Electronics Weekly Electronics Design & Components Tech News
On Wednesday Arm sacked the CEO of Arm China, Allen Wu. On Thursday Arm China said Wu was still CEO.
Arm had replaced Wu with two execs but yesterday Arm China claimed one of them was sacked last month.
It’s the latest folly in the saga of Arm beginning with the UK government agreeing the sale of the company to Japan’s Softbank in 2016, succeeded by the folly of selling 51% of Arm’s China operation to Arm China in 2018 at a time when problems were brewing with the tech trade with China.
Ultimately the boss of SoftBank, Masa Son, can determine the outcome of any dispute between China and Arm.
But the underpinning of Son’s increasingly debt-threatened business is its shareholding in China’s Alibaba worth $125 billion.
Like everything else in China, the government decides if Son can liquidate all or part of this holding and if the China government says No, then the value of the stake as a guarantee of Softbank’s debts is diminished or eliminated.
So China holds all the cards.
I think we kept the flag flying Mike, the secret IMHO was not to spend too long with the Russkies – they could go all night,
Thank you DontAgree, that’s most kind and much appreciated
The blog was based on the sort of conversation one has round the bar of a conference hotel sometime after about 10:30pm, Mike, so one expects and accepts the odd insult as all part of the give and take.
Compared to the rest of the online world, these comments are squeaky clean … ok that is not saying much
Anyway, thank you both for your enlightening discussion.
I recall it used to be that by midnight it was often just me, you and a couple of others keeping the bar staff busy. Days long gone now 🙁
I was talking about UMC Mike, not Apple. And I have noticed that you’re not unfailingly polite to me. Nor, actually, would I want you to be.
Okay – that’s the problem with your replies not being threaded – I assumed you were replying to a different comment. But yes, by the A.T. Kearney principle and Malcolm’s analysis UMC are headed for the deep end of the pool as well.
And I don’t believe I am rude to you, or at least not trying to be, just trying to correct incorrect statements. But I’ll try to be more diplomatic in future. The only people I am deliberately rude to are unpatriotic Britons who I really wish would go and live elsewhere and give up their UK passports once and for all.
If you think a company which has lasted 40 years, has annual revenues of $5bn and is the fourth largest in its industry, as ‘not a major success’, then I increasingly wonder what planet you’re currently inhabiting, Mike.
I assume you mean Apple, which as you say is fourth biggest in its industry. But most industries except automotive have proven over time to only have room for three players – see your copies of Malcolm’s IFS slidesets for lots more details – so in fact it could be argued that Apple may well be on the long slow path out.
But anyway the fact that the first, second and third players don’t use their own baseband processors and make phones in much higher volumes indicates that making your own baseband processor is a minority event at most so your statement ‘I thought the big users of baseband processors make their own’ is definitely inaccurate. And if you want to do it you have to pay so much for a licence is it really worth doing anyway ? Apple has to use Qualcomm chips in one of its next phones as per the legal ruling so in fact none of the big 4 will be using their own baseband chips for a while, presumably with the introduction of the next iPhone.
And no need to be rude with your last statement either. That’s not professional.
I thought the big users of baseband processors make their own, Mike
Not back when you were talking about. And even now Samsung has given up and gone back to using Qualcomm, whilst most others never left them. It’s mainly Apple who make their own and they had to settle a court case by paying Qualcomm $4.5bn and agreeing to use Qualcomm ICs in some future products.
Well Pepe, IMHO Wintel so appalled the industry in the PC market that I think the industry decided never again to let a single-sourced processor dominate an industry sector – hence ARM when mobile came along – and then along came datacentre with x86 dominating!
So we let a single company dominate the far more lucrative baseband processing slot instead 🙂
ITRON and TRON were both part of the same over-arching strategy Mike – to have an architecture, an OS and ICs to disseminate a standard microprocessor across the digital universe.
That was the original intent but it was fast overtaken by the reality that Japan couldn’t make high end processors, but had all the world’s leading automotive and robotic manufacturers who needed a standard OS long before WindRiver came along. Hence it was ported to all sorts of processor – I think SPARC was one of the few it wasn’t ported to.
If Philips hadn’t contributed the process, ITRI might have done, Duncan, ITRI had been refining that CMOS process it bought from RCA in 1976 and it was leading edge in 1987. After all UMC was founded in 1980 with a process contributed by ITRI.
I think that would have failed as I wouldn’t regard UMC as a major success story. Remember we are talking 3um for the Philips process and 2um for the ITRI process when the Japanese were introducing 1.2um, so the ITRI process most definitely wasn’t leading edge, far from it.
But the 3um Philips process used less vertical height than the ITRI process (which Morris had managed of course so knew all its problems) or most of the Japanese processes. TSMC shrunk the process to 2um in 1988 which combined with a good metalisation strategy (can’t remember if 2 or 3 layers) and the introduction of the Tancell EDA tool which the Japanese were slow in adopting made medium volume high NRE full custom ASICs more cost competitive with the low NRE gate arrays the Japanese were pushing at the time.
However the big market then was PC chipsets and market leader Chips and Technologies were getting their ICs made at Hitachi on a process more advanced then anything at TSMC, possibly 0.8um but definitely better than 1.2um. But TSMC got some IP blocks designed which allowed others to enter this market and the rest is history.
Well Pepe there was good reason. Tron had onsiderable success in the industrial market and was moving into the computing market. Ken Sakamura billed it as a universal processing architecture for every market – a bit like Arm does today as ‘the architecture for the digital world’ – Tron could have been Arm 20 years before Arm. MS- with its x86 buddy – didn’t want the world going that way!
You’re mixing up TRON and ITRON here. The Industrial TRON OS (ITRON) was a huge success but was purely software that could run on any microprocessor, not just G-Micro. Ports to x86 and 680×0 were extremely popular, as were ports to run on the internal architectures at Hitachi and Mitsubishi. It was also the first OS for automotive use running on most Japanese architectures as each car manufacturer had a pet semiconductor supplier.
Very interesting the TRON project. I feel the Microsoft has been in the wrong side history since the beginning, but they clearly along side Intel underestimated ARM. because once ARM got their hold in the mobile market Intel and Microsoft are now nowhere to be seem in the mobile market. I remember buying some pocket PCs for a costumer, but never again.
It’s all catch up catch up catch up Mike and has been for decades.The VCs always used to want something 100x better than what’s available in the market if they were going to back it. You have to shoot for something in the future – not stumble along in the wake.
Japan, Taiwan and Korea never used VCs. They were all government funded in same way as China. And as you’ve pointed out below, there are strong parallels between China now and Japan in the 80s.
If Tron was as bad as you say it was, Mike, why did Microsoft pressure the US government to force the Japanese government to drop the project?
To be fair Americans overreact to everything.
this is a excerpt from Wikipedia.
“from the point of view of the United States trade officials in the 80s and early 90s, the TRON project was seen as a plot brought up by the Japanese bureaucrats to “control the world””
Its really sounds similar to today rhetoric.
Probably because Microsoft weren’t a hardware company back then. Intel were far more worried about the 188 and 386 clones Fujitsu produced at the same time as G-Micro (TRON) as the 386 outperformed the Intel version and the 188 could be integrated with Sea of Gate ASICs for custom telecomms controllers which were a big Intel market back then pre the Motorola QUICC controllers.
And as Pepe says, Americans do tend to spout a lot about processors. Some even think ARM processors are designed in the US !
I think you’re absolutely right Pepe, for decades the Chinese have tried to build a domestic tech industry the easy way – copying, buying, stealing – instead of mastering all the relevant technologies like the Japanese, Taiwanese and Koreans did. Where I differ with you is that I think this rather slovenly approach of the Chinese has failed. They don’t show up with significant market share in any of the technically demanding markets.
Japan, Taiwan and Korea also copied, bought and sometimes stole technologies, as indeed did most US or European startups. Nothing unique to China – that’s how the semiconductor industry has always operated !
And isn’t r.f. (wireless) technically demanding ? Many of the best chips for this area are Chinese with over 50% market share in many cases.
Where would TSMC be if Philips hadn’t taught them to make IC?
Even back in the 1970s, Sharp licenced Rockwell Techniques to make calculator ICs.
If China had the same inventiveness and dedication as Japan they’d have developed a domestic core as good as Arm and RISC-V by now, Pepe, like Japan did with TRON.
Chinese companies have created some interesting architectures in the past like this one http://icubecorp.com/ but they failed to get traction, because using establish architectures like ARM is easier and less riskier, but those risks could be attenuated with a free open source ISA.
I really wish they embrace more open source because that is the biggest selling point of RISC-V,
That being say from my perspective the big problem with the Chinese is not that they lack “inventiveness and dedication” they are pretty dedicated and inventive. Ironically the problem with the Chinese is that they are cutthroat capitalists, the issue is to make money fast and with the least investment possible, they will always go for the easier route and become complacent. That is one of the reason why the government money has failed in creating dominant players in China the easier route is to buy establish technologies, why wasting time, money and effort in creating a reliable local supply chain when you can just grab the low hanging fruit.
But don’t be fooled by that because when they have enough pressure they can work like ants and they can archive goals quite fast.
Hi Pepe. I have a sample quad-core RISC-V SoC here which was Chinese designed. I don’t think it’s on the open market yet but it performs almost as well as the ARM-A53 quad core in a Pi-4.
So the Chinese are definitely embracing open source. But whether that SoC will appear on the Western market is hard to tell. Many Chinese ICs only have Chinese datasheets so it’s a struggle to use them unless you have a native Chinese speaker to help you as I have.
I agree Mike language is a big barrier in the adoption on Chinese ICs, i have found some nice cheap components LSCS electronics but with the datasheet in Chinese is an extra step to have translate. Anlogic fpga are another example, with everything in Chinese is somehow hard to work with their boards.
The language barrier is also a problem even with open source projects like the PaddlePaddle AI framework which can decrease some adoption outside of China.
But there are also Chinese companies with excellent internationalization like Espressif and their nice ESP IOT microcontrollers which have pretty solid documentation.
Hehe – trust me the Espressif documentation used to be just as bad !!! Various people discovered this IC around the same time and worked together to form an early English language datasheet. This triggered usage in the West and now the company maintain a proper English language website and everything. But for most Chinese companies it isn’t worth the effort – all their clients are Chinese as probably 50% of all electronic design work now takes place in China or Taiwan.
And yes lots of good stuff on LSCS. I get all my power MOSFET drivers from them as they are far more rugged than the ones Infineon or ST sell. Also various switch-mode PSU chips which use far less external components that Western designs.
Being one of the few Westerners to use it, I can state quite categorically that TRON was a dog so to compare it with ARM or RISC-V is just plain silly. A more sensible comparison would be with a 68030, which it was also poorer than.
There are however several Chinese processor architectures. Sunway is the most interesting but isn’t available in the West, whilst Loongson builds upon and extends the MIPS64 architecture that has floundered in recent years here. But most interesting are the Chinese AI startups as they do things differently to the Western ‘1000s of GPUs is good, millions is better’ approach. These are so good, Xilinx bought the best one, whilst Horizon Robotics looks like becoming a real star in the coming years.
Trade wars, export controls, corporate drama and even a physical brawl in Bitmain. Looks like the semiconductor industry is reaching drama levels only see by social media influencers.
But is worth noting the decision of creating ARM China was in part to slow the adoption in China of what i think is the biggest threat to ARM future and that is RISC-V, if the Chinese perceive that they going to lose access to ARM IP, they would have throw everything they have to RISC-V, i mean government, academic and corporate. So if RiSC-V become big in market of the size of China then could possibly make the Softbank acquisition of ARM Holdings looks like another WeWork or worse.So there is hundred of Billions in future profits at stake.