Dylan Patel — Deep Dive on the 3 Big Bottlenecks to Scaling AI Compute

Bullish:

• $NVDA - NVIDIA has locked up majority of advanced semiconductor capacity and memory, creating massive competitive advantage as AI compute demand explodes.

  • “NVIDIA is now the largest customer at TSMC, and NVIDIA is the largest customer at SK Hynix, the largest memory manufacturer.”

    Dylan·36:26

  • “NVIDIA just did it way earlier than Google or Amazon.”

    Dylan·28:45

  • “by '27, NVIDIA is gonna have like 70 plus percent of N3 wafer capacity or something like that, or around that area.”

    Dwarkesh·25:02
• $INFRA - AI infrastructure spending reaching unprecedented levels with massive returns on invested capital as revenue growth accelerates.

  • “Amazon, you know, 200,000,000,000, Google 180, you know, so on and so forth. All these companies are spending hundreds of billions of dollars of capex. You know, there's there's, you know, nearly a trillion dollars of capex being invested in data centers in America this year, roughly.”

    Dylan·1:13:08

  • “the ROIC on these data centers is super high.”

    Dylan·1:13:51

  • “If Anthropic was not on track to have five gigawatts by the end of this year, but it needs that to serve both the revenue that's gone crazier than expected.”

    Dwarkesh·4:04
• $ASML - ASML controls the ultimate bottleneck for AI chip production with EUV lithography tools, with demand far exceeding supply through 2030.

  • “ultimately by 2829, the bottleneck falls to the lowest rung on the supply chain, which is ASML.”

    Dylan·37:09

  • “ASML makes the world's most complicated machine, I. E. An EUV tool, and the selling price for those is $304,100,000,000 dollars.”

    Dylan·37:16

  • “Even under very aggressive supply chain expansion, they only get to a little bit over a 100 by the end of the decade.”

    Dylan·37:29

Bearish:

• $AAPL - Apple facing severe headwinds from memory price increases and losing priority access to advanced semiconductor nodes as AI companies outbid them.

  • “Apple has to either pass it on to the consumer, a, or b, they have to eat it. I don't see Apple reducing their margin too much. Maybe they eat a little bit. But at the end of the day, that means the end consumer is paying $250 more for an iPhone.”

    Dylan·1:25:00

  • “Apple will become a smaller and smaller and smaller percentage of TSMC's revenue, and therefore be less relevant for TSMC to cater to their demands.”

    Dylan·2:20:09

  • “Not only will Apple not be the first to a node, they will also not be the majority of the volume to the new node, and then they'll just be like any old customer.”

    Dylan·2:22:02
• $RETAIL - Smartphone and PC volumes collapsing as memory prices triple, with low-end phones being cut by 50% as AI consumes semiconductor supply.

  • “smartphones and PCs are not gonna get incrementally better year on year, and in fact, they're getting incrementally worse.”

    Dylan·1:24:13

  • “Used to be 1,400,000 smartphones were sold a year. Now we're at like 1.1, but our projections are we maybe get down to like 800,000,000 this year, and next year like 600 or 500,000,000.”

    Dylan·1:25:44

  • “Xiaomi and Oppo are cutting low end and mid range smartphone volumes by half.”

    Dylan·1:26:05