Why Kai-Fu Lee thinks companies need an AI boss

Kai-Fu Lee’s AI Superpowers, an influential book published in 2018, largely predicted the current tech battle between the US and China. Now, Lee’s forecasting that AI will unleash more radical changes inside the world’s companies than most CEOs expect — and this could reshape the competition between those two economies once again.

The former head of Google China, who has also held senior executive roles at Microsoft and Apple, understands intimately how Silicon Valley leaders think. But now that Anthropic’s Dario Amodei has called on the industry and its regulators to “pace the frontier,” Lee says he would reframe that call: “Move fast, but make sure safety keeps pace.”

Lee says he agrees with the principle that developments in AI capability must be matched by advances in safety and governance, adding that industry collaboration will be “particularly important at this stage,” given how governance and regulatory frameworks “inevitably lag behind technological development.”

He is now deeply embedded in China’s AI industry as chairman of the Beijing-based venture capital firm Sinovation Ventures and founder-CEO of 01.AI, a Chinese “AI tiger” that has pivoted from building AI models to helping enterprises deploy the technology effectively.

Lee’s transpacific perspective has given him a unique insight into the challenges that global companies face in realizing AI’s potential. In a video interview, he says two things are clear from his conversations with other chief executives: “One is that they all want to do AI transformation. Two is they’re mostly doing it wrong.”

Lee is pitching his new book, AI Native, as a road map for doing it right. CEOs hoping to build an “AI-native” company must take personal charge of their technology strategy, he says, and use “radical transparency,” enabled by AI, to give them unprecedented insights into what is actually happening inside their organization. That conscious echo of the management mantra of Bridgewater Associates founder Ray Dalio, he adds, is an idea that Chinese employees will embrace faster than Western workforces.

Why transformation starts at the top

CEOs everywhere face “a Darwinian choice,” Lee says: Embrace AI or accept the consequences. They have little time to do so, he adds; companies whose assets are largely digital, like entertainment, gaming, and internet businesses, will be “dead in the water” if they do not adapt, he says. CEOs operating businesses further from the AI frontier — like coal mines, farms, or manufacturing plants — may have a three-to-five-year “buffer” before they feel the pain of AI disruption.

That means executives need to ask now whether they are passing “the earnings call test,” Lee says: If their company’s AI efforts have not yet boosted their profits, then they are mere “AI theater.”

Where chief executives go wrong, in any industry, is when they delegate and defer to their chief information officer, Lee argues. “Only the CEO can define the ambition, concentrate resources, redesign the operating model, remove barriers, and remain accountable for the result,” he writes in AI Native, so AI transformation “can only be done from the top.”

That also means that each CEO needs to be using AI effectively themselves, he adds. Doing so demonstrates that the company’s leader understands AI’s importance to the bottom line.

The radical transparency of Boss AI

Lee has taken his own advice, designing a “decision assistant” made up of 19 agents to help him run 01.AI. His company has since packaged this up into a product to sell to other companies, called Boss AI.

At its heart is what Lee calls a “truth engine” that records, organizes, and analyzes every internal meeting, sales call, and customer service request to show what is getting in the way of the company meeting its goals. That gives a CEO visibility into all corners of their company — and a “promise ledger” to hold people accountable for following through on what they say they’ll do.

“Middle managers often hope that the CEO will forget. And the CEO does forget, but the CEO’s AI does not,” Lee says. It is at this tier of most companies that the bosses’ plans meet the most resistance, he adds.

“There are a lot of antibodies in the company, people who just don’t want to do AI transformation. Middle managers feel vulnerable because they’re not technology experts, [and] they’re afraid the boss will put in someone who is,” he explains. Those managers also fear losing their power if their company lays off their direct reports, he adds.

Still, even as he says that AI is dividing workforces into “irreplaceable humans” and “displaceable humans,” Lee does not think that CEOs should measure AI success by the number of jobs they eliminate.

“The winners will not measure success by payroll cuts. They will measure it by what their best people, amplified 10 times, can now do that was previously impossible,” he writes in his book. Most people will not be laid off if they’re willing to learn new AI skills, he adds, “but they might be paid three or five times more” than before.

But he still believes that companies should flatten the hierarchies that protect people who resist change. “When the company is transparent, there’s no need to hide, nowhere to hide, and everybody talks about the real issues and tries to solve the problems.”

Where clashing cultures meet AI superpowers

Lee presents his ideas as “exhilarating” for employees who are willing to sign up for them. “The company will have a better morale, better culture, better execution, better results, better stock price, and everything will be a virtuous cycle and the company will grow,” he predicts.

Even so, Lee admits that radical transparency “takes [some] getting used to.” Some corporate cultures will be more reluctant than others to expose themselves to the boss’ all-seeing AI, he adds, bringing his argument around to a point which converges with his earlier predictions about AI’s impact on geopolitical competition.

The idea behind Boss AI “is more acceptable and more suitable for companies where the CEO has immense respect and power, and where companies are run by consensus,” he says. “Everything else being equal, Asian companies will tend to adopt it earlier and faster. And we’re seeing that.”

01.AI has been selling Boss AI in China and across Asia, and has seen “no resistance” to the idea of recording meetings or capturing employee communications, Lee says. Privacy is a priority in his part of the world, but not the overriding one it is in many Western markets, he says, describing this as “a second Asian advantage” over the West.

“Asian companies are more collectivist,” he says, predicting that Chinese employees will tolerate companies recording “everything all the time” if they can see the benefits to the business. Workers in the West will not let that happen, he predicts, which will leave their CEOs without the vast amounts of data available to their Asian rivals.

“This goes back to AI Superpowers,” Lee says, noting how his earlier book helped popularize the idea that “data is the new oil.”

Where Lee sees the AI superpowers now

Since AI Superpowers, Chinese companies such as DeepSeek and Moonshot have offered meaningful competition to the likes of Anthropic and OpenAI, without fully matching the capacity of US frontier models. Lee says it would be unwise to assume that China’s AI industry will continue to lag Silicon Valley.

“If you look at the top 1,000 people in AI in the US, they are significantly better than the top 1,000 people in China. However, if you look at the top 100,000, the two countries are very comparable,” he says. “So I think it would be a terrible mistake to underestimate the capabilities of Chinese companies.”

The US will not win the AI race by restricting exports of Nvidia’s most powerful chips, he adds. “The export controls do not work.” Chinese engineers’ skill at making “good enough” solutions has been underestimated, he says, adding that he finds the idea that the US will retain semiconductor dominance permanently “highly questionable.”

Chinese chipmakers’ products might not be as good as Nvidia’s, but they are used to competing on lower margins than Nvidia is, Lee notes. And US export controls can create a powerful incentive for investment in Chinese innovation, he adds. China, he predicts, will be “completely self-sufficient” in chips within two to three years.

As the US tech industry’s leading CEOs debate whether the potential dangers of AI warrant slowing down development, Lee says their Chinese counterparts do not dismiss the “hypothetical future risks.” But the industry’s focus has been “much more practical” in China, he adds: “building applications, driving industry adoption, and finding ways to monetize AI.”