All through 90s and most of 2000s intel had the FAB process lead over TSMC and Samsung. It was the 14nm and then 10nm transitions that hugely derailed intel. However they are at least talking about a roadmap (for two years now) with the same message of catching up to competitors by 2025. They also have a plan to fab third party designs which they previously never did under the IFC name (intel foundry services, I think).
Currently they are transitioning to 7nm, and the roadmap talks about 2nm as well. For reference, nVidia 3080/3090 series uses Samsung 8nm process, AMD Ryzen third gen and Radeon 6x00 series is TSMC 7nm, Apple M1 was TSMC 5nm, and M2 TSMC 3nm.
One thing to remember is that a current “top end gaming PC” can easily be put together with an intel fabbed CPU and a Samsung fabbed GPU, so TSMC’s dominance, while very real, can be overstated.
IIRC the problem with Intel isn’t that they can’t figure out the tech to get the chips small enough by itself, it’s that the entire design philosophy of the x86 architecture doesn’t lend itself to the smaller die sizes w/o them completely redesigning the whole thing from scratch so they’ve been reluctant to do it.
Wow, cool post, and very informative! Sounds like you have had a seat on the 50 yard line for some interesting stuff.
Overall though I think globalization is a powerful force. Systemically, it makes no sense to invest in making things you can get cheaper from elsewhere. That of course is in the absence of friction like nationalism, war, politics, and what not. But structurally, it makes as much sense to spend trillions on building a chip fab capability in the US to rival that in Taiwan as it does to pay Americans three to five times as much money to make cheap clothes as it costs to have them made overseas.
Thanks for sharing that. The two new Intel factories are about to begin being built about 8 miles north of me, with eight more to follow after that. My son will graduate high school next year, so I’m pushing him to think about how transformative this could be for this area and his future possibilities.
To be honest, as dyed-in-the-wool free market capitalist I’d be skeptical of the CHIPS act if I didn’t bleed Intel blue. There are two solid arguments why semiconductors are different. First, national defense. How much money would Putin pay to have a state of the art semiconductor factory(s) capable of producing all the chips needed for their missiles, smart bombs, and planes? The war in Ukraine would different if they did.
Second, economic of Fab are way different than almost any manufacturer, especially something like clothing. The press release says the $20 billion Ohio Fab will employee 3,000 with an average wage of $135K. Roughly 2/3 of the jobs will be technicians which only requires a 2 year AA degree at acommunity college. When we do the math we end up 20 billion/3,000 or $6.7 million in capital expenditure per job!
So the total payroll for the factory will be a $400 million, or put another way a 2% increase in interest r rates will cost Intel as much as the payroll. The most important criteria for Fab is getting it built and permitted ASAP, because they are only state of the art for about 5 years. Second, is workforce. Incentives are probably 3rdl Wages are almost an afterthough.
In contrast you can probably buy a factory fill with sewing machines in Bangladesh for $10 million (wild ass guess) and have 300 odd workers for shift, so a capital expenditure of $10,000/worker. In this case a 2% increase of interest rates cost you another $200K/year. But a $1 hour increase in wages costs you $2 million! Wages are everthing.
Or put it another way if the US can’t be competitive in semiconductor manufacturing we are really screwed.
Which is why I mentioned friction; this is one of those cases where it probably makes sense to do something that is usually contraindicated. I just wish we had a real policy over time to do things like this rather than panicking and throwing money at it when it’s possibly too late.
I think at this point you pretty much need national investment to work at the cutting edge of process node size. It is just so damn expensive. I wonder just how much money TSMC does get from the Taiwan government through the “Development Fund of the Executive Yuan”.
Apparently a problem TSMC is seeing with the fab they are building here is:
One obstacle TSMC is encountering: There aren’t enough trained semiconductor engineers in the area to staff the facility,
Yes. Having something so critical to our military and economic strength sitting on an island that is currently being threatened by a totalitarian government just seems like a really bad idea.
I read somewhere that the whole reason why the US is no longer the world leader in semiconductor manufacturing is that other countries’ governments have made it economic policy to pump money in and subsidize semiconductors. I can’t evaluate the accuracy of that, but if true, it certainly seems like we would be foolish not to follow suit on such a critical technology.
Economic policies alone aren’t enough. Many, many countries have put money into semiconductor manufacturing and have all failed even at larger nodes (even when those larger nodes were already obsolete). Semiconductor manufacturing is extremely hard to do at even a decade’s old node sizes.
Sure it’s hard, but the US simply doesn’t have the option of not being able to do it. Even the EU, it’s a bad idea not be to able to do it.
At least to whatever standard that’s required, maybe not the latest, but even not being cutting edge is a bad idea, especially if it’s crucial to that thing you’re already spending more money than the next what is it, x countries combined?
These were both great, thanks for linking them. I was awestruck learning that 80% of the new workforce in Taiwan there have college degrees, fully paid. What a fantastic way to become a powerhouse for technology.
Also this graph stuck out on TSMC directly and where their customers lie:
Which kind of gets to the issue with us being so reliant on things there. Even if China doesn’t try to pull Taiwan in, anyone else can and would use it as a target for global effect. It truly is a gem now for what they do, but a one relying on trust that any bad actor won’t try to mess with it to the detriment of us all.
This thread is moving fast but I wanted to thank you for sharing your history within the industry as well as the explanations and detail regarding things about it.
Coverage of this kind of stuff is why GPS is the only cable news show I watch anymore. He hits a lot of topics that aren’t widely touched elsewhere in US media(not just TV news).
And I’ll second your thanks to @Strollen for sharing!
It’s been that way for a long time. There are a number of things mentioned when it’s pointed out but many of the things that make the EU a strength and also wish lists for us wanting more from our own democracy run afoul of getting a bigger tech investment within EMEA.
Notably, things like labor laws, pro-tech laws and regulations, government funded research and investment, tech centric education, and very notably, venture capital. The US and other locations have taken on a lot of risk along the way to produce the successes we do have, Strollen even mentions that, above. But the payout can be great.
Also for what it’s worth, it isn’t just the tech industry where this is a problem in EMEA, it’s multiple industries. There is a real sense that the EU hasn’t just fallen behind in tech within the region, but also any significant hold on global business at scale. If you tally up the market cap and income of the largest companies in Europe they pale in comparison to even one or two of the giants in tech that we seem to have here in spades. I’m not trying to demean any EMEA investments here, it’s just the reality of where we are now. We took huge risks in the past that got us here, for the most part, no single EMEA country has done the same. We’re like the lottery winners here with all the stupidity that new money seems to bring with it.
All this being said, it is something that is changing, there are new startups in tech there. It will take significant investment to expand upon them, however. And we could just as easily make huge investments that fail.
If you ever played the game Axis and Allies you can remember paying for the die rolls that granted a random chance at a new technology. That’s essentially what got us here.
The UK is in EMEA, although it may not feel like it sometimes! Plenty of stuff got invented and initially commercialised here, but it all seems to end up in the hands of US corporations sooner of later.
Plenty of tech jobs though!
I dont know about the EU, it feels like EU companies are less willing to sell out to big tech, but then dont get much global reach.
