Pando

In six years, Kai-Fu Lee says China has gone from mere copy cats to potential global AI domination. What changed?

By Sarah Lacy , written on May 24, 2017

From The China Desk

Kai-Fu Lee is a master of transcontinental tech self-promotion.

Everyone in Beijing and Silicon Valley will tell you this, some of them with an envious sneer. I mean it as an unabashed compliment.

I’ve known Lee for the better part of a decade, first when he was getting attention as a high profile Google hire in China, away from Microsoft where he’d spent a good deal of his career, and later doing early reporting on his new incubator and fund Sinovation Ventures.

We even partnered together to produce the very first Disrupt Beijing while I was still at TechCrunch. It was a remarkable event produced and streamed in two languages with real time translation headsets and speakers like Kevin Systrom, Phil Libin, Tencent founder Pony Ma, and Xioami’s founder Lei Jun

I caught up with Lee yesterday for the first time in person since that event -- a lifetime ago in startup years. He was in the Bay Area this week, producing an event and doing a ton of press around his rebranding as China’s premier investor in Artificial Intelligence.

You can read about his bullishness about AI in most tech publications this week. Like I said, he’s good at getting the word out. But there are a lot of China bulls. What struck me most in our conversation was his rapid transformation from someone who believed Chinese startups were mostly copycat artists six years ago, to someone articulating a credible theory for China surpassing the Valley in the next five to ten years.

That’s a more interesting transformation than Sinovation’s positioning from “all about mobile” to “all about AI.”

Ah, yes, AI. The buzzword of the year.

It’s tossed around these days the way “big data”, “the cloud”, and “the algorithm” were in the past. You can seemingly work it into any pitch. But Lee isn’t just tossing it around for marketing. His companies have pooled all their anonymized data into one spot, so that small startups can have enough critical mass of clicks to work with. He offers them computing power to process all that data too. It’s a whole AI collective his startups can plug into when they take his money, and he hopes to get even more startups into the group so that that data cash can grow even more. 

And promotional or not, Lee comes by his AI cred honestly. He studied AI at Columbia University more than 30 years ago, and at Carnegie Mellon University, where he got his PhD in computer science. He says his papers still hold up; he regularly distributes them to his team.

Those credentials have only recently come back into fashion. While at Microsoft, he was told to drop the PhD title from his business card; it was considered a detriment when you were selling business software for a living.

His firm has invested in a handful of AI powered companies-- everything from a payday lending company called Smart Finance Group to a question and answer site called Zhihu. Even Meitu-- Sinovation’s first IPO which he expects will return the entire fund’s capital once they’re allowed to sell their shares -- uses AI to power….. Selfies. That right: Selfies. They know through a brute force of data what their audience considers “beautiful” and can apply those filters and tweaks for you automatically. (Let's set aside for a moment, whether this is a good or bad thing for girls and my broader feminist objection to a single "standard of beauty.") 

Because AI is mostly all about brute force of data. You can be the second tier researcher with 2x the data and outperform the top researcher handily, Lee says. 

And in that lies the biggest reason Lee is so bullish on AI in China, and the biggest risk he thinks every startup around the world faces right now: It all hinges on whether or not you have access to shit loads of data.

First, the bull case. China long had a disadvantage when it came to building consumer web businesses, and it’s been absolutely nowhere on enterprise. As I’ve written, that gap is narrowing. 

Not only is Chinese money one of the biggest forces propelling unicorns to higher and higher valuations, but Chinese companies are competing with US startups in ways they never have before. Just look at two of the highest valued startups of all time: Uber and Didi. Didi has long ago passed Uber in ride volume and if Uber had a more liquid stock, it likely would have passed it in valuation.

More impressive: Didi is running a better game when it comes to global domination. It has invested heavily in local players who understand their markets better than Uber, like Lyft, Ola, Grab and Brazil’s 99. Uber, meantime, is draining its capital and splitting its focus at a crucial time trying to compete directly in these markets. Didi is funding the locals to run their own races, and will likely just purchase any winners later on.

While Facebook, Google, and Twitter were thwarted in China by government regulations, Didi just plain out won. It out-executed Uber, fair and square.

We are used to Silicon Valley companies failing in China. We are not used to Chinese companies beating Silicon Valley companies in valuation and global prowess while both are still in the pre-IPO stage. This is a game changer moment in how the Valley thinks about China— and how China thinks about China.

And, as Lee points out, a lot of the companies he is funding aren’t these derivative “the Google of China”, “eBay of China”, “Uber of China” models. There is nothing in the US like his online lending company, he argues, because the US has more ready access to credit, it doesn’t have as large of a consumer market as China, and has far more government regulation. An authoritarian, “pro-tech” government can allow startups to run a lot faster in the beginning and deal with potential regulation later, he says.

There are a lot of Chinese VCs who argue China’s sheer force of demographics will propel its tech sector, but Lee has never been one of those. He argued six years ago that China was still in its copycat phase, buoyed by a massive consumer base and frequently government erected barriers to entry when it came to foreign originators. I argued back at the time all of tech were copy cats: Facebook hardly invented the social network, any more than Apple produced the MP3 player, or Google built the first search engine. Execution, details, design, business models were all where innovation lives-- whether Google or Tencent. AOL certainly never found a way to build an empire out of IM the way Pony Ma did.

But it was a very different Lee at lunch yesterday, several years into running his own firm with his first multi-billion dollar IPO under his belt. One even more bullish on Chinese innovation than I am, which is saying something.

Mobile broke the fast-follow paradigm in China, he argues, because there was a more open playing field and mass consumer adoption in China from the beginning of the smartphone era in a way there wasn’t in the desktop world. AI has the likelihood to propel Chinese startups even more, thanks to all that data, far less government restrictions in a lot of markets (if you are a Chinese company, that is), and all the assets China has already built in terms of talent, wealth, and a pool of managers who’ve built high scale Web startups before. Lee’s research has shown that 43% AI papers published have at least one Chinese author.

I asked why the Chinese government-- which typically worries more about how to employ its population-- wasn’t freaked out more about AI taking jobs, a spectre consistently raised in the US. Lee shrugged and agreed it’s strange given AI will displace a lot of jobs. But he said, so far in his conversations with regulators, it hasn’t been a concern.

China has gone from a laggard to neck-and-neck with many US rivals. In another five to ten years, China could easily surpass the US. And the two ecosystems will likely divide up the rest of the world. US companies will dominate Western Europe, the English speaking world, and older economies, while Chinese companies expand rapidly in South East Asia, India, Africa, and in some cases, the Middle East. It will not be the same playing field Silicon Valley has enjoyed in the past. Whether it’s driven by AI or not, on that Lee and I agree.

So what is the risk to his bold vision playing out and AI giving China the extra edge it needs to go from “other dominant” consumer tech hub to the dominant consumer tech hub? Again, it’s access to data. While Chinese companies are lucky to have so much more data, driven by such larger, always-on, mobile audiences, both China and the US are dominated by several large Internet giants who are only getting stronger as they age: Google, Facebook, Amazon, Apple, Alibaba, Tencent, and although weakened lately, Baidu. As the former head of Google China, he knows the big company game well.

One thing both ecosystems share is how dominant their largest tech companies keep getting as they defy middle age. In the Valley, this has lead to a little bit of an existential crisis: If Google owns everything data, Facebook owns everything people, Amazon owns everything logistics, Apple has a stranglehold on mobile access and discovery, and Netflix/Amazon/Apple/YouTube own everything entertainment… where exactly do we build new products?

In China, the approach has been to take money from the corporates. Corporate venture capital has been on the rise everywhere, but in China it’s one-third of all deal flow. So far Alibaba, Tencent and Baidu are deciding what startups can essentially live and die by flooding the ones they like with capital and in some cases, valuable partnerships and distribution.

But no one-- not in the US, not in China-- is sharing data. And as the largest eight (or so) tech giants in the world-- worth trillions in combined market cap-- get bigger and bigger, that data imbalance will get harder and harder for startups to compensate for with their nimbleness and smarts.