TSMC has acquired a lead in this area through a number of different methods. One of the main things is that they focus deeply on manufacturing. Another is that they work 24 hours a day in R&D, running 3 shifts so they basically have the lights on all the time. And as mentioned above, the upfront costs are incredibly high with a fab costing on the order of 20+ billion dollars to construct.
Intel is attempting to catch up, but it will likely be another 3 to 5 years before they are able to do so. Honestly just having R&D up and going all the time is probably a huge advantage for TSMC and probably a big reason behind their success. Regardless, suffice to say basically all cutting edge product shipments would cease in a matter of months if TSMC fabs were destroyed.
https://news.ycombinator.com/item?id=35667282
It also has to do with competitive boasting: theirs goes to eleven.
Honestly, I'm not sure anything of that is real. I can't believe other fabs don't run tests 24/7, and I can't believe they have people that rarely meet changing the same machines instead of only running tests without changing anything.
Another advantage of TSMC is how they have enough fab space dedicated to R&D that they can run prototypes through quicker because they aren’t competing with manufacturing to get processed.
Maybe the culture thing is really pervasive. It wouldn't be a first.
do you have examples?
Your addition of "abundantly" makes things more ambiguous, but I think the choices available in most large US cities is _decent_ after midnight. Can you get absolutely anything like NYC? No, probably, not. But burgers, diners, and fast food are all generally available. That's also ignoring 24/7 grocery stores like Walmart that can fill in the gaps if you're craving something hyper specific. Perhaps you can't buy it ready-made, but you can buy the ingredients!
What about Samsung? I thought they also made leading edge.
According to them they will do it in next 12 months.
The question is how easily could they migrate and prioritise the workload?
It's also worth noting that in the event of a war the US is very, very likely to bomb the shit out of the TSMC plants.
https://www.businessinsider.com/us-would-destroy-taiwan-semi...
https://www.theregister.com/2023/03/14/us_china_tsmc_taiwan/
Taiwan being the core producer of the super high end chips is guaranteing them that if a war ever happens, they won't be left as sacrifice to the opponent while the rest of the world goes business as usual.
They critically need to be a strategic and non replaceable producer.
No, Samsung has some in South Korea also.
this isn’t correct.
UMC, SMIC and TI have more than 10 fabs each.
Nanya 3 fabs.
Micron > 10 fabs.
Intel > 10 fabs.
Tower Semiconductor 7 fabs.
Rohm > 10 fabs.
Fuji 4 fabs.
Fujitsu 5 fabs.
Kioxia 10 fabs.
Renesas 9 fabs.
and the list goes on and on.
[1] https://en.wikipedia.org/wiki/List_of_semiconductor_fabricat...
Texas Instruments has its own factories in the USA but AFAIK they make other kinds of ICs than the popular ones that TSMC is known for, they do mostly analog and embedded (still important stuff).
The sibling post to yours lists other factories. There are definitely other companies.
And it is TSMC, Samsung and Intel.
ARM doesn't make chip
on the low-end and mid-tier chip, you have plenty of fabs like Global Foundries, UMC...etc.
on the high end, you have Intel, TSMC, and Samsung
I don't think it's as strong dependance as many commenters assume here. Samsung is 2 years behind TSMC and intel 3-4 years behind in terms of fab capability. While losing 2 years of progress is not great, it's definitely nothing like world can't function without TSMC.
Obviously they need few years to ramp up but I assume it's not like Taiwan geopolitics situation would change in a day.
This is quite hyperbolic. While losing TSMC will have a massive global impact, there are alternatives in many different market segments (e.g. Intel for CPUs and GPUs for computing, be it personal or datacenter), STMicro/NXP/Infineon/Bosch for industrial applications, etc. They may not be as advanced, or as good, or as cheap, or in enough quantities, but it's flat out false to say that TSMC is the entire semiconductor industry. Just a major portion of the bleeding edge.
Samsung is behind a generation (or thereabouts). Not sure why exactly.
AMD spun off their fab as GlobalFoundaries. The separate company bowed out of the bleeding-edge fab business. Their 14nm node was licensed from Samsung and they completely cancelled their 7nm node.
As bleeding-edge fabs kept getting more and more expensive lots of companies decided to go fabless. They were able to do that because unlike everyone else TSMC doesn't make chips for itself so it doesn't represent a competitive threat. This is the real key to TSMC's success.
Other companies like GlobalFoundaries and ON Semi are filling volume with older process nodes like 14nm. There's a large number of chips needed by millions of manufacturers large and small. Only a small proportion of those need the power/perf you get from the latest and greatest. Manufacturing on an older node is cheaper thanks to lower capex and much much better yields.
Eventually I think we'll see litho equipment make its way into the niche market similar to 3d printers. We've seen some people hacking together stuff to make their own chips in a garage... they're tiny 5-2000 transistor affairs but it would be quite interesting to be able to churn out a custom chip with a million transistors on it.
Why didn't they just make art with the same quality?
It's not just what tools you have, but how you use them.
You don't just go plug in one of these machines and it starts pumping out working chips.
Increasing density means better controls and processes, better knowledge of running and calibrating the machines.
Tons of processes before and after the ASML machines.
Each fab is the development of an entire "process". Furthermore how the machines are used varies greatly. Everything from exact doping and cleaning processes to metal deposition parameters, plus the masks you feed into the system which these days are all interference-based - that is you rely on the interference patterns because you can't actually make masks nor use light of a wavelength small enough to create 14/10/7nm feature sizes. The mask you make is nonsense but the interference pattern it creates results in the actual patterns you want.
How many layers, how you can intersperse metal vs silicon layers, allowed dimensions for gates, wires, distances between components, and a whole host of other things are unique to the specific process. Oh and parameters vary based on what you are making on a given process (eg SRAM vs logic). Different layers can have different requirements as well. You might not even be able to take a design to another fab from the same company on the "same" process node, let alone a competitor.
This stuff is all extremely closely guarded by every fab. If you're a serious customer (read: $$$$) the fab will share these parameters with you so your engineers can design to that spec. Then your first run will likely have problems and need to work with the fab engineers to tweak layouts and such to get usable yields - depends on how much you are relying on the fab for design/layout. TSMC will happily license you certain blocks to use in your design along with guarantees on yield since the design is made for their process from the start. You can also license designs (eg from ARM) that are pre-customized for a given fab's process.
The nirvana of a household printing capability never really materialised.
Won't you be better served with a FPGA? Are there any cases, besides hobby projects, where making you own silicon on a process many generations behind the state of the art would make sense?
With 3D printers, there are parts that are simply impossible to make without them, and there is no programmable plastic, so it makes sense to use them. In fact, we are starting to see entire factories churning out plastic parts from FDM printers not so different from what consumers buy. In the case of Prusa, they even make the parts using the same printers they sell to consumers.
However, TSMC has the largest capacity worldwide. Intel's probably second for capacity. Samsung's probably third. Micron, Texas Instruments, Infineon, Epistar, and several others are lower down yet significant.
If you need small chips, out of plants already constructed, based on the data in the list, Intel and Samsung appear to have the "current" smallest node size at 7 nm. TSMC has three plants under construction with 3-5 nm node size. Intel and Samsung also have a lot of capacity in the 7-20 nm range.
Most companies are in the 20+ nm size ranges, although definite exceptions with Nanya (10 nm), SK Hynix (10 nm), nanoPHAB (10 nm), Win Semiconductor (10 nm), GlobalFoundries (12 nm), SMIC (14 nm), UMC (14 nm), Kioxia/SanDisk (15 nm).
There's also technology and product variety, although most list 3-5 for each plant, so there's a bunch of specializations.
[1] https://en.wikipedia.org/wiki/List_of_semiconductor_fabricat...
This is an interesting but not so obvious question, because it is so basic, I would thought it wont be on HN at the same time judging from comments on Hardware topic it really should have been spelled out clearly in the first place.
So first you have IP companies like ARM. They have lots of different IPs. ( Intellectual Properties ) From ISA ( Instruction Set Architecture ), think x86, ARMv8 etc to the actual CPU design, think Intel P-Core, E-Core, Skylake, AMD's Zen core etc. Along with many other IPs from GPU to Network, CPU Interconnect or even DRAM controller. ARM will work with TSMC ( or any other popular Fabs for specific market ) to have those design specific for the Fab and specific to certain Fab technology.
So in an extremely over simplified situation, you buy a few IPs, or there are even reference designs. Stitch them together. Send them over to TSMC and say hey can you print them ( or Fab ) for me.
What TSMC doing that nobody can do ( in the same time frame ) is that they could Fab it using industry leading technology, or the smallest transistor currently available as long as you are willing to pay for it.
There are plenty of chips that doesn't require TSMC to print. From Automotive to low cost semiconductor using very very old semi-conductor Fab's technology. Think about the small chips that goes into toys or electronic appliance. Or simple calculators.
But then what are Intel, Arm, etc in this picture? I don't understand semiconductor manufacturing in enough detail -- I assume TSMC occupies a different part of the supply chain? But chip manufacturing seems like a pretty integrated process top to bottom; what's the division between them? In concrete terms, what is it that TSMC is doing that nobody else is?