A lot more people are organizing China tech trips these days, which I actually think is great. More people should come see what’s happening here firsthand. But here’s what I think makes @techbuzzchina different: deep domain expertise. Our trips didn’t start as tours. They grew out of our work with fundamental investors who needed to meet companies face-to-face, ask detailed questions, pressure-test what they were hearing, and understand the environments in which these companies actually operate. Today we work with a much broader group: VCs, founders, executives, and others across tech and investing. So the trips have expanded too. We might visit a factory, an R&D facility, or a robotics lab alongside closed-door company meetings. Every trip is built around what the group actually needs to learn. But the underlying model hasn’t changed. We’re not general business, marketing, or China consultants who moved into tech because AI, robotics, semiconductors, and other deep-tech sectors happen to be where the action is now. I’ve worked in tech, finance, and investing for more than 20 years. I know the companies, the technologies, the investors, and the ecosystem. And because I live and work in Silicon Valley as well, I understand what people coming from overseas already know, what they think they know, and where the real knowledge gaps are. That means we’re not just arranging access. We know who is worth meeting, what is worth seeing, what questions to ask, and how to put what you hear into context. That’s the product. More info below:

Hung out with a friend in the Bay whose teenager is in a single club sport now, and I learned that even in the off-season, their child is doing at least three, usually four days of practice in the evenings lasting at least two hours. During competitive season, it is pretty much every day, including plenty of travel on the weekends. And their kid is not national level or anything like that. I find this absolutely insane. The grandparent was complaining to me about how the kid was actually very gifted in a few other things (or so they thought; they showed me some work). They were unhappy about their grandchild having to give up all those other interests simply because there is no time. Academically, while the kid is doing well at grade level, they do not have time for any acceleration. Compared to what they know about kids in Asia, the grandparents feel their grandchild is actually quite a bit behind. The main thing for me was just hearing how much time my friend spends on this one sport. We talked about how much value this actually adds to the kid's life, and they said it does not really matter anymore: it has become a way of life, all the kid's friends are in it because of the time commitment, and they cannot really leave. In addition to school, it completely defines their existence. My husband and I are seriously discussing this because my friend basically fell into it without realizing where the journey leads, and we do not want to make the same mistake: starting off with an innocuous sport, having a child who is decently but not remarkably talented at it, and ending up spending all their time on it to the exclusion of everything else just because those are the rules of the game now. It feels like there is no choice. I would love to hear how other parents are solving this. It is not that I want to keep my kid from playing sports, but I want it to genuinely serve my child's growth and personal development, rather than just enriching whatever organization or ecosystem benefits financially from all these "competitions" and "camps."

If you follow AI, you probably know most of the information in this presentation, but it was nonetheless helpful to see how a top fund communicates the narrative. After all, it is all about making people understand, convinced, and aligned with you on what is happening. The most useful "proprietary" data point in here was easily the seven-company sample, where Sequoia noted they were able to get a $110 million post-money valuation on average, while the company raised just a month later at a $3.4 billion valuation from what they call a "capital partner." Sequoia is positioned as a firm that helps you build the company rather than just finance it, so their capital comes at a much more "strategic" valuation. They pointed out that this is froth in the market, and it really seems like there is a lot of dumb money sloshing around. You have to be careful which valuation you are going in at, because, as they are suggesting, you are not getting in at the Sequoia price, which is obviously something Sequoia really needs its LPs to know. There's at least one point I disagree with, but keep in mind that this is a marketing presentation (as all presentations to LPs are). I'm appreciative of the share.

I don't understand these people who can't find uses for their Muse AI agent. They must not be moms. P.S. I actually don't trust Meta with my data security, so I created a whole new mailbox and some forwarding rules so it can only see a portion of my communications. But just connecting that, there are so many household tasks and organizational things that you can do. I'd probably have it doing 20x what I'm doing now if I fully trusted it. And I'm someone who consumes a couple hundred million tokens a day running AI agents for my business.

Both @Zai_org and @AlibabaGroup have made pretty explicit statements about RSI recently: Zhipu founder and chief scientist Tang Jie said last week that they’re already seeing its earliest form. Their example is actually pretty concrete: a GLM-5.3-powered agent helped build and optimize the inference infrastructure for GLM-5.3-Flash across 100,000+ domestic chips, and in under two weeks increased end-to-end throughput to ~3.2x the initial baseline. The model was basically diagnosing bottlenecks, modifying code, running experiments and iterating from the feedback. Tang Jie is careful to say this is NOT RSI yet. Humans still choose the objectives, boundaries and risks. But he says the endpoint, if this continues, is a system that can autonomously design and train its own successor. Zhipu is calling the broader roadmap “Fully Self-Training.” OTOH, Kimi and DeepSeek have been much less explicit about this publicly. For now, anyway.

Somehow I am pretty confident that this idea has been pitched and that people have funded it.

So you know how yesterday I was oohing and aahing a little bit over Alibaba CEO Eddie Wu's statement that they want to 10,000x human intelligence by leveraging machine intelligence? Well, just talked to an AI-native startup who is convinced that their handful of engineers was doing the work of a couple thousand. (It's headed by someone who used to manage a few thousand engineers so I believe them.) Already like nearly 1000x leverage. I need to get my own business / self there!

I feel like I only have two modes of existence now: either rolling out of bed and putting on sweats (and maybe remembering to comb my hair), or getting really dressed up for an event or TV appearance. I know this confuses the heck out of my kids' daycare teachers, who often can't recognize me: "Are you sure you're so and so's mom?"

Dinner with America's top CEOs

What Xinhua state media is reporting occurred on US-China AI discussions. I don't think anything substantive happened nor was it possible to. "Stability" AKA "let's reset and not make things a lot worse" was probably the main goal. And I don't want to read too much into it but "each side should play to its strengths" sounds like more open source, more domestic chip ecosystem building to me. "Xi Jinping noted that both China and the United States are major powers in artificial intelligence. While there is competition between the two sides, there is also room for cooperation. Each side should play to its strengths rather than erect barriers against the other. The two sides can continue their dialogue on AI, exchange views on its risks and benefits, and work together to prevent its misuse and abuse. Humanity must maintain control over AI technology, uphold a people-centered approach and the principle of using AI for good, and ensure that AI serves human progress."

From someone in the comments whose posts are private: "you should describe what actually happened, which is usually that an optimization process exploited a vulnerability in the sandbox design" I personally think the word "exploit," while accurate, even has some anthropomorphic connotations. I think maybe a better way of saying it might be: "A model optimization process utilized a vulnerability in the sandbox ..."

Oh, look! That person funded by U.S. EA AI money telling you that the Chinese government doesn't care about cyber risk, when literally the government has warned the entire public about it, like just last week. I LOLed Oh, to be clear, I'm not talking about the author, nor Matt or Kyle

So apparently, President Xi said in the next five years they hope to welcome 100,000 American youth to do exchanges and study in China. China has amazing resources for STEM kids. If you're one, you should consider taking advantage of the ecosystem, and we'll take you there. As you probably know, we completed two successful pilot cohorts for our @TechChinaTrek this past summer for 10 to 18-year-olds. Right now, it is just visiting facilities and doing some simple STEM workshops. But for the more hardcore kids who actually can and really want to make something, don't worry: we're working on programming for you.

So Alibaba $BABA had their big Apsara conference earlier this week in Hangzhou, and I thought CEO Eddie Wu’s speech was actually pretty interesting, so just pulling out some of the numbers that jumped out at me, and my thoughts: • <3% → 1,000x: By their estimates, machine thinking today is still less than 3% of human thinking. Alibaba thinks eventually machines will produce more than 1,000x as much thinking as all humans combined. When you put it that way, wow, I honestly can't wait to see how the world changes. • 99.9%: Eventually machines could do 99.9% of all thinking. That's just an arithmetic rephrase of the above statement, but might be easier for people to remember, lol. • 10,000x: Wu thinks each individual could have 10,000x the intellectual leverage we have today. I was thinking about this, and I definitely feel like I've probably doubled or tripled my output with AI, but I hadn't thought about the possibility of getting to 10,000 times. That would be insane. Has anyone tried to measure their "intellectual leverage"? I'd love to hear about it. • Millions of agents: His example is designing a spacecraft to Mars, where the problem gets broken into tens of millions of subtasks carried out by millions of agents. It's gonna be wild when this happens. • 5–10 trillion parameters: This is the scale Alibaba says it is targeting for future Qwen models. This is what ByteDance and DeepSeek are allegedly looking to do as well. • 20 GW: Alibaba Cloud says it plans to have more than 20 GW of global data center capacity by 2032. For reference, OpenAI's Stargate is supposed to be about 9 gigawatts by 2029. Also liked this line: “AI coding is simply the light bulb of the Machine Intelligence era.” Basically coding is one of the first obvious uses we have found for all this intelligence, but that doesn’t mean it’s what most of it will ultimately be used for. If we go with the metaphor that AI is electricity and coding is a light bulb, electricity is going to be used for way more things than just light bulbs. We know it's the first use case for electricity, but when's the last time you noticed a light bulb?

We just had our back-to-school night for the Montessori that we send our kids to, and the teacher talked about how proud the kids get when they are able to learn and finish something. They feel really grown up. Because it is Montessori, they are explicitly explained a job and then asked to do it, but for much of the day they get to choose which jobs they play with. They're called "jobs," but they are really just tasks or games, and we love this. We explicitly did not want to send our kid to a "play-based" daycare, and recently I'm ever more convinced we're correct. A friend who did send their kid to one said the teachers there were super proud that they don't teach the children anything. They told them, "By the age of five, maybe we'll let them know what 1, 2, 3, 4, 5 is, but don't worry, we will never ask them to learn to write it or anything like that." I find that totally ridiculous. Research shows that if you challenge children, they actually love it. When you define play so narrowly as not learning anything even slightly "academic," just playing with toys, props, or whatever is on the playground, it misses the point: learning should be play. To kids, learning IS play. It's nonsense to define learning how to write, for example, as "not play," and in the US, somehow increasingly "not OK." And by the way, some kids like worksheets. There's this one five-year-old in the class that, in the pictures I've seen, they make up math problems for themselves to solve during free time. And others might be taking a book and trying to read it together. To me, refusing to teach them how to write with a pen and paper is just as stupid as forcing them to write all day. And mastering something is how you gain real confidence, not through this fake "I am great, I am special, I am smart, I am unique" mantra that I see teachers having kids repeat to themselves. The more you figure out how to acquire a new skill and do it independently and autonomously, the more confident you become. That's true for everyone, whether you're 3 or 80.

EA seems pretty pro-America to me. Also, what? Why this specific mention of shrimp? Is it just to heighten the absurdity of it all? "The memo goes on to criticise the movement for prioritising "foreigners over citizens, shrimp over families ..."