Hacker News new | past | comments | ask | show | jobs | submit login
SoftBank Group Nears Deal to Buy ARM Holdings (nytimes.com)
357 points by anyfoo on July 18, 2016 | hide | past | web | favorite | 98 comments

This is the opportunistic genius of Masayoshi "Masa" Son as one of the most successful technology investors of our time. He's taking advantage of the weak GBP thanks to the whole Brexit fiasco. As a major carrier in Japan and an owner of Sprint, this move makes total sense.

Since ARM had been a mainstream acquisition target for years now, I'd suspect this owes more to their M&A team rather than their CEO.

Estimating ARM's future earnings is relatively straightforward, and Softbank, like Berkshire, loves free cash flow. It doesn't require a visionary to go through the calculus of an acquisition under these parameters; it requires a team of shrewd finance professionals.

Intel doesn't exactly shy away from purchases either; is it likely that the only reason they haven't made an offer anti-trust regulations?

Intel has already sold off their successful line of ARM chips ages ago and instead focused on domination via the x86 architecture (Which only Intel and AMD are licensed to make).

I'm not sure Intel would want to also take full control of ARM chips regardless of Antitrust regulations.

Whoever buys ARM can keep collecting the licensing fees, perhaps even increase the fees a bit. No platform is particularly tied to ARM in the long run, they just use it because ARM chips are cheap, fast and power efficient. If Intel tried to change the ARM licensing deals in monopolistic ways, then everyone would simply move to new chips. For example, Android has always been reasonably platform agnostic with their Java layer and Apple have been introducing platform independant binaries.

Wouldn't be allowed, right? That would position Intel as absolute monopolist on desktop, server and mobile CPU.

Intel CPU cost already way to much due to lacking competition from the only left x64 vendor AMD which is years behind. And Intel slowed down the release of faster CPUs. Where are the 8 and 16 core CPUs? Where are the 5+ GHz CPUs? Why hasn't the single core performance advanced that much (if you discount the advance of memory speed) since 2004 (Pentium 4 era).

Are you sure about the single thread performance? Because the jump from Pentium 3 to Pentium 4 was not that large, but the jump from Pentium 4 to Core 2 Duo (2006) was. All the jumps after that (Nehalem etc.) were larger than the jumps we have nowdays, and the jump to Sandy Bride (2011) was really large. Only since then has the single thread performance stagnated.

Cinebench R10 32 bit single thread scores:

  Intel Pentium 4 660    - 3.60 Ghz - 2245 Points
  AMD Athlon X2 6000+    - 3.00 Ghz - 2848 Points
  Intel Core2 Duo E8400  - 3.00 Ghz - 3647 Points
  AMD Phenom II X6 1075T - 3.00 Ghz - 3820 Points
  AMD FX 8320E           - 3.20 Ghz - 4056 Points
  Intel Core i7 950      - 3.06 Ghz - 4353 Points
  Intel Core i7 3930     - 3.20 Ghz - 6268 Points 
  Intel Core i7 5960X    - 3.00 Ghz - 6950 Points

I'm not sure that cinebench is a good single thread benchmark because its basically a benchmark of how many vector ops you can run in parallel. But using it allows people to print pretty charts like you have there, as will benchmarks doing AES encryption, or any other "special purpose" instructions that have been added to the CPU in recent history.

A better traditional single thread benchmarks might be spec int base minus libquantum. That tests things like how good the branch predictors are relative to pipeline depth, how fast dependent instructions or pointer chasing can be handled, etc.

Yes, I was not really sure how representable Cinebench is, hence my question how sure the parent is. But the benchmarks were easy to come by (back when I posted them already a while back). The only other benchmark I could find that had CPUs from Pentium 4 up to Haswell was Geekbench.

Would love to see someone do a benchmark like that (would do it myself if I had enough spare money), with equal clock and memory bandwith etc.

EDIT: after reading the critique about libquantum on wikipedia I'm even more interested in the benchmarks.

"In the SpecInt2006 benchmarks, the 462.libquantum benchmark is highly vectorizable. The baseline computer for all benchmarks is a 1997 Sun Ultrasparc server computer. Whereas most of the spec sub-benchmarks turn in a performance improvement of about 5x to 80x times faster than the Ultrasparc, the particular 462.libquantum sub-benchmark turns in a result that is up to 4082 times faster than the Sun Ultrasparc .[3] This suggests that for this sub-benchmark, most of the improvements over the Ultrasparc are due to vectorizing compiler improvements, NOT due to CPU hardware improvements, since 1997."

Intel already sells 18-core processors, granted they are server chips, not desktop chips.

48-core ARM processors already exist[1] but via the vendor Cavium. The takeaway is that multi-core is continuing very much so.

edit: I just looked; Intel has Haswell-based server processors with 24 physical cores[2]

[1] http://www.cavium.com/newsevents_Cavium_Introduces_ThunderX_...

[2] http://ark.intel.com/products/93790/Intel-Xeon-Processor-E7-...

> Why hasn't the single core performance advanced that much (if you discount the advance of memory speed) since 2004 (Pentium 4 era).

It has, but you have to use AVX and other new instruction sets. Singlethreaded, a fifth-gen Core m5 tablet CPU can be as fast as a first-gen Core i Xeon workstation CPU, and a sixth gen i7 desktop CPU can be twice as fast as both: https://admin.tao.at/post/evolution-of-cpu-efficiency/

(Sadly I don't have any Pentium 4s left to compare on. And of the Core2 generation only laptops.)

It seems like a lot of new tech went into AVX and increasing both width and number of vector registers. Kind of makes you doubt whether single-threaded performance is just a mask for wider execution pipe underneath.

Buying ARM wouldn't help Intel, that's why.

I assume you ask this question because ARM is absolutely killing Intel on mobile CPUs. Worse, Intel has tried to get into that space and ARM has handed Intel its head more than once. So this is not a dumb question.

But ARM makes a trivially small amount per CPU manufactured, making it up on volume. Intel takes a large amount of both revenue and profit on each piece silicon that goes out. And you can see the result: Intel's market cap is still 5X ARM's ($160 Bn to 30 Bn).

As an individual investor ARM is (was -- assuming this buyout goes through) probably a better investment than Intel since Intel is struggling to find new markets to expand into, with most of its revenue coming from a dead end business (PCs) while ARM is being pulled by many people into new, growth businesses. But because of its business model, an ARM acquisition would dilute Intel's results (its margin would drop).

For smaller companies it's OK to "eat your own lunch." This was in fact Sun's principle (which didn't work out for them so well) and Apple's (which has, so far). But when you get to be huge, it's hard to move the needle in a way that appeals to investors. Look at Apple: they have a crummy iPad business that, were it standalone, would be one of the phenomenal businesses of the decade. But when mixed in with the rest of Apple, it's simply blah.


Separately, I remain astonished that Intel killed its StrongARM business line. They definitely had the best ARM on the market and it played, I think, to their strengths. I think they were just afraid that it attacked their brand, which at the time was very expensive CPUs. But they could be dominating the mobile market now.

> As an individual investor ARM is (was -- assuming this buyout goes through) probably a better investment than Intel since Intel is struggling to find new markets to expand.

Well, ARM doesn't have those pesky chip fabs on their balance sheets, that's the licensee's problem. Accordingly, they have a profit margin of 30 percent, compared to INTC's 20.

Why is it Anti-trust? Intel makes chips and ARM does not make chips.

Control of the intellectual property. Intel already controls what's in most desktops since Apple ditched PowerPC. No competition there across the globe. Tablets and mobiles were the challengers eating into that market share. ARM dominates there with its I.P. in almost everything that's mainstream with over 10 major companies licensing the I.P. for their own chips. If Intel owned that, they'd control the I.P. behind most of their rivals. Finally, there's the server situation where, although more ISA competition, x86 still dominates.

Letting Intel have ARM would certainly be an antitrust situation given the sheer dominance Intel's I.P. would have at that point.

But they both design chips. And combined, they design most of the chips in the PC, server, mobile and tablet market.

    > He's taking advantage of the weak GBP thanks to the whole Brexit fiasco.
More likely: SoftBank Group is taking advantage of its ability to issue debt at historically low interest rates, thanks to the monetary policy of the Bank of Japan.

SoftBank's net-debt-to-equity ratio has risen from 0.6 to 3.5 from FY2011 to 2015: http://www.softbank.jp/en/corp/irinfo/financials/results/hig....

That being said, SoftBank purchased ARM using GBP (I assume), so they would need a way to convert their cheaply acquired JPY to GBP. I wonder if they are able to purchase foreign companies for JPY at, say, a 10% markup. The sellers would be able to sell this JPY on the market as soon as they receive it, and SoftCorp would be left with a brand new company in its portfolio and some JPY liabilities (they are issuing unsecured, subordinated (5-year?) JPY bonds at 2.5% p.a.: http://www.softbank.jp/en/corp/irinfo/stock/bond/).

FYI, much of ARM is held via ADRs (American Depositary Receipts) denominated in USD. This is how ARM shares trade on NASDAQ (under the ticker ARMH) in addition to trading on LSE.

Brexit didn't make ARM cheaper. On referendum day it was 1567 yen/share. On Friday was 1670 yen.


The referendum was announced in early Feb 2016 and markets started to move then, http://www.xe.com/currencycharts/?from=JPY&to=GBP&view=1M. See that step that looks like the half-profile of Table Mountain, that's referendum day.

It looks like that although the GBP has declined against the Yen over the last year (and dramatically since the EU vote) [1], ARM's share price has increased to more than counteract this (again dramatically since the EU vote) [2].

[1]: http://www.bbc.co.uk/news/business/market_data/currency/11/1...

[2]: http://www.bbc.co.uk/news/business/market_data/shares/3/8729...

Yes but ARM shares are up since Brexit, more than offsetting the GBP loss.

Even more so when you look at the Yen vs GBP in particular. The Yen is up ~40% against the GBP in the last year.

Value of ARM should stay the same when GBP goes down (as viewed in yen or USD) since it's selling to the global market. Or maybe go up since ARM's UK expenses are now lower. It's the pound that's depreciating, not ARM.

Good point, ARM's revenue in the form of licensing fees and royalties is denominated in Dollars.

You think the deal was started, and nearly completed in a few weeks? Surely the deal had been going on for a while longer than brexit turmoil

It's quite possible to do these deals in short time.

For one, the young investment bankers and lawyers are accustomed to staying up late when there's a deal on. I used to live with one, and it was simply part of the job that the job was everything, all the time. Staying up an unhealthy amount is what they expect.

Also, this isn't a completely out-of-the-blue deal. Everyone in the M&A space would have given some thought as to which firms would potentially merge, and they would have made presentations to the bosses of the involved firms airing the pros and cons, the possible financials, the antitrust side, and various other issues.

It seems like SoftBank is very willing to do deals with very little due diligence (based on some, uh, experience).

In this case, at this size, my guess is they've been keeping tabs on ARM for a long time and maybe decided to just pull the trigger with Brexit going on.

The BBC it was done in two weeks.

Wanted to mention this elsewhere but can't find a reference, do you have one?

Paywall :( But thanks!

They might've been planning and negotiating the deal as a 'just-in-case' far ahead of the vote. It's not like people didn't know the pound was going to plummet in the event Leave won.

This makes me wonder when did the acquisition talks begin.

Don't forget a stronger yen which could face enormous volatility in the future.

Apparently Arm's Founder isn't particularly into it:

"ARM is the proudest achievement of my life. The proposed sale to SoftBank is a sad day for me and for technology in Britain."


I once read an article where the CEO of ARM reasoned against raising prices because its partners would then have less healthy businesses. I never forgot that piece of wisdom.

But now with this acquisition, is there any room for such generous mentality?

It is not generosity. It is knowledge of your own business model.

ARM wants to earn 1penny on every chip sold, but to put chips everywhere. So increasing royalties, but preventing market expansion is bad. Especially with the internet of appliances rapidly approaching.

Not just that, it also prevents other companies from entering this market. Maybe forward thinking is a better term that generous.

I'm curious. Does ARM get royalties even on custom designed chips like apple's? Or only on reference cores?

> Does ARM get royalties even on custom designed chips like apple's?

Designing custom cores requires holding an architecture license which is ARM's highest-level license (beyond the perpetual multi-use — use a core in any product indefinitely — and the subscription — use any ARM product for a set duration). There are about a dozen architecture license holders.

That's standard supply chain theory. You will read it in books about the Toyota Way.

The FT claims that the deal has been done for £23.4bn.


>Japan’s SoftBank has agreed to acquire Arm Holdings, the UK’s preeminent technology company, for £23.4bn in an enormous bet by the Japanese telecoms group that the smartphone chip designer will make it a leader in one of the next big tech markets, the internet of things.

Apparently that is a 40% premium on market cap. [1]

[1] http://www.bbc.com/news/business-36822272

Does anyone know how (if) currency conversion is done with amounts this large? I imagine with the volatility of the GBP right now this could have been more expensive if the conversion was done a few weeks ago.

It's more likely that currency con version at this scale will be largely on paper accounting because 10s of billions of pound sterling is probably more than most foreign currency reserves of countries much bigger than SoftBank usually carry.

For example, a SoftBank subsidiary can probably take out huge loans from UK/international institutions and collateral/pay the loan with other assets in other jurisdictions. I believe this is how many companies avoid repatriation taxes: by taking out a loan in the US and making the payments to financial institutions overseas.

It's pretty easy - the nominal amounts for derivative trades can be extraordinarily high and currency markets are smaller but in the trillions of dollars per day.

They can hedge this (lock in the price in Yen or USD) now and be sure of not losing on currency shifts when it comes time to make the transaction work.

It's not hard to do an FX trade of this size. I've written programs that trade billions of FX a day, and that was for a teeny tiny fund.

For large amounts, you call the broker, which I suppose whatever investment bank is running the deal will have figured out how to do.

But 24B is a drop in the ocean.

> in an enormous bet

It's a lot of money but I'm not sure it's that much of a bet.

Maybe this explains why SoftBank recently chose to sell its controlling stake of Supercell for about $7.3B USD -- they needed the cash for something bigger.

SoftBank had purchased Supercell for about $2B only a few years ago, so it was a genius investment. Will be interesting to see if they can turn ARM into another one of those.

I did wonder why they would sell their Supercell stake considering they just introduced a new game which is looking to be another cash cow. This certainly does seem to put the sale into some perspective.

If they've just introduced a new game which demonstrates continued earning potential, then now is an excellent time to sell.

Uhh, what did they just introduce?

Clash Royale - one of the top grossing games of the past few months.

Clash Royale.

This could be great for RISC-V's prospects, especially if Softbank decides to be more partial to which customers it gives the license, or if it increases prices, and so on - basically anything Softbank might do in the future that would piss customers off, could be a win for RISC-V.

Don't expect any major shakeup over the next 3 years though. Even if some customers decided to drop ARM right now, it would still take 3+ years before they come up with a strong RISC-V alternative. First, Android would have to give RISC-V its blessing (support it), too, but Google is a member of the RISC-V foundation, so I assume that's not an impossible task.

Mixed feelings on this - huge congrats to ARM friends but I feel it's 1 act too short for an exit like this. I'd love to have seen a British owned tech company higher up the Global 2000 before transferring ownership.

Yes I too wish that there was a big British tech company I could work for in the UK (being British myself). Unfortunately I can't imagine ARM ever being competitive as an employer.

When I was applying for internships during my degree a couple of years ago I looked at US companies like Google and Oracle and ARM in the UK. I think ARM paid about a quarter of what Google and Oracle were offering. One quarter. And that's just cash before I included the housing, car costs and massive relocation grant that the US companies included and ARM didn't.

Today Glassdoor tells me that ARM pays interns about half, and graduates about 60%, of what Microsoft does, in the same city (Cambridge) a few miles apart.

But they have good people working for them so it must be enough I guess!

A decade or so ago, ARM was offering graduates about £25K. I believe they still offer the inflation-adjusted equivalent of that: ~£32K.

Does this mean that Microsoft Cambridge are offering graduates ~£50K? It's interesting that I consider that to be very high for a graduate salary yet, at USD 66K equivalent, it is significantly less than would be available in an area of the USA with a comparable cost of living...

Glassdoor says £70k to £90k for the same grade ('senior' but of course it's hard to compare).

That's about what I would expect. This is for PhD interns and grads I should have made more clear.

I'd say £32k is beyond taking the piss for someone with a CS undergraduate degree and having to live in Cambridge (you couldn't buy a house there on that) but if they're getting the people they want then it must be right for them.

>I'd say £32k is beyond taking the piss for someone with a CS undergraduate degree and having to live in Cambridge

Whilst it's not giant it's not super low either. I'd be surprised if you could find that many tech (that is not banking/consultancy) companies in Cambridge (or indeed London) offering signficantly more than that. There's also things like Bonus, Share Scheme and pension (private health insurance too but that's a pretty common perk) that needs to be brought in the comparison.

Don't Facebook and Google offer more than that in London to new graduates with just undergraduate degrees? I know those are outliers then, but it's a shame that the outliers are all US companies. And ARM isn't just some tech agency, they're working on hardware, simulation, compilers, etc, hard tech stuff.

I'm sure FB and Google offer more but they're not most tech companies!

London is also more expensive than Cambridge (I doubt you'll buy a house a short distance from the office in London on a mid-level engineers salary at Google/Facebook, you can happily do that in Cambridge at ARM).

I'd also note hard-tech stuff just seems to attract less money than web/social (see how the recent linked-in acquisition by MS wasn't for much less cash than the Softbank ARM deal).

I'm not saying ARM is super generous with their compensation compare to other tech giants, but you get a pretty reasonable deal, and it's a nice place work with interesting things to work on.

(I work at ARM)

>I'd say £32k is beyond taking the piss for someone with a CS undergraduate degree and having to live in Cambridge (you couldn't buy a house there on that) but if they're getting the people they want then it must be right for them. The average CS grad salaries in London are below that. But I agree, it is still terribly low.

What about Imagination?

I was surprised ARM hasn't been bought by a larger conglomerate yet. SoftBank is way ahead of the game.

The problem with ARM is that the most likely suitors don't dare buy them -- if Intel, Apple, Microsoft, or Google bought ARM Holdings, the others (all rivals) would start looking for an alternative architecture to run on and the licensing fees would take a nose dive.

It's one thing to license an architecture from an independent, scrupulously neutral third party: another entirely to gamble your business platform's future on a rival's goodwill.

Softbank isn't an obvious direct platform-level competitor for Intel or Apple or Samsung or the others, so may be an acceptable acquisitor who won't frighten ARM's licensees into abandoning ship.

That assumes someone would buy them as an ongoing business. Apple could have bought them specifically in order bring the ARM designers in to Apple and force their rivals to switch - an acquihire on a massive scale. Using an ARM designed chip is an advantage, and taking that advantage away from your competition could be a very good thing for your business.

I'm glad that didn't happen though.

Apple completely design their own chips. The real mystery is why Apple stuck with the ARM architecture when they switched to 64 bit (since aarch64 is a brand new design). My guess is because they still had to license the 32 bit architecture for compatibility, and because the amount that Apple pays to ARM each year for the license is peanuts.

For the same reason they haven't switched to ARM for macOS yet - compatibility. If Steve Jobs was still alive, he would've probably switched macOS to ARM by now, despite all the hassle. He's done it before. I'm not really seeing Tim Cook pull a move like that anytime soon.

Very much this. If this deal goes through, I have a feeling that we may see ARM begin to give way to RISC-V. Perhaps it might happen slowly at first, but the mobile space is already prepared for a switch given that Android is virtualized and Apple is switching to abstracted "bitcode."

An Apple switch is unlikely due to their significant in-house advantages being geared towards ARM directly. PA semi was an ARM shop and their innovation in their A$ line of chips is driven by significant experience working with the ARM ISA... They have a licence which gives them the freedom to design their own chips and unless the new owners decided to do something disastrous with respect to existing contracts, that license will remain valid and its terms unchanged. Something ephemeral to the facts of the existing contracts is unlikely to motivate Apple to do anything so drastic. They already have the ongoing Intel sky lake laptop debacle and a future of less and less certainty regarding Intel's promises to deal with.

Tldr: Apple are more likely to move macOS to ARM than iOS to RISC-V

I stand corrected. My memory failed me trying to compose a reply on my phone.

P.A. Semi's PWRficient processors were Power based, not ARM. They learned to adapt :-)

Financially the deal doesn't make sense. ARM has been around for a while their FCF ~500M a year. At $30 billion purchase price you are paying 60x FCF. Apple and Qcom just license the instruction set and design the processors themselves. To justify the $30 billion price they really need to increase FCF to 1.5 billion a year. For a relatively mature business I don't see how it is going to happen.

Mature in the mobile and current marketplace - as has been mentioned in the thread previously, the gamble they're taking is that ARM is going to dominate IoT. With a (potential) monumental increase in the number of connected devices, as well as updating older gen embedded systems, that's where they're aiming for that >3x growth to come from.

Maybe it's just me, but I see this as a foolish move. ARM was better suited to being an independent company due to their licencing model. With the purchase by SoftBank now there's greater incentive for licensees to move away from ARM. I suspect we'll see much more investment in RISC-V as a result of this, which would be a good thing for most of us, but a less good thing for ARM.

That depends on how Softbank manages the acquisition. If they keep the company separate and just keep cashing in dividends, I don't see the problem. Softbank afaik is also very much focused on the Japanese market only.

Foolish for the downstream integrator, maybe even the longevity of the platform, but shrewd for creditors and shareholders.

I'm laughing because integrators flocked from MIPS to ARM, and events could unfold such that they run back to MIPS etc

Apparently, $32 billion is how much they'll pay for this deal. [1]

[1] http://www.wsj.com/articles/softbank-agrees-to-buy-arm-holdi...

They bought Sprint before. It's costing them an ARM and a leg.

Finally, a hardware-related acquisition that is more expensive than the Whatsapp acquisition.

I'm kind of surprised Apple didn't swoop in and buy ARM long ago. I guess as others mentioned that might mean all the Apple competitors using ARM chips would flee and leave them with a lot less partners.

They seem happy to screw their suppliers right down instead and leave the risk to them - the sapphire screens saga being an example.

I wonder what SoftBank can add to ARM technologically?

Obviously besides swimming in dollar

They took 95% ownership of Aldebaran Robotic a while ago, and it's a huge success in Japan. I'm no expert but isn't that building a vertical?

AFAIR the robots are running on Intel Atom, but well... embedded... ARM... (not saying that's the only reason though)

huh, that sounds cool, energy-efficient factory robots, sounds like something the Japanese love

Isn't he already bankrupt due to the "wrong" acquisition of Sprint? This is hard to believe. I did notice he was preparing cash for something(selling Supercell, and Alibaba shares for cash), I thought those cash are used to cover Sprint's losses. Gosh this is really, really hard to believe.

Neat! now my company can no longer worry about FIVEEYES having control of the spec, or the company that makes it. probably the best news of the year for me.

First sentence: "SoftBank is nearing a deal to acquire ARM Holdings, the British semiconductor company, said two people briefed on the matter who asked not to be named discussing private information."

Editing standards at the Grey Lady have declined.

I find that sentence grammatically correct. What is it do you think editing has missed?

Disclaimer: I'm not related to the BBC

Maybe I'm not thinking straight because I'm up to my ears in ARM chips (my cell phone, tablet, half a dozen homemade gizmos, etc.). And I know they don't actually make the chips themselves, but GBP 2.2e+09 doesn't seem like a lot for a technology that has so much influence.

Correction: 2.2e10, as noted. Now I know I'm not thinking straight.

It's 2.2e+10

It's 40% above the current market cap.

SoftBank WiFi access points are the most user unfriendly on the planet. You have to call them to be able to register to get online. And guess what? If you are not a Japanese resident you can't get a Japanese phone number. And even before registering the WiFi craps out 80% of the time within 60 seconds of connecting, no matter where you are in Tokyo. I am extremely disappointed. SoftBank will shake you down for all you are worth.

Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact