastic. So like the article says, customers keep using Arm until the cost makes them look for alternatives.. Which is business 101. But then so is valuation, like traditional N x Revenue, and Softbank isn't alone in re-imagining company valuations. See Tesla for more info. I'm guessing their decision was just based on device count, and an assumption Arm's customers could be squeezed into paying $1/device royalty... And I'm guessing Softbank didn't ask Arm's customers whether they'd be willing to pay that.

[Anonymous Coward]

IMHO $32B is not reasonable

Being an old git I was always taught to look for a 10 year payback on any business investment, so $18B would be reasonable, not nearly twice that.

Not as bad as HP but still a spectacular failure.

[BebopWeBop]

Re: IMHO $32B is not reasonable

Warren Buffet, another 'old git' takes a similar approach. With ups and downs, it has not done him or his investors a lot of harm.

[Jamie Jones]

Re: IMHO $32B is not reasonable

Maybe it's Warren Buffet you're replying to!

[ckm5]

Re: IMHO $32B is not reasonable

That's not really how investors think. It's all about P/E ration and, at 16x, it's actually reasonable for a software company, which ranges from 10 to 20....

[gs4avs]

Re: IMHO $32B is not reasonable

16x P/E is not unreasonable, agreed. What is unreasonable in this case is that SoftBank paid $32B which is 16x of sales not earnings.

[Peter2]

That was my first thought honestly. If they were expecting to make a quick buck then they shouldn't have bought a company for twenty years worth of its yearly revenue and ~64 times it's current yearly profits.

I frequently question these days if a lot of these people have any better idea what they are doing and should be doing than I have. :/

Next mistake; sell off a profitable company that will eventually make back the money you put into it to raise "working capital" to throw more money into WeWork & Uber, neither of which have ever made a profit.

Nobody else is going to buy ARM for anything like what they've paid for it if they have any sense, which is going to leave them having lost so much money that they would have likely been better withdrawing it in low domination notes and then burning it to heat their offices.

[aki009]

At a buck a pop I'd say there'd be quite a few alternative architectures that'd make their presence felt. I particularly liked the AVR32 from Atmel that's dead now, but not forgotten.

[Henry Wertz 1]

Careful!

Careful! I'm a big fan of ARM, the low power ones are very low power, the fastest ones are quite fast while still having excellent performance per watt. BUT, for a lot of devices it's just a matter of needing some kind of CPU that runs Linux (usually, but WindRiver or VXWorks or something possibly.) MIPS is kind of dreary (usually not even an FPU) but plenty of access points and such have them, RISCV is completely free and has a full Linux and toolchain (GCC etc.) support. Apparently (at higher power points) the POWER CPUs are coming back into their own.

One of the things that has kept ARM successful for so long (Qualcomm does the same thing really) is to make sure to keep their various prices and fees high enough so they are rolling in the dough, but low enough so it's not seriously worth redesigning existing products to switch CPUs, probably not worth looking into other CPU designs for future designs either, or god forbid worth it for them to roll out their own CPU designs to potentially compete with ARM.

[Martin an gof]

Re: Careful!

Far be it from me to write a "me too" post, but I came here to say pretty much the same thing. I have followed the progress of ARM since the Acorn days (being the owner of an Archimedes) and felt proud to have been a part of something that the UK can claim has been a roaring success rather than "invented here, monetised elsewhere" as so often seems to be the case.

I had thought that following such farces as the Cadbury sale, the UK government was beginning to understand that raw capitalism wasn't always in the country's best interests, but the sale of ARM to Softbank seemed to prove otherwise.

I'm currently working with ARM M0 processors in the BBC micro:bit and some Adafruit Feather devices, and having previously worked with the older Atmel-based Arduinos, the extra "elbow room" is fantastic (compare the price and performance of a "Mega 256" with a "Feather M0"). It would be a shame if the micro:bit increases in price. I hope that rather than squeezing the golden goose (as someone else has said), ARM can find a new owner who understands the business and gives it the freedom to carry on in the way that has been successful until now.

Not that I don't want competition! RISC-V is very interesting, but as you point out, one thing that no other vendor seems to be able to do (yet) is to produce largely compatible devices across such a huge range of price and performance.

M.

[Anonymous Coward]

Re: Careful!

Honestly ARM is not a great investment from an ROI perspective with current expectations from investors (this includes anyone with a pension/retirement account). The premium paid by Softbank is a bit too high.

So the best thing is to probably for Sotbank to float it back on the stock market?

[whitepines]

Re: Careful!

Apparently (at higher power points) the POWER CPUs are coming back into their own.

And they're just as free as RISC-V now that the ISA was made open some time back. I have a Power desktop and it's nice to be using an open ISA without sacrificing (too much) speed / usability over Intel/AMD. Of course I have been using Linux since the stone age, so probably was an easier transition than some others.

[Anonymous Coward]

Typical Myopic story

Ive always maintained that any mega multibillion pound/dollar merger/takeover is NEVER EVER in the interests of the consumer. They end up paying for such indulgences. Return on investments demands (at premium prices) is thus a one way price rise street, for consumers.

It is solely engineered (pun intended) by the stockbrokers for the benefit of themselves & shareholders, by pandering to the egos of CEOs, who have spare cash OR lenders who are happy to leverage such deals and earn fat commisions. Cash up now and debt fill the company balance sheets ! Its a game played over and over again in many scenarious worldwide, with many examples.

Witness VirginMedia takeover by Liberty Global. How many price rises since takeover? Boots prices after takeover? eg. current own brand hayfever generic 7 tablets being sold for £4.95 in their stores, when you could buy 14 generic tables in the next door poundland for £1. AOL / Time merger. What happened? Dixons/CPW ! AFter collapse of Comet, they had no high street competition, YET? The greed has killed the golden goose as mentioned in above posts.

Games played by big boys.eh

[Anonymous Coward]

Re: Typical Myopic story

I'm not sure that I've ever seen it to the advantage of the company that has been taken over, either.

[Anonymous Coward]

Re: Typical Myopic story

Businesses that get acquired are of two kinds

Those that are dying themselves - so a shit show after isn't a surprise. All your examples are in this category.

Those being acquired for a ridiculous multiple such as this ARM acquisition.

Or put another way, a company with a good product at a fair price isn't going to be up for sale without the ridiculous multiple.

Which I suppose is at it should be? The dying business would go bust sooner.

Let market competiton deal with prices - just like walking to poundland did for the hayfever tablets.

The consumer benefits - via pension pot valuation and maybe transaction stamp duty?

Asset stripping and LBO are the things I think should be stopped - here it is neither yet.

The positive of Softbank is that ARM went on a *massive* hiring spree after the acquisition - it seemed fishy (I know some terrible political types that got hired). It looks like Softbank don't know what they are buying with their yen carry trade.

[Anonymous Coward]

Re: Typical Myopic story

The massive hiring spree was followed by a somewhat less massive firing spree, in certain areas of the company. I know a fair few people who thought they were on to a good thing but were shown the door because ARM hadn't realised that they were actually cannibalising their customers' own business models.

[Marco van de Voort]

MIPS

Risc V might be fashionable and the "open" community's darling, but there is only one decent source for production level silicon exists , SiFive.

MIPS is relatively cheaper, and more regular and used in great variety, used in the past in many routers, game computers, settopboxes and even microcontrollers using it exist (e.g. PIC32).

I'd at least expect the various ARM license fugitives to be divided over the two.

[diodesign]

Re: MIPS

I think in all honesty MIPS is looking a bit dead in terms of future generations using it. Sure it's still out there in a load of chipsets...

...but university courses are teaching RISC-V as an example CPU architecture now as it's pretty close to MIPS and all open-source and royalty-free yadda, yadda. In fact, the Berkeley team came up with RISC-V because they didn't want to teach MIPS internals any more, Arm wouldn't provide an academic core they could pull apart in an open way, so they made their own.

If you want to go beyond FPGAs and simulations, RISC-V does have a downside in that you're fairly limited for SoC suppliers. SiFive, Greenwaves, Microchip, Kendryte and NXP spring to mind, though.

C.

[John Sturdy]

Re: MIPS

I get the impression that MIPS, in the form of Loongson, is alive and well in China.

[DavCrav]

Re: MIPS

"Arm wouldn't provide an academic core they could pull apart in an open way, so they made their own."

Didn't anyone learn anything from Microsoft?

[Zippy´s Sausage Factory]

Re: MIPS

Didn't anyone learn anything from Microsoft?

Other than how to play Minesweeper, do you mean?

[Anonymous Coward]

Re: MIPS

Trump with his Huawei ban certainly didn't!

[Anonymous Coward]

Re: MIPS

> RISC-V does have a downside in that you're fairly limited for SoC suppliers.

As in none. SiFive made a limited number of development boards in 2018 - the SiFive Unleashed. They are sold out, no longer available, and they were ridiculously expensive to begin with: USD $3000 for a development board.

Combined with the fact that the RISCV spec is (a) unfinished and (b) incomplete. And that no chipmaker has announced an intention of fab-ing a RISCV chip. And that no-one has announced an intention of making a RISCV board.

You can boot Fedora 32 RISCV64 in QEMU. That's about it. It's barely usable, even by QEMU virtualization performance expectations.

So yeah, it's nice to talk about RISCV as an alternative to ARM, but the reality is that there is no RISCV. It's just a collection of PDF's.

[Jason Bloomberg]

Re: MIPS

The reality is that there is no RISCV. It's just a collection of PDF's.

Not true at at all. I actually have a $5 commercial RISC-V chip on a breakout board ready to use.

There are quite a few silicon implementations out there and available commercially. I even recall some sort of Arduino clone based around RISC-V.

True; things have not reached industrial scale yet but likely will given the hype it has managed to generate, pressures on countries not to be reliant on foreign entities, and higher license fees and royalties for ARM will help.

[Anonymous Coward]

Re: MIPS

Yeah, OK. $5 board. And Arduino. Thanks.

The Amazing Blinking LED Show, powered by Arduino RISCV.

Is the difference between "someone makes a $5 RISCV Arduino" and a board with a RISCV chip that can be used to build a PC that can boot a usable Linux distro, and can be used for kernel or compiler development, really that obscure?

No-one is going to be able to develop and/or test a RISCV compiler backend or the Linux kernel on a $5 Arduino board. If you don't have a compiler, you can't have a kernel, or a distro. It's a catch-22.

I built LLVM on Fedora 32 RISCV running in QEMU, just for kicks and giggles. I gave QEMU 128GB RAM and 8 RISCV CPU's. It took 9 days with gmake -j4. Does that sound like a usable development environment?

[sw guy]

Re: MIPS

Running a PC is not the sole purpose CPUs, you know it.

I have participated in setting specifications of several SoC with various CPUs on they, and then writing firmware while other engineers actually created the chip. In every case, the development platform was a x86 PC either on Windows or Linux.

[Martin an gof]

Re: MIPS

Running a PC is not the sole purpose CPUs, you know it.

And the key here is that ARM is possibly (I don't really know) the only CPU designer which provides devices suitable for everything from extremely low power / basic performance up to devices suitable for desktops and servers which share (subsets of) the same instruction set and similar architectures such that swapping one part for another doesn't mean a complete redesign.

And they have the stuff now, and it works, and the licencing means that there's usually something affordable for everyone. I gather that the basic M0+ core can reach around 1 Dhrystone MIP per MHz at around 4µW per MHz in a die of 0.007mm² - that's a square with sides eight hundredths of a millimeter long. Ignoring the stuff required to make an actually usable system you could potentially fit over a hundred of these cores on a sliver of Silicon just 1mm on each side. The M0+ runs the Arm v6, Thumb and Thumb 2 instruction sets.

At the other end of the scale the sky seems to be the limit with partners producing ARM-based processors having performance characteristics similar to x86 devices but at a fraction of the power budget - this is why Apple seem to be moving away from Intel and towards ARM. Yet even the latest Arm v8 processors have largely backwards-compatible instruction sets, and most can execute the exact same Thumb 2 instructions that the M0+ uses.

Can anyone else offer that breadth of product?

M.

[martinusher]

Re: MIPS

>So yeah, it's nice to talk about RISCV as an alternative to ARM, but the reality is that there is no RISCV. It's just a collection of PDF's.

The RISC-V is just a rationalized version of the common 32 * 32 bit register RISC. I've used these as a MIPS processor and soft processors, IP blocks for FPGAs. The RISC-V hasn't made too much of an impact yet only because there's been no compelling need but switching to it will be no different from moving from a Microblaze to a Micos, a 'meh' sort of project (...speaking from experience.....)

The ARM is common for the same reason that Intel's x86 architecture is common. It may not be the best but its common, well understood and widely supported. People won't move without a reason so its in the vendors' interests not to give people that reason. Back pricing the ARM licenses -- adjusting them to provide an expected yield rather than maintaining a competitive presence -- will work in the short term because of the costs involved in shifting architectures but its never a good long term strategy. Its tempting to think that because you're the best that customers have no choice but unfortunately if there's one thing customers hate is being stuck without competitive options -- and there are always hungry vendors circling waiting for their opportunity to strike.

(Incidentally the Microblaze/Micos is a living example of this problem. We have a long running product line built around Xilinx FPGAs. We needed to upgrade to newer parts but Xilinx's pricing options were not very attractive. We were locked in, it seemed. Lattice was hungry for the business.....they didn't have the big chips and so on but for our products we don't need $2K FPGAs, we're looking at a few dollars each. Neglect the commodity product at your peril!)