Alibaba Chairman Joe Tsai Spoke at Edward K Y Chen Distinguished Lecture 2025

We won one game and lost one game.

Professor Tang, before I get into the question, I first just want to express my deepest appreciation. I feel extremely honored to be able to be on this stage. I'm no Jack Ma, but to have this opportunity to share ideas — when you guys approached me, this is a lecture series and I said I didn't want to stand up on the podium to lecture people. I would prefer to do a fireside chat format so that we could sort of exchange ideas, in honor of Professor Chen. And so I look forward to this exchange, and also I understand there's a little bit of Q&A later on with the students in the audience, so I really look forward to that.

The NBA has been in China for a long time. They've actually played games in China for many, many years. And then there was a hiatus after 2019. This year was the first time in six years that the NBA brought two teams — two NBA teams, one of them is mine, Brooklyn Nets — to play in Chinese territory. They chose to play in Macau. They have a deal with the Macau Sands organization to play in Macau for the next five years. But part of the deal is that there's a chance for the NBA to also opt to play in mainland China. So at some point I expect that in the not-too-distant future the NBA will be back in mainland China. And from the NBA's perspective, the logic is very clear. China has probably the largest basketball fan base in the world, and everybody follows the NBA. Why is that? Because the NBA is the league where the best players in the world play. About 30% of the NBA's players are non-American. They come from all over the world — Europe, Eastern Europe, Australia. Not many Asian players, but of course we hope that in the future China would develop its basketball program and we can see another Yao Ming in the NBA. So the fan base is huge in China. It makes a lot of sense for the NBA to come back to China and play games so that the players can interact in person with the fan base. And of course, during all these years the NBA's games are always broadcast in China through Central Television and also on streaming media. From China's perspective, it's equally as important for China to be engaged with the rest of the world and to bring the best parts of sports and other aspects of culture from around the world to China to interact with the fan base in China. So this time we got a huge, very warm welcome from the China side because the NBA is so popular, and the games are a huge success. We actually gave the fans a very good showing in terms of the competitive level. One game went into overtime. The other game — I think we lost the overtime game but we won by about three points the second night. So it was a great experience and I'm just very positive about this kind of cultural exchange and sporting exchange.

I just wanted to mention this idea of using sports to bring cultures together is very, very important. A big part of my philanthropy in China is all about sports — investing in sports as part of education. And I have a program to bring middle school kids, eighth graders that are about to go into the ninth grade, to send them to the United States for high school — four years of high school — to play basketball, but more importantly to get an education in the US. There's a little bit of replicating my own experience, because when I was young I left Taiwan when I was 13 years old and then I went to the US for boarding school and university and so on. And this program is actually both developing the next generation of basketball talent in China, but also giving the kids an opportunity to see what's outside. And also I think the biggest beneficiaries are the high school kids in the US. The Americans, they see these kids coming from China. They've all read about China but they haven't interacted with people, individuals. And we select the best of the best, both from an academic standpoint and also basketball skills standpoint, and they come into the US environment and they become part of the community and they're well loved in the community. So this is great. This kind of people-to-people exchange is so important. And I believe if I have the resource and ability to do that, I will always support that program.

It's very selective, competitive. So every year we select between six to eight kids.

Yeah.

We can always try. You're holding him back.

Just a little bit of historical context. When I joined Alibaba in 1999, China's GDP per capita was $800 and today it is $13,000. The growth has been tremendous. And I always tell my friends that I'm in a very fortunate position because I lived through the confluence of the rise of China as a manufacturing power and the economy of China rising, and the advent of the internet — the growth driven by the internet. Sitting at the intersection of that, I was able to, from the Alibaba vantage point, witness this sort of dual engine growth. It was quite tremendous.

Alibaba's original idea, like you said, was a B2B website, because Jack thought that he wanted to use the internet to level the playing field for all the small businesses in China — trading companies, small manufacturers — which if you recall, before China entered the WTO in 2001, international trade had to be done through state-owned trading companies. But once China entered the WTO, it opened up trade, and that was the beginning of China's development as a manufacturing base. People started to set up factories. They could see opportunities to trade using the internet with the rest of the world. So Alibaba's inception was a business-to-business website. We helped small businesses sell things wholesale. And by the way, Alibaba's first website was in English language, because it was facing buyers around the world — it was an outside-facing website. And then we evolved from B2B commerce to consumer commerce, which became the largest consumer shopping platform, Taobao today. We also developed payment because there was a pain point — buyers and sellers didn't trust each other. Buyers didn't want to pay first and sellers didn't want to ship the product. So we invented a thing called Alipay, which in its original form was an escrow system to allow the trade to happen. And then we got involved in logistics and so on and so forth.

And you ask about the secret sauce. I think the secret sauce is that a good company always develops when they follow customer needs and customer demands. We basically followed what our customers wanted and we developed everything organically. I would always — if you people in the audience here, the students, later on you will go out and found companies — you would always want to favor organic development over acquisitions. Now, of course, we've also made some acquisitions. Some of them are successful and some of them failed spectacularly. But always favor the organic development, because you're developing it through your team, and your team has the best DNA in terms of the Alibaba culture and innovation.

So that's how we sort of evolved. And then the reason we got into the cloud computing business was also to service our own need. It's not because someone decided, 'Oh, cloud is a good business, we should be in it.' That's not the case. We developed the cloud business 16 years ago. Sixteen years ago, nobody was talking about cloud computing, and we said we were looking at all the data we were managing and the consumer platform was handling massive amounts of transactions. And our CTO came to us and said if we continue to use third-party software and hardware — we were using servers from Dell and IBM, storage devices from EMC, database software from Oracle — if we continue to use that, we will later on hand over all of our profits to these technology vendors. So we developed cloud computing really out of necessity, out of the need to become self-reliant in technology, which later on we'll talk about as a sort of national thing. And so we had a team that focused on that development of an operating software — not a single computer operating software, but one that works across data centers, across multiple computers, because we need to do parallel compute in order to manage the large amount of data that we have. That's the genesis of our cloud business. We really ate our own dog food, right, and used the technology ourselves. And then later on we decided, why don't we — this technology is so good, why don't we open it up to third-party customers? And that's how we went into the cloud business.

Well, I think this is a very topical question because China just published the 15th Five-Year Plan from the Central Committee. And I think some of you might have read the whole plan. I read a summary of it. But there's really two major takeaways from the five-year plan. Number one, China wants to continue to be a manufacturing powerhouse. I think the emphasis on the manufacturing economy, which is part of the real economy, is right there. That's what the leadership in China has clearly stated — this is where we want to be. You know, you compare China to the rest of the world, consumption is actually a very small percentage of GDP, less than 40%. Whereas in the United States, consumption is 70% of GDP. So China is actually making things. China is looking at GDP growth from production and to an extent exporting that production to the rest of the world. Because China is, I believe today and 10, 20 years from now, China will continue to be the manufacturing base to supply the rest of the world. So that's number one.

Number two, the plan says we want to become technology self-reliant. I think this is very much both the foresight of the leadership as well as a reaction to what's going on with geopolitics, where the United States and some of the European countries are restricting critical technology from China. So China believes that we have to develop our own technology, which China is already today very much well on the path of doing that.

Now, going back to the first goal of maintaining a manufacturing base — if you look at the history of China, how China has become a wealthy country, I said $800 GDP per capita to $13,000 today, and we'll get to $30,000 in the next 10 years — the path of wealth creation has been in production and making things and supplying that to the rest of the world. And I actually think that this is — but there's a lot of criticism around that. They say, 'Oh, China has excess capacity. They're exporting excess capacity to the rest of the world as though that's a terrible thing.' Well, if you think about it, did everybody criticize the German car industry for having excess capacity when Germans exported all their cars? I mean, definitionally, excess capacity simply means you have production capacity that your domestic economy cannot absorb. Therefore you resort to exports. But the use of the word 'excess' seems to be like a dirty word, but it's not. It shouldn't be, because how a country gets wealthy is to make things and collect money from the rest of the world to make their own citizens wealthy. And I truly believe that continuing on this path of being a manufacturing center for the world — and now high-tech manufacturing, not making shoes or T-shirts — if we continue on that path, China will continue to grow its economy and the citizens will continue to gain wealth and gain disposable income, and at some point consumption will come. Today through the Alibaba platform we see a lot of consumption. I'll give you a data point: there are something like 56 million people on our platform that are spending more than $6,000 US per year on Taobao. That's way more than the average disposable income of Chinese citizens. But we already have the scale that many people are spending that amount of money. So at some point consumption will come. But I think the leadership in China understands that the foundation to healthy economy growth and wealth creation stands with having a strong manufacturing base, and today it's high-tech manufacturing. China is the best in the world when it comes to making electric vehicles, batteries, solar panels, and all those things that the rest of the world needs.

Well, obviously when you talk about national policy, China — the government has identified areas for investment. For example, in semiconductors, in semiconductor manufacturing process, in equipment that makes semiconductors, all that — I think that's all good. But the part that I focus on that I really thought was very interesting is: a couple of months ago the State Council came up with an AI plan. And Chinese people are very practical, it's very goal-oriented. The AI plan simply said by 2030, which is five years from now, we should see 90% penetration of AI agents and devices. That's what the State Council said. So they basically said, here's the goal. We'll let the market figure it out — whether the market is state-owned enterprises or private entrepreneurs, they will figure out how to proliferate and achieve high penetration of AI in China. And I think that's really a great policy, because all of us — we are all part of the so-called AI race, right? The race between China and US AI. At the end of the day, you don't keep score by looking at how good these large language models are. The score is being kept by the adoption rate. The more people that adopt AI, the more society will benefit. So China's whole approach is the proliferation of AI, and I think that's a great policy.

So right now, I think the way Americans want to define who's winning the AI race is purely by looking at how good the large language model is. One day it's OpenAI, the next day it's Anthropic. And then the Alibaba Qwen model — by the way, the Alibaba Qwen model just won a two-week contest of trading cryptocurrencies and stocks. They asked something like 10 different models, American models and Chinese models, and the Alibaba model actually became the winner of the whole thing, with DeepSeek second. By the way, huge amount of respect for DeepSeek, our neighbor in Hangzhou — they're doing incredible work.

If you compare, we don't look at the AI race according to the American definition. We look at the entire stack of where China has the advantage. You start with energy. China has an advantage in electricity generation, and that's because 15 years ago the government had the foresight to invest massively in energy transmission. When you generate electricity in the north, it has to be transported to the south, right? And when you generate electricity, especially in clean energy where there's sun, wind, and water, it's not necessarily where the electricity is needed, so it has to be transmitted somewhere else.

The State Grid in China — there's two grids, north and south — they make $90 billion of capex annually. The United States, $30 billion only. So the United States is massively underinvesting in electricity transmission, but China has been doing this for the last 15 years. The result is China's installed base of electricity generation capacity is 2.6 times that of the United States. And even better, the net additions to capacity that China is putting in is nine times that of the United States. China's growth in electricity capacity is much, much faster than what's going on in the US, and most of the net increase is in clean energy, solar power.

The whole end result of all of this is that China's electricity on a per-kilowatt-hour basis is about 40% cheaper than that of the United States. So China has an energy advantage in AI because when you burn all these GPUs to train large language models and run inference, you're burning lots of energy. The big difference between machines and the brain is machines eat a lot of energy but the brain operates on very low energy.

Then we look at how much it costs to build a data center. It's 60% cheaper to build data centers in China — this is before you buy the chips, the cheap GPUs. And then you look at model development. I actually think the Chinese models are not very far behind the United States.

There's a reason for that. China has a lot of engineers, and it's the country that produced the most STEM students every year. You would think that developing AI and training models is a very high-level research exercise, but a lot of the work is actually in the engineering.

You have to work on the system to make the system very efficient in training the model with hundreds of billions of parameters, even a trillion parameters. If the system is not efficient, it will cost you a lot of GPU resources. China being lacking in GPUs actually creates an advantage of starvation — when you don't have a lot of resources, you are forced to innovate at the systems level, and this is where China is strong, with lots of engineers and talent.

Globally, almost half of AI scientists and researchers have had a degree from a Chinese university. Globally — whether they work in US companies or Chinese companies or anywhere. What that means is you go to a company in the US, many of them are ethnic Chinese.

And I just recently saw a social media post — someone working at Meta, which is Facebook, complaining — this is not a Chinese person — complaining that the AI team that he's working in, everybody is speaking Chinese and they're sharing ideas in Chinese that he doesn't understand. But they're so smart in creating these AI tools, they should be able to real-time translate Chinese into any language. But still, in a sort of water cooler conversation where you sit down in the cafeteria, it's hard to capture everything.

So what that means is a lot of the idea sharing and exchange is happening in Chinese globally in AI. This is the first time Chinese language is an advantage. It had been a disadvantage for Chinese companies to expand overseas, because when Alibaba goes to open up an office in Italy or Japan or the US, the people we hire locally can't speak Chinese, so they speak English. Our people sitting in Hangzhou have to communicate in our second language, which is not ideal.

That's a huge hindrance for Chinese companies who expand overseas. But now knowing Chinese has become an advantage in the AI world, and that's very interesting.

Anyway, I've just listed a number of advantages that China has. But the biggest advantage is that I believe the approach that Chinese companies are taking toward large language models — which is open source — will accelerate the adoption of AI and will really realize the proliferation of AI that would benefit broader society.

And the reason open source is so important is that it is cheap. It actually doesn't cost anything to use open source models. Alibaba has published many versions in open source, and we're available in all the open source marketplaces in the United States and around the world. People can just take our model, download it, put it on their infrastructure or their notebook computer, and start running AI at no cost.