I remember the first time I visited China. We landed around 11pm local time in Beijing in the midst of the summer heat wave. As we landed, humidity fogged the Boeing’s windows, and the runway lights projected mirages from the haze. I could have sworn that heat was the product of a billion people’s fervent labor to advance their country and pull themselves into a new era.
Since that trip, when I visited RenRen, Autonavi and a few other blossoming startups, the Chinese startup ecosystem has grown tremendously. Chinese startups raise nearly half of all venture capital dollars and nearly 100 are valued at $1B.
Kai-Fu Lee’s book, AI Superpowers, provides some of the best history and perspective on the Chinese startup ecosystem I’ve read. Dr. Lee was President of Google China, and has held executive roles at Microsoft, Apple and SGI, among other places. He’s a venture capitalist in China and knows the ecosystem well.
There are two ideas in the book that will remain with me. The first is his view of the influence of machine learning in the world. The second are his descriptions of the fierce competitive dynamics in China.
On machine learning,
The AI world order will combine winner-take-all economics with an unprecedented concentration of wealth in the hands of a few companies in China and the US…China’s data advantage extends from quantity to quality. The country’s massive number of Internet users- greater that the US and Europe combined- gives it the quantity…Unlike American companies which gather data about online interactions, Chinese companies gather data about the physical world.
This quote spans a few chapters but I wove it together to make a few key points. First, the book embraces the idea that machine learning creates monopolies based on data aggregation. Second, China will have more data. Third, Chinese data captures a greater fraction of human behavior. More of it measures the physical world because the Chinese internet blossomed first on mobile in a wave called O2O, online to offline.
On startup competition, Chinese founders differ from their US counterparts in that the Chinese entrepreneurs what Dr. Lee dubs “gladiator entrepreneurs.”
As we enter the age of AI implementation, this cutthroat entrepreneurial environment will be one of China’s core assets in building a machine learning economy…In my view, the willingness to get one’s hands dirty in the real world separates Chinese technology companies from their Silicon Valley peers.
The book argues startups in the US prefer to solve problems with software rather than with people or by moving atoms, and casts them in the light of operators as armchair generals, compared to Chinese founders who are infantry.
For example, the competition in the Chinese ecosystem exemplified by the War of a Thousand Groupons in the mid-2010s in which hundreds of GroupOn clones fought for survival first by vying to acquire the most customers at the lowest margins and then to establish competitive barriers to entry via diversification, created Meituan, a $320B conglomerate that’s less than 4 years old.
The book argues a piquant point of view and one that deserves to be read and considered.
To me, the Chinese and US startup ecosystems differ because of their history.
One started with PCs and laptops, the other with mobile phones.
One began with expensive labor and the other with inexpensive labor (and this is the reason for the differences in go-to-market approaches mentioned above; many Chinese strategies are simply too expensive to pursue in the US, hence the obsession with the leverage software provides).
One has a government that has embraced - until recently - laissez faire policies toward technology, while the other has a government investing enormous sums to accelerate technology and entrepreneurialism. Success in technology is a national priority for China.
AI Superpowers shares the Chinese startup history and provides a contrast between that ecosystem and the US, and is a starting point for good debate about how we should think about the next 20 years of startups: in a far more global context than at any time in the past.