Full episode: • Dylan Patel – Two labs will soon control m...
Me on twitter: https://x.com/dwarkesh_sp
What Dylan Patel said
Written from the verified transcript and checked against it. Every figure links to the moment it was said.
Dylan Patel said that since 2022, US export controls and financial systems have dramatically shifted compute deployment. Today, 70% of watts deployed for data center AI compute are in America, while China is sub-10% and will remain so, likely reaching 30 gigawatts of AI compute or less by 2028. He thinks China will hockey stick, with domestic production from SMIC and CXMT adding 5-10 gigawatts in 2028 alone, and it's reasonable they could add 50 gigawatts in 2029, though those chips are worth much less than Western ones. Patel noted Chinese labs like 01.AI have at most 100-200 megawatts of compute, while Anthropic is nearly 5 gigawatts. He argued that compute differences matter less now because pre-training runs like Anthropic's Mythos use sub-200 megawatts at any point, with most compute going to research, and this may shift as automated coding and continual learning increase training share.
Key takeaways
China will have 30 gigawatts of AI compute or less by 2028, sub-10% of global incremental new compute.
In 2028, China's domestic fabs (SMIC, CXMT) will add 5-10 gigawatts of chips, which are worse than Nvidia's or Google's.
China could add 50 gigawatts in 2029, but that would be worth only about 20 gigawatts of American chips.
Anthropic is nearly 5 gigawatts by end of year, while leading Chinese labs (except ByteDance Seed) have 100-200 megawatts total.
Anthropic's Mythos pre-train used less than 200 megawatts at any point, with most compute going to research, not training.
Numbers and commitments
Figure
What it refers to
Type
At
70%
Share of watts deployed for AI compute in America today
7:14Pre-train compute of Mythos and research allocation
Questions asked in this interview
6
0:00And how does like—yeah, how does Chinese compute continue increasing through this whole trend?
1:27And then how fast does their hockey stick go up?
2:17Basically, I want to know: do they just hockey stick at the point at which they are able to start shipping large amounts of compute, or is it still going to be less than US plus allies?
3:26But do you think China could do adding 50 gigawatts by 2029?
7:21Do you think the RL was less compute than the pre-train?
If you think there's 200 gigawatts globally in 2028, how much is in China by that point? And how does like—yeah, how does Chinese compute continue increasing through this whole trend? Because if the RSI stuff kicks off in the West before China has a large amount of compute, maybe we're living in a different world than when it doesn't.
Yeah, so China today, if we sort of level set back to 2022, the US was adding about 45 to 50% of the world's compute. China was adding about 30 to 35% of the world's compute, and the rest being taken up by the rest of the world. Since 2022, we've had big regulations against China and a dramatic increase in America. So today, 70% of watts are being deployed in America. And, you know, China is really a very small number. It's sub-10% of watts being deployed for data center AI compute is in China. And as we step forward, they're still at a very small number. Their domestic production is quite small. Their purchasing from Nvidia is still quite small. And a lot of that ends up in other places as well, right? Malaysia or what have you. So ultimately, China domestically still continues to have sub-10% of incremental new compute. So in 2028, it might start to inflect up, I think, but it's pretty easy to say China will have like 30 gigawatts of AI compute or less.
I do think in 2028, they have a big uplift in what compute they're able to deploy. In 2026, they're still mostly relying on a lot of the smuggled chips, a lot of the chips that TSMC made for companies that they thought weren't Huawei, but ended up being Huawei, or a lot of HBM that Samsung is shipping. But in '27, and in '28 especially, fabs start to go up from SMIC and CXMT and such where domestic production is actually reaching many millions of units a year. And now they're incrementally adding 5 to 10 gigawatts in just 2028 of domestically produced chips. Those chips are definitely worse than the chips that Nvidia will have in '28, or Google will have in '28, or OpenAI will have in 2028.
So even the gigawatt number overstates things, you're saying. Like it's 30 gigawatts, but it's really much worse chips. But then, if you think the world is going to add 100 gigawatts the following year, or something—I know you said you can't really say that far out. How much is China able to add the subsequent year? Basically, I want to know: do they just hockey stick at the point at which they are able to start shipping large amounts of compute, or is it still going to be less than US plus allies?
There's a lot left to whether or not the US passes the MATCH Act, whether or not tools continue to get export controlled, and how fast China can build their new equipment that they're starting to be able to produce domestically. But ultimately, China is definitely going to hockey stick. If there's anything China's really good at, it is scaling manufacturing really, really quickly. And I imagine China will start to be able to extract more and more purchasing of even foreign chips into domestic China, or at least close the gap in what the US is allowing Nvidia to sell them or what have you.
But yeah, I think it's completely reasonable that China in 2029 can do 50 gigs. But if most of those are domestic chips, there is some factor there where that 50 gigawatts is really worth as much as 20 gigawatts in America, or from American chips.
Right, right. So you're actually projecting a world where maybe the leading lab in 2028 has more compute than all of China will have in '29 or even '30, if you weight gigawatts by their quality.
14 more exchanges in this transcript
Sign in free to read the rest of this interview. No card required.
