China Tech Talk

I did a very early morning quick take on my trip to China. I’m sure there will be more on this since so much stuff happened while I was there. But I really wanted to just the message that China is becoming a really livable place and people are moving there for the quality of life. I met several people in Beijing this time who moved there from America from different reasons. It was interesting to hear about why they did it. I spoke about my own thoughts on what I enjoyed about living in Beijing and why I would consider moving there more permanently.
I think there are still many reasons to not move to China. For example, the intensity of life and work is much higher there than here. The competition for everything is so much higher and this drives people toward addictions that are all about the short term consumerism satisfaction like looking at micro drama on their phones and snacking/drinking milk tea all the time.
If I had to talk about what I think is problematic in Chinese society, the phone addiction and increased short termism seem to be the most obvious ones.
And those are also problems that likely will get worse with more competition and AI. But in many ways, the quality of live in Beijing is also just so much better than here.


This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tphuang.substack.com

I had a chance to go over Tau Scaling and reusable rocket with Lei. As Lei is the person who shares my point of view the closest on Tau Scaling and Huawei’s efforts on this front, we really spent a lot of time explaining why what Huawei did in their design and production is quite special.
Even if transistor cell itself can continue to scale smaller, that does not apply to memory cell sizes or analog components like copper connections and such.
This episode was recorded before I went to China. Now that I’m back and Kirin 9050 has been unveiled, it’s a good time to publish my discussions on it.


This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tphuang.substack.com

I am currently in Beijing and have the pleasure of getting together with some friends in Tsinghua University in the AI school. I was welcomed by Alex and Qian Yu. We sat down talked a lot about the academia, AI research in China, getting talent to come to China and future of AI both in LLM and embodied AI.
It’s quite the interesting experience to come here and see Zai in one corner, Unigroup building in another corner, the school for automation (using unreal new technology), Galbot robot booth, Infinigence, Vidu AI, Robotera and many other AI and semi startups.
Pay attention to the tips they have about the transitioning toward Ascend from Nvidia chips by researchers. Also, check the part about researchers in China finally finding Kimi K3 to be good enough for vibe coding compared to GPT and Claude. There are still some concerns about inference sped for Kimi. I think they need to get more compute.


This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tphuang.substack.com

I was fortunate to be invited by Angelica Oung last week to talk about China’s AI progress in light of Kimi K3, DUVi lithography new, CXMT IPO and the KOSPI market crash.
Things have changed a little bit since the podcast, but CXMT remains very important, K3 remains the top Chinese model and lithography is still a big story.


This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tphuang.substack.com

My quick thoughts on DSV4-Flash: Huge improvement over the original. It took 3 to 4 months, but we essentially got something close to GLM-5.2 in software engineering tasks.
I did some testing today on VS Code using cline plugin. DSv4-Flash is essentially free, so I’m going to keep it as a backup to GLM & Kimi’s latest models, since those coding plans are going to get used up quickly.
Going forward, I see 2 paths for open src/weight models:
* Smaller models that can run locally with some souped up hardware.
* Larger models that can need SuperNode to run and can compete with closed source flagship models.
All the developers are going to want to run these smaller LLMs locally. You don’t have to worry about AI company getting your data and can fine tune them anyway you want. This is going to drive an acceleration in “souped up” local hardware. Very bullish in my opinion for memory chips.
For the larger models, I think there really isn’t a lot of room for middle ground. You need to make them larger to compete fully with the Anthropic and OpenAI models.


This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tphuang.substack.com