[aurareturn]

Silicon Valley is arguing that TSMC isn't investing enough. They should be investing hundreds of billions to build fabs, like how big tech is investing in the AI buildout.

$45 billion for new fabs is peanuts compared to Amazon's $200b and Google's $180b investment in 2026.

Can't really blame TSMC though. It takes years for fabs to go from plan to first wafer. By the time new fabs go online, demand might not be there. Who knows?

[runjake]

According to Elon during his recent Dwarkesh podcast appearance[1], TSMC is limited by resource constraints (fab components, contractors, etc). His claim is that TSMC is building as fast as they can and they are unable to meet industry demand.

Seems legit to me. Nonetheless, I think it's a solvable problem.

1. https://www.youtube.com/watch?v=BYXbuik3dgA

[throwworhtthrow]

If this is actually true, I think you can find a more reliable source than Elon Musk.

I'm not saying you should never listen to a word he says. His actions shape the world after all, so it's important to understand how his words precede his behavior. But I'm baffled why anyone would take Elon at his word, or even slightly hedge their perception of reality based on Elon's claims of fact.

[runjake]

I was leaving an HN comment, not writing an essay. I'm not fond of Elon's personality, but I listened to the context of the conversation and believe him.

Did you listen to the conversation? There was a great amount of detail. Which parts of the conversation seemed unbelievable to you?

Regardless, it's also been reported in the press over the past quarter, and TSMC's previously largest customer, Apple, notably has had to make fab adjustments and form new partnerships with Intel.

https://stratechery.com/2026/tsmc-risk/

https://www.eetimes.com/tsmc-will-struggle-to-meet-ai-demand...

And even the TSMC CEO himself has acknowledged it on multiple news sources. Here's just one:

"Demand is 3 times higher than what TSMC can produce"

https://wccftech.com/tsmcs-ceo-admits-chip-production-is-ins...

Hopefully, the CEO of the company in question is good enough for you?

[throwworhtthrow]

> Did you listen to the conversation?

No, and I'm sorry for derailing your point. Thank you for the additional links. I skimmed them all but didn't see anyone corroborate the claim that TSMC is limited by its upstream component suppliers, rather than its own factory underinvestment in prior years. Am I misunderstanding, and those two things are the same?

[runjake]

If you dig deep into the links I posted, they are reportedly constrained by construction of fabs along with acquiring enough ASML machines.

But, you are certainly correct that factory underinvestment is likely the primary cause for being in this predicament.

[anonymars]

Ah, that "lays off 50,000 workers because of overhiring" oracle-of-farsight big tech?

Little easier than "laying off" a billion-dollar fab, isn't it?

[rasz]

> Amazon's $200b and Google's $180b investment

Last time I checked you cant build Chip Fabs with cloud credits.

[rwmj]

"Silicon Valley" doesn't get to make the decision unless they are willing to send some of those hundreds of billions to TSMC up front. (TSMC isn't going to want future promises of business either since those are worth very little.)

[aurareturn]

I don't disagree. I wrote the top comment here basically saying the same thing: https://news.ycombinator.com/item?id=46764223

If big tech prepays for the entire fab, I think TSMC would do it.

[toss1]

And if the Big Tech companies think it is so important to get all those compute and/or memory chips sooner and in larger supply, it should be no problem at all for those Big Tech companies to pay for the costs and then have priority access to all (or their portion of) the output for the future years.

OTOH, if they are insisting on not investing their funds or stock, and it is simply pressure on TSMC to take on the risk, TSMC should be very wary of taking on risk for those players (unless TSMC sees another advantage of producing into a likely glut or supply canyon shortly after the new fabs come online).

[baq]

if what Elon recently said is true (if - but he might not be... inaccurate... on this particular thing) they already have and bought the forward production capacity of those new fabs and it still isn't enough.

[aurareturn]

I believe that. TSMC would have to start another fab or two.

PS. I'm pretty sure Intel is also at max capacity. They cancelled a bunch of fabs a few years ago when they were on a spiral.

[slumberlust]

Actual spending already out the door or pledges? Big difference vs. money spent and money planned.

[kldg]

it all takes years. it takes years for permitting to open up the power plant to run the chips. at the scale the Big 3/4 (google, amazon, microsoft, and meta-ish) are going, we don't actually have the capacity to BUILD the capacity, despite a forecast of just 1% national electricity consumption growth this year, partly because we were expecting electricity demand to slow down and for an orderly shutdown of our fossil fuel plants. we couldn't even fill >100GW of gas/coal turbine orders over the next 5 years if we had to, and we might have to, because some of our grids (notably PJM's) are forecast to be under their safety margin of over-production in the following years.

meanwhile, regional grid operators are faced with Big Tech driving tens of % of total power into private contracts where there's only one customer; they are making the decisions normally reserved for nation-states, right? reopening Three Mile Island sounded like a pipe dream a few years ago. I hear They have something like 50 more experimental, small-scale NPPs they want to fire up across the country in the next few years, too (but despite sounding like a big boon for energy, they're ~meaningless short-term in the face of how much demand we're looking at). -so this power (uh, literally) gets wrested away from the grid authorities and from what was largely the domain of government, to now be managed by techbros and a select few partners who will be reliant on their money; I'm sure that will work out fine.

anyway, part of the reason it does make some sense in the US for the government to push for more coal/LNG turbines, is because they're already there and we need them now; the permitting to un-mothball, prevent mothballing, or expand facilities, is far less arduous than what a company'd have to go through for a new facility (tho again, we don't have capacity to build all the turbines we require inside 5 years anyway). I'm not saying it's a good idea to start sending up more GHGs, but it's maybe better than pricing out electricity for residences and "real" industry. hey, who knows? maybe they'll simply build natural gas pipelines that don't leak this time.

-oh, and then there's the problem with these new datacenters disrupting the traditional power demand curve, because they don't really do as much peak draw anymore; their peak draw is approaching base load, as LLM batching (when a company has a bunch of stuff they want processed and can wait a day for it to run in "off-hours") is sold, and if unsold, that time can be used as training time; so the modern datacenter is a 24/7/365 organ; the heart, powering our society, Moltbook. the importance of this is it makes solar less financially attractive, because now we need to be able to bank more energy since more demand's shifting to overnight. we might also want to consider just getting the moon really, really hot? then we can get a truly substantial haul of lunar light for our panels. you know, we decided against nuking hurricanes again recently; maybe we could build some new ones and nuke the moon, a lot.

[phil21]

> because they don't really do as much peak draw anymore

This is the same for more or less all major industrial users of electricity. Typically it's a boon for a power grid and overall lowers prices due to the stable consumer that helps you achieve very high capacity factors on your generation side. Large industrial users typically pay for a "max usage rate" (e.g. they commit to 200MW and will always pay for 200MW even if they only use 180MW average that billing period) due to the infrastructure needed to serve them - so it's as close to guaranteed money a power grid operator is likely to ever get.

If we want to re-industrialize the nation to any meaningful degree, we are going to need more baseload. AI datacenters may be a bubble, but if we can't somehow leverage the unlimited free money being poured into this space to augment our electric grid and build generation capacity for the first time since the Greatest Generation, we will have entirely failed as a society. The fact we have made it more compelling for folks to work out private deals with nuclear power plant operators to go behind-the-meter vs. just taking it from the grid is utterly absurd and shows how absolutely impossible it is to get anything done these days. These were last-ditch options after operators got frustrated trying to do things the usual way with years to decades of delays.

What is really happening at a very high level zoomed out: As a nation we decided to stop investing in energy infrastructure for over 50 years, and we are now reaping what we have sown. Eventually you run out of the previous generation's infrastructure investments, and also run out of cheap parlour tricks like sending industry overseas and focusing on energy efficiency vs. actually building stuff.

We get to figure out how to build things again or die trying. The AI bubble has only brought the demand forward a few years - anyone paying attention to this sector knew grid instability was effectively written in stone without major changes. Take our electric grid out of the hands of politicians and put it back in the hands of engineers and planners that actually can do things. You can't even build a transmission line of any length or size these days without a decade or more of legal battles and NIMBY. Good luck with the actual size of investment we need today.

tldr; We deserve all the pain we collectively get. You can only ignore problems for so long to take short term gains. Chickens coming home to roost in this arena, among many others. Once you stop investing in the future, the future eventually comes for you.