transcribe

The Export Controls Are Working - Dylan Patel

Dwarkesh Patel · 1m · transcribed 19d ago
More from Dwarkesh Patel Business
𝕏 Share ▶ YouTube 📥 PDF 🤖 .md

Section Insights

# 0:00

Global Compute Landscape in 2022

What was the distribution of global compute capacity between the US and China in 2022?

In 2022, the US accounted for about 45% of the world's compute capacity, while China contributed around 30%. The remainder was shared by other countries.

  • The US was the leading contributor to global compute capacity in 2022.
  • China was the second-largest contributor but significantly behind the US.
  • The rest of the world accounted for the remaining compute capacity.
# 0:14

Shift in Compute Deployment

How has the deployment of compute resources changed since 2022?

As of now, 70% of compute watts are being deployed in America, with China contributing less than 10%.

  • There has been a dramatic increase in compute deployment in the US.
  • China's share of compute watts has significantly decreased.
  • The US is dominating the data center AI compute landscape.
# 0:29

Future Projections for China's AI Compute

What are the projections for China's AI compute capacity by 2028?

By 2028, it is projected that China will have around 30 gigawatts of AI compute or less.

  • China's AI compute capacity is expected to remain low through 2028.
  • The projected growth for China in AI compute is minimal compared to the US.
  • Future developments may lead to an inflection point for China's compute capacity.
# 0:43

Cooperation vs. Competition with China

What are the implications of cooperating with China in the tech sector?

There is a perspective that cooperating with China could be beneficial, especially given their control over supply chains necessary for robotics and other technologies.

  • Cooperation with China could lead to mutual benefits in technology development.
  • China's control over supply chains is significant for various tech sectors.
  • The compute situation may influence the dynamics of cooperation.
# 0:58

Impact of Compute Capacity on Technological Advancement

How does China's compute capacity affect its technological progress?

If China's compute capacity remains as projected, it will lag significantly behind in advancements like automated coding and research.

  • China's compute stock may hinder its technological advancements.
  • The difference in compute capacity could lead to notable successes for other countries.
  • Technological progress is closely tied to compute resources.

Transcript

0:00 So, China today, if we sort of level set back to 2022, the US was adding about 45% of the world's compute. China was adding about 30% 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 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.

0:32 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. >> Wait, during 2028? >> Yeah, in 2028. >> Honestly, when I interviewed Jensen and asked about export controls, I was still manning what is like the opposite view that he has. But, I'm like, yeah, maybe there's a world where we if we just cooperated with China, it would be better for us, especially since they control so much of the supply chain and the other things that will be needed for robotics and other things. But, I didn't realize the compute situation was as as you're saying. If they ship the amount that you're saying, that's a huge difference. By the time we have automated coder and even getting into like automated researcher, China is like way far behind on the compute stock. If that ends up being the case, that would have worked. I I think that's actually a notable success.

Summary

The discussion highlights the significant shift in global AI compute capabilities, particularly between the U.S. and China. As of now, the U.S. dominates with 70% of the world's compute resources, while China's share has dwindled to under 10%. Projections suggest that by 2028, China's AI compute capacity may only reach around 30 gigawatts, indicating a substantial gap in technological advancement.

- In 2022, the U.S. contributed 45% of global compute, while China contributed 30%.
- Recent regulations have increased U.S. compute deployment to 70% of the global total.
- China's share of AI compute has fallen to less than 10%.
- Predictions for 2028 suggest China may only achieve around 30 gigawatts of AI compute.
- The disparity in compute resources could hinder China's progress in automation and AI research.
- The conversation reflects on the potential benefits of cooperation with China, despite current competitive dynamics.
- The speaker acknowledges a notable success for the U.S. in maintaining its lead in AI compute capabilities.

Questions Answered

What was the distribution of global compute capacity between the US and China in 2022?

In 2022, the US accounted for about 45% of the world's compute capacity, while China contributed around 30%. The remainder was shared by other countries.

How has the deployment of compute resources changed since 2022?

As of now, 70% of compute watts are being deployed in America, with China contributing less than 10%.

What are the projections for China's AI compute capacity by 2028?

By 2028, it is projected that China will have around 30 gigawatts of AI compute or less.

What are the implications of cooperating with China in the tech sector?

There is a perspective that cooperating with China could be beneficial, especially given their control over supply chains necessary for robotics and other technologies.

How does China's compute capacity affect its technological progress?

If China's compute capacity remains as projected, it will lag significantly behind in advancements like automated coding and research.

© transcribe · For agents Built with care and craft by Gokul Rajaram