Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Businesses Intel

TSMC, a Company Few Americans Know, is About To Dethrone Intel (bloomberg.com) 195

For more than 30 years, Intel has dominated chipmaking, producing the most important component in the bulk of the world's computers. That run is now under threat from a company many Americans have never heard of. From a report: Taiwan Semiconductor Manufacturing Co. was created in 1987 to churn out chips for companies that lacked the money to build their own facilities. The approach was famously dismissed at the time by Advanced Micro Devices founder Jerry Sanders. "Real men have fabs," he quipped at a conference, using industry lingo for factories. These days, ridicule has given way to envy as TSMC plants have risen to challenge Intel at the pinnacle of the $400 billion industry. AMD recently chose TSMC to make its most advanced processors, having spun off its own struggling factories years before.

TSMC's threat to Intel reflects a sea change in chipmaking that's seen one company after another hire TSMC to manufacture the chips they design. Hsinchu-based TSMC has scores of customers, including tech giants Apple and Qualcomm, second-tier players like AMD, and minnows such as Ampere Computing. The explosion of components built this way has given TSMC the technical know-how needed to churn out the smallest, most efficient and powerful chips in the highest volumes.

"It's a once-in-a-50-year situation," said Renee James, the former No. 2 at Intel who heads startup Ampere. Her company is less than two years old and yet it's going after Intel's dominant server chip business. That Ampere thinks it can compete is a testament to stumbles by Intel, and TSMC's ability to benefit from those mistakes. It's been a decade since Intel faced major competition and its 90 percent revenue share in computer processing will again deliver record results this year. But some on Wall Street are concerned, and rivals are emboldened, because TSMC has a real chance to replace Intel as the best chipmaker in the business. Last year, the Taiwanese company amassed a bigger market value than its U.S. rival for the first time.

This discussion has been archived. No new comments can be posted.

TSMC, a Company Few Americans Know, is About To Dethrone Intel

Comments Filter:
  • by jellomizer ( 103300 ) on Wednesday November 28, 2018 @12:28PM (#57715438)

    Intel Designs and makes their own chips. TSMC just takes your designs and makes the chips. Which is a perfectly fine business model, but comparing Intel to them isn't really the same. As Intel puts a lot more R&D in designing the chips then making them.

    • by Joce640k ( 829181 ) on Wednesday November 28, 2018 @12:35PM (#57715478) Homepage

      Nope, most of the R&D work is in the fab, not the microcode. Making the chips work reliably is the hard part.

      AMD is currently shipping 7nm chips thanks to TSMC, Intel is stuck at 10nm.

      • by alvinrod ( 889928 ) on Wednesday November 28, 2018 @12:51PM (#57715580)
        Intel isn't even on 10nm yet, at least not in terms of having mass production of shipping products (I'm not even going to count the pathetic i3 they trotted out and relegated to the bottom end of the Chinese market). However, if you look at the characteristics of each process, it becomes pretty clear that TSMC is (and historically always has been) playing loose with their naming. Not that it really matters anyways since it's just a marketing term. Intel's 10nm has roughly similar characteristics to TSMC's 7nm process. Even that is bad for Intel though, as historically they tended to have at least a year (and more often two years) lead over the competition.
        • by epine ( 68316 )

          Not that it really matters anyways since it's just a marketing term. Intel's 10 nm has roughly similar characteristics to TSMC's 7 nm process.

          Everyone who follows this even a little bit already knows that Intel's processes are the real thing, and everyone else's numbers are measured with parallax and a fat rubber ruler.

          Not that this normally matters, as you point out. It only matters when some idiot comes along crowing about how Intel has been "left in the dust".

          It's also the case that Intel is designing tr

        • I'm not even going to count the pathetic i3 they trotted out and relegated to the bottom end of the Chinese market.

          I don't think they even make it any more, they never did manage to get the defect density to a commercially viable level. I think they essentially just bulldozed that fab line and went back to square one.

      • Not necessarily. The Fab is very important. It's the other half of the equation. You need the fab processes to create the devices necessary for your designs. However, Intel got to where they are not only because of their fab technology, but also their design technology. Creating design tools capable of handling nearly a billion transistors is no trivial matter. They did not exist. How do you standardized logic cells? How do you lay them out so they actually are capable of the speed (timing) you expect. How
      • Hmm.....

        Think this might be a company worth buying stock in?

      • by Aighearach ( 97333 ) on Wednesday November 28, 2018 @02:32PM (#57716222)

        Intel's "10nm" is the same as 7nm. In fact, one of the dimensions is even 7nm, and they could have just measured from that side and called it 7nm. The numbers don't have meaning.

        You're conflating chip design with process design. Obviously, the company that owns the fab has to do the process design. And that is separate from the design of the chips.

        You were wrong as soon as you decided to start with the word "nope." Obviously, Intel does R&D for process, and for the chips they make, so they're doing a lot more R&D than a company that designs a comparable process, but not the chips. The difference in business models, and the fact that they're both competing on the latest generation of process, guarantees that Intel must be doing more R&D.

        The mystery in your comment is why you single out microcode to represent the whole chip design process.

        • Intel's "10nm"

          What are you talking about exactly? Because no one has seen a successful run of this process yet. The only [wikipedia.org] Cannon Lake part is a partially failed product with a disabled GPU/bad thermals/and no advantages over a comparable mobile Kaby Lake CPU.

      • by Solandri ( 704621 ) on Wednesday November 28, 2018 @02:51PM (#57716334)
        In the wonderful world of fab marketing.

        Each company seems to be measuring a different thing when they report a process is "x nm." So while you can compare nm within a single company's offerings, you can't compare them between different fab companies. TSMC and Samsung's 7nm processes leapfrogged Intel's 14nm process (37.5 million transistors per mm^2). But they're still behind Intel's 10nm process.

        • It is like modern retail 'sales' - up to 90 percent off, which means 1 item is 90 percent off and everything else is 5 percent or so.

          Their 7nm process is really just down to 7nm, but most of it is over 10.
      • Intels 10nm is not even there that much (and already 4 years late), they produce a handful of very simple chips. TSMCs 7nm is pretty much the same ans Intels 10nm, both are marketing numbers anway. The main difference, TSMC is producing already millions of rather complex processors for Apple, Amd and NVidia and others while Intel had a stockholder launch 4 years late producing a handful of the simplest celerons they have with high failure rate and apu disabled to keep the stockholders calm.
        2019 will be inte

    • by Shaitan ( 22585 )

      What on earth gave you that idea? You are suggesting Intel pays a handful of code monkeys more than they spend building and developing custom materials and hammering at cutting edge physics with insanely expensive built from scratch equipment and then retrofitting that into fabs? You must be kidding.

    • by sh00z ( 206503 )

      ... Intel puts a lot more R&D in designing the chips then making them.

      and yet, there's still not a chip on Intel's horizon that won't be susceptible to Spectre/Meltdown. Seems to me the R&D is pointed in the wrong direction.

      • by gtall ( 79522 )

        Oh, have you been getting their memos? Do share!!

    • by RhettLivingston ( 544140 ) on Wednesday November 28, 2018 @02:07PM (#57716080) Journal

      You are correct but you can compare Intel to TSMC plus its customers who are doing their own design work. The article did so.

      Historically, the company has squashed rivals using a research budget that dwarfed anything else in the industry. But TSMC’s approach is even undermining this advantage.

      While Intel still outguns TSMC in capital spending on new plants and equipment, the tables are turned when you combine the research budgets of TSMC customers like Qualcomm, Apple, Nvidia Corp. and Huawei Technologies Co.

      According to Goldman Sachs, the combined budgets of TSMC’s customers are not only larger than Intel but the gap is increasing. By 2020, they will spend almost $20 billion, according to its estimate, at least $4 billion more than Intel.

      IMO, the rise to dominance of TSMC's business model is inevitable and probably being driven by the industry's fall off of the Moore's Law curve.

      For decades, companies have been able to keep increasing the capabilities of their product by just buying the next-generation general-purpose chip. They got lazy in the process. I'd say this transition occurred in the '87-'97 time frame, a time when the need for engineers to design custom hardware plummeted in favor of buying COTS. But the general purpose approach is starting to fall short of the increases necessary to drive new consumer purchases.

      But innovation is still possible. Our laziness has created a deep untapped well of performance growth that can be had by equipping the domains to create domain specific designs. If we can reignite domain-specific engineering, many domains can achieve order of magnitude changes in performance by rolling their own designs.

      TSMC is enabling the larger of these domains to achieve purpose-built silicon designed by the domain's engineers for the domain.

      • dominance of TSMC's business model is inevitable and probably being driven by the industry's fall off of the Moore's Law curve

        Mostly driven by the mobile ARM market eclipsing the x86 PC market. But flattening of Moore's law is certainly a contributor, it means everybody is working with exactly the same fab equipment, it all comes from the same place.

    • Obviously these are not comparable companies. We're comparing the efficiency of a vertically and horizontally integrated manufacturing models.

      TSMC has proved that sticking to your knitting is a more efficient business model than Intel's vertical integration. In the past it was always Intel's Wintel-fueled R&D budget that kept it one process node ahead, but the ARM wave gave TSMC a comparable budget and they focused it more efficiently. With TSMC now a business of comparable size and growing faster, ther

  • by xxxJonBoyxxx ( 565205 ) on Wednesday November 28, 2018 @12:35PM (#57715490)
    Intel saw how Microsoft got PC builders to beat each other up to carry the One True Operating System, and decided to avoid that racket by building their own parallel brand: Intel Inside (with music) so that CPUs from AMD, etc. didn't push their pricing down. It largely worked, to the point where almost everyone in America knows who Intel is: "it's the guys who powered my Compaq 10 years ago - dun dun dun DAHN".
    • When even Weird Al parodies you [youtu.be] then I guess you know you are popular. =P

    • Intel saw how Microsoft got PC builders to beat each other up to carry the One True Operating System

      The difference is, running macOS, Windows, Linux or OS/2 actually makes a difference compatibility wise. Intel vs. AMD, not really (there were some differences at the cutting edge, like SIMD, but most software waited until those were pretty standard anyway.)

      • >> Intel vs. AMD, not really

        Duh - that's why they did (or anyone else would) create their powerful "Intel Inside" brand: to create perceived value where there is little actual difference between themselves and their competitors. See also: Nike, McDonalds, Red Hat, etc.
  • It's not only chips (Score:5, Interesting)

    by bogaboga ( 793279 ) on Wednesday November 28, 2018 @12:40PM (#57715520)

    Last year, the Taiwanese company amassed a bigger market value than its U.S. rival for the first time.

    I am afraid the USA is [quickly] becoming an entity of little consequence. It's sad. When we lost manufacturing to China, folks here were ebullient, saying we surrendered cheap labor intensive jobs to China. They were happy that when it came to technology, we are "up there."

    From this piece, it now appears that we're not safe. All our president can do is to apply sanctions - which hardly work by the way.

    Given that china still owns a significant amount of our debt, we're surely in trouble.

    Not so long from now, Russia and China will introduce the C929 [wikipedia.org]. Then our serious remaining industry will be threatened.

    Suggestion: Let's stop fomenting chaos in far away lands and concentrate on making the USA a beacon of prosperity once again.

    • Given that china still owns a significant amount of our debt, we're surely in trouble.

      That could actually be worse for them than it is for us.

      There's a saying: if I owe you a thousand dollars, I have a problem. If I owe you a trillion dollars, *you* have a problem.

      • Re: (Score:3, Informative)

        by Aighearach ( 97333 )

        Given that china still owns a significant amount of our debt, we're surely in trouble.

        That could actually be worse for them than it is for us.

        There's a saying: if I owe you a thousand dollars, I have a problem. If I owe you a trillion dollars, *you* have a problem.

        Well, that is actual debt though.

        Holding bonds isn't the same as giving a loan, though; you get no power over the bond issuer. Defaulting only means people wouldn't want to buy that bond in the future.

        Also, the US government issues the bonds, and they all get purchased. What the market does with them after that doesn't even affect the US Government. There is no reason for the US Government to care who buys them. Only a country that has limited demand for bond purchases would need to care about that stuff.

        Wh

        • The U.S. can threaten to selectively default on the bonds that China holds. Who that hurts the most is an open question. Note that inflation is a continuing partial default.
    • by dgatwood ( 11270 ) on Wednesday November 28, 2018 @01:14PM (#57715706) Homepage Journal

      Meh. Airbus is a European company, and competes head-to-head with Boeing. The difference between one main competitor and two is not nothing, but it also isn't the end of the world. Besides, Boeing is also competing with companies like Bombardier (CRJ) and Embraer, with at least four more regional jet makers starting to gain popularity as well. Sure, those companies build only smaller, regional jets, but every one of those routes is one that could have been flown by a 737, but wasn't. The implication that Boeing is somehow going to go from no competition to crushing competition is kind of silly in that context. They have a *lot* of competition already, and one more player almost certainly isn't a big deal.

      Also, I've seen Russian manufacturing quality control, and I've seen Chinese manufacturing quality control, and I wouldn't fly on a plane built by either one of them unless there was an American company running the show, with employees doing random drop-in checks to keep them honest. I've seen way too much appallingly bad quality control (we're talking loose screws rolling around inside, unauthorized part substitutions causing a 70% DOA rate, premature failure caused by overheating critical components while soldering, etc.) out of Chinese manufacturers to trust them with my life. And Russian heavy industry seems to do well up until they start cutting back on the rate of manufacture, and then those last few off the line are death traps.

      If they make it fifty years without a significant uptick in crashes, I *might* start trusting them. And even then, it would still just be a "might", not a "will". And that's also true for any airline. They're going to be very wary of any new manufacturer until it has proven itself.

      • Also, I've seen Russian manufacturing quality control, and I've seen Chinese manufacturing quality control, and I wouldn't fly on a plane built by either one of them unless there was an American company running the show, with employees doing random drop-in checks to keep them honest.

        Your sentiments hardly matter given that when it comes to human space travel, our government currently relies on the Russians at 100%. In fact, when it comes to rocket engines, the USA also relies on the Russians.

        All it will take is appropriate financing and the rest will be history. Americans used to say the same about Japanese motorcars in the early 80s. The story is very different now given that GM is busy closing factories now.

        • In fact, when it comes to rocket engines, the USA also relies on the Russians.

          Utterly false, except for that for first stages, one batch of mothballed NK-33s Orbital Sciences used/is still using for a while for their Antares craft, one of which caused a nasty total failure in 2014, and is to be replaced with the RD-181.

          ULA uses the RD-180 for the Atlas V, a tremendously successful albeit now expensive rocket, with 78 successful launches and one partial but not complete low orbit failure for the NRO, an arra

          • by dgatwood ( 11270 )

            Indeed, the recent failures of the Russian space program are precisely what I was thinking of when I expressed concern over the quality control in Russian heavy manufacturing.

      • What about an American airplane run by Asians? Say a Boeing 737 Max?

      • It sounds like concerning competition for the 737 until you realize that the design has long outlasted what was intended, and they're only doing small amounts of R&D for the new models. And even though it uses a lot more fuel than the regional competition, it also gets lower insurance rates because of its service history. It is easy profit for Boeing.

        Lots of western companies with quality manufacturing are competing, and Boeing mostly wins due to trust. I don't doubt airlines who simply can't afford Boe

      • Boeing is competing with neither. They more or less own Embraer now and the CRJ line will last for just a few more years.

        • by dgatwood ( 11270 )

          Heh. Fair enough. I hadn't noticed their controlling stake in Embraer. I wouldn't rule out Bombardier, though. The Global 7500 is going to ruffle some feathers on the bottom end, and if some of those enhancements make it into a future CRJ cabin upgrade, it could shake up the mid-range, too.

    • Did you know that Russia was always building aircraft? Apparently not.

      They always cost less to buy, too. And yet. ;)

      Expect that plane to be very popular; in China.

    • It's sad. When we lost manufacturing to China ...

      Over the last 30 years, manufacturing in America has doubled [marketwatch.com].

      The "deindustrialization of America" is one of those things that "everybody knows", but is actually nonsense.

      • It's our working class that's been devastated. I mean come on, that article counts refined petroleum products as "manufactured goods". The day our working class has jobs again is the day I feel better. Until then, it's a crisis.
        • by gtall ( 79522 )

          That's a fool's errand. The only way to put the American working class back in manufacturing is to raise the cost of everything in the U.S. drastically. That would cause an inflation. Interest rates would rise, government deficits would rise even faster than the King of Debt, el Presidentie Tweet ever hoped for in his wildest dreams.

          • Let's create a win for them. It's the least we can do after deliberately ruining them the past 30 years. Afflict the comfortable, comfort the afflicted.
        • It's our working class that's been devastated.

          Actually, they haven't. This is another thing that "everybody knows", but is not true.

          Working class incomes, adjusted for inflation, have stagnated since the 1980s. Stagnation means they stayed the same, not that they declined.

          Is it a problem that working class incomes have stagnated, especially when people higher up the income ladder have prospered? Sure, most people would agree that is a problem. But by using hyperbolic words like "devastated", you aren't contributing to the solution.

    • It's about a trillion out of 15.6 trillion. We owe most of the debt to ourselves. Meanwhile what we own to other countries can largely be thought of as tribute. Folks don't seem to realize that with a big military comes an empire, and America has an empire like any other nation with a big army. You don't need 19 aircraft carriers to defend yourself against Canada & Mexico...
      • by gtall ( 79522 )

        Check the deficit figures, it ain't the military busting the budget. It is the blue hairs and I-have-to-get-mine mentality.

    • by gtall ( 79522 )

      Manufacturing in the U.S. is roughly north of $2 trillion a year. 2nd to China out of a roughly $19 Trillion economy. China's manufacturing is roughly $6 Trillion. Theirs is about 42.6 % of their GDP, the U.S. is roughly 10 % of GDP.

      So the U.S. has lost manufacturing, I suppose you could argue it went to China. But the SE Asian countries are attempting to do to China what some argue China has done to the U.S. That's one of the reasons China is swaggering around the S. China Sea claiming to have the Biggest

    • Initially, the engine is likely to be supplied by Rolls-Royce or General Electric that already have products in this class.

      And the most expensive part of the plane will still be made in the US or Britain, as will probably most of the other components that you would attach to these engines.

      And also I fail to see what the C929 has to do with TSMC. Plus Taiwan and China are currently very different places / political bodies.

  • by PCM2 ( 4486 ) on Wednesday November 28, 2018 @01:31PM (#57715822) Homepage

    I don't get what this article is trying to say. TSMC is about to dethrone Intel ... how?

    In value? TSMC's market cap is half that of Intel.

    In innovation? TSMC is a manufacturer of outsourced chip designs ... think a Chinese factory that produces car parts for Ford. The article references the combined research budgets of Qualcomm, Apple, Nvidia and Huawei. That's cute, but what does it have to do with anything? In real life, those budgets don't actually "combine." They're all competing against each other. Does the outsourced manufacturer benefit from having all these customers? Sure, but so what?

    Intel, meanwhile, designs its own chips, which still dominate the PC and server industries. It also makes chipsets to go along with these, and most processor customers use these integrated chipsets these days. Add to that wireless devices, compilers, SDKs, and all the other stuff it produces that comprise an entire ecosystem.

    Does Intel face challenges? Sure. I don't think anybody can argue that it's kinda fallen on its face in the mobile market. But does anyone see AWS or Boeing approaching TSMC to produce custom CPUs tailored for specific workloads?

    So in that sense, you might as well say TSMC is about to dethrone IBM. It just doesn't make any sense to me.

    • And doesn't Intel still make what are pretty much the best Ethernet chips. They work, and are well and openly documented? And if you don't need the highest performance, aren't their on chip GPUs also good on both metrics?

      Anyone have some real facts, or pointers to the quality of AMD's current chipsets? I've heard ugly mutterings, but have no knowledge, still using stuff that's many years old now.

    • What the article didn't emphasize is that Intel's main market is internal consumption (Intel-branded chips), while companies using the foundry model that TSMC pioneered sell to the industry as a whole. (I note that Intel does have a small, and not successful, attempt at a foundry business [intel.com].)

      Intel must amortize the cost of its IC process development, plus the cost of new fabs every generation, based solely on the revenue it can generate from the sale of its own chips, while TSMC can spread that cost over th

    • TSMC is about to dethrone Intel ... how?

      TSMC already dethroned Intel in fab capacity. TSMC is now in the process of dethroning Intel in transistor density, and after that comes dethroning Intel in clock frequency.

    • But does anyone see AWS or Boeing approaching TSMC to produce custom CPUs tailored for specific workloads?

      Yes, Amazon's Graviton for one.

  • Just like magic rock gardens a spire can rise only so far and then it crumbles down to be replaced by others on top of the ruins.
  • by Zontar_Thing_From_Ve ( 949321 ) on Wednesday November 28, 2018 @02:20PM (#57716154)

    That Ampere thinks it can compete is a testament to stumbles by Intel, and TSMC's ability to benefit from those mistakes.

    Maybe. But from personal experience, I can also say that they might just be stupid. I work in IT for a Fortune 500 company. I never name them here. We do a lot of software as a service type things. We're pretty good at what we do and in some areas we are likely the top dog for US based solutions. The business segment I support isn't sexy at all. But every year we have to deal with multiple startups who try to take our business. Many fail. We often see new companies say they can do everything we do for half the cost. Half the cost? Sure. But in reality they do like 1/3 instead of everything we do. So we have customers who leave over cost and then come back because our competitors really suck. Again, this is just not a sexy business segment in what I work with. This isn't it, but imagine you work for Turbo Tax. You don't just have H&R Block, Tax Slayer, etc. to deal with but every year some punk upstart company says it can do taxes better for way less than you charge. How much better can it really be? Either tax software works or it doesn't. You can't really make it "better". But we still have competitors who claim that this somewhat stagnant business can be done better and cheaper. And every year we watch as they go out of business.

    Another problem I've seen with startups is the "We can't possibly fail because we're geniuses!" attitude. Most startups do fail. I know a guy who has spent most of his IT career chasing startup glory and failing. He does make really good money, but he's always having to find a new job with the next startup. The older product I still support was started by a successful startup that my current employer bought out. I can tell you that a large number of people associated with that startup left us within a year or two of the acquisition going through and they went to a few different new startups that various people associated with the original company started. All those startups failed. And a large number of those people returned to my employer, tail between their legs. So sure, maybe what is said about Ampere is true and will happen, but I'm not ruling out that it's a dumb idea and they'll fail at it.

    • Quick question if I may impose? In very general terms, you and your (would be) startup competitors, SaaS offerings, who uses 3rd party cloud vendors and who in-house, and "why?" for all the combinations except the obvious of startups using the cloud.
  • Hsinchu, Taiwan is also a mini free trade zone. It's fenced off and guarded. What goes in gets shipped out and not taxed.

    It's been around for quite a long time (over 20 years that I know of). I've visited the area.

    They really know what they're doing. They have on campus dormitories for employees if they choose to live there (and work long hours).

    Note that this is NOT mainland China and is a sort of democracy. China insists they're a rogue colony though and this has been an item of contention for severa

    • Clarification: Hsinchu is not all fenced off, only the development TSMC and associated areas. They call it "Science Based Industrial Park".

Beware of all enterprises that require new clothes, and not rather a new wearer of clothes. -- Henry David Thoreau

Working...