High Signal Podcasts Evidence ledger
Method
Browse
← Back to evidence

Evidence receipt / belief

Published · transcript-backed

Dwarkesh Patel: belief

13 Mar 2026 Dwarkesh Podcast Dylan Patel — Deep dive on the 3 big bottlenecks to scaling AI compute

“If you’re Jensen or Sam Altman, or whoever stands to gain a lot from scaling up AI compute, there are these stories that they’d go to TSMC and say, “Why can’t we access Y and Z?” But I think the point you’re making is that it doesn’t really matter what TSMC does in some sense.”

— Dwarkesh Patel

Source trail

Everything needed to verify it.

Speaker
Dwarkesh Patel
Attribution
Verified speaker
Claim type
belief
Recorded
13 Mar 2026
Publisher
Dwarkesh Podcast

Transcript context

…Right, but the percentage of cost— It’s a proxy, yeah. If you’re Jensen or Sam Altman, or whoever stands to gain a lot from scaling up AI compute, there are these stories that they’d go to TSMC and say, “Why can’t we access Y and Z?” But I think the point you’re making is that it doesn’t really matter what TSMC does in some sense. In fact, even if you have Intel and Samsung building more foundries, in the long run, you’re going to be bottlenecked by ASML and other tool and material makers. First, is that a correct interpretation? Second, should Silicon Valley people be going to the Netherlands right now to try to pitch ASML to make more tools so that in 2030 they can have more AI compute? It’s a funny dynamic we saw in 2023, 2024, and 2025. People who saw the energy bottleneck before others asymmetrically went to Siemens, Mitsubishi, and of course GE Vernova, and bought up turbine capacity. Now they’re able to charge excess amounts for deploying these turbines in places because of energy. In the same sense, this could be done for EUV, except ASML is not just going to trust any random bozo who wants to buy EUV tools. These turbines are much cheaper than EUV tools, and there’s many more of them produced. Especially once you get to industrial gas turbines, not just combined-cycle but the cheaper, smaller, less efficient ones, people put down deposits for these. Someone could do this. Someone should go to the Netherlands and be like, “I’ll pay you a billion dollars. You give me the right to purchase ten EUV tools two years from now, and I’m first in line.” Then over those two years, you go around and wait for everyone to realize, “Oh crap, I don’t have enough EUV tools,” and you try to sell your option at some premium. All you’re effectively doing is saying, “ASML, you’re dumb. You weren’t making enough margin on these. I’m going to make a margin.” The question is, will ASML even agree to this? I don’t think so.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.

Search evidence