Former Google China President on the CFO-CEO AI Divide
Kai-Fu Lee, CEO of 01.AI, explains how CFOs can win over a skeptical boss and train their successors as AI takes on more of the work.

Sign up for smart news, insights, and analysis on the biggest financial stories of the day.
Kai-Fu Lee, the former president of Google China, has two arguments a CFO can make to a CEO who’s skeptical of AI and one piece of career advice if neither lands.
Lee, who now serves as chief executive of his own AI company, would start by showing the CEO how AI could change the way the company runs, down to who does the work and who oversees it. Then he’d use AI to answer a question already on the CEO’s desk, like which divisions might miss their quarterly revenue targets and where the CEO could step in.
If neither pitch lands, Lee told CFO Upside before his Nov. 5 keynote at WOBI’s World Business Forum in New York City, “the CFO should resign and get another job.”
Authority and Accountability
A CEO or the board can approve an ERP the whole company depends on and leave the day-to-day work to the CFO and their team. “I don’t know a single CEO who uses ERP,” Lee said. He expects the same reaction to Claude Cowork if the CFO pitches it as another tool the finance team uses to get its work done.
A CFO who wants more room to use AI in finance could start by showing the CEO what it can do for them. Lee says Boss AI, a product his company sells, can draw on meeting recordings, sales calls and other internal records to give executives a “God’s eye view” of the company.
A CEO could ask, “What are the biggest debates in the company in meetings that I didn’t attend that I need to know about?” Lee also imagined a CEO whose company had just survived a crisis asking, “How do I install process to stop the recurrence of such disasters?”
When CFO Upside asked why AI should be a CEO mandate when finance has to fund it and shoulder a large portion of the risk, Lee pointed to who can change the org chart. “AI first and foremost is about transformation of a company that includes the strategy, the priorities, and the reorganization, and that belongs to the CEO,” he said.
His forecast of “many times more AI workers than human workers” would change who manages the work. “You need to identify who can manage AI workers,” he said. “You need to reduce the layers of management.”
For the finance chief putting together the business case, Lee would start with where AI could help bring in more revenue. “I think it would be too little for the CFO to think of it as approving an expense or using it to save costs,” he said. “Yes, those are all fine, but what’s important is how does it help the company grow?”
Lee’s distinction between buying AI and changing how people work matters as budgets rise. Of the 102 CFOs Bain & Company surveyed, 56% said they were increasing companywide AI investment by more than 15% in 2026. Bain found that some finance teams ran AI-generated forecasts alongside their existing planning cycles, leaving them with two processes and little of the time savings they’d expected.
Elsewhere, the zero-day close OpenAI CFO Sarah Friar is working toward shows how the division of labor could change inside finance. AI would flag exceptions and draft variance explanations, reducing “the scramble to reconstruct the business after the period ends.” As Friar put it in a recent essay, “Finance validates the numbers, applies judgment, and owns the final sign-off.”
Tomorrow’s Problem
When finance chiefs talk about where they developed their judgment, they often point to work they did early in their careers. Lee expects “many fewer entry-level jobs,” and he has a different apprenticeship in mind for the people who get them.
His case for keeping people in those jobs starts with who owns responsibility for errors made by AI. “When AI makes a mistake, you can’t punish it. AI has no passion, no desire for promotion, no fear of getting fired.” He would encourage junior hires to “be that accountable person who’s partnered with one or more AI workers working in tandem.”
At Anthropic, one AI workflow produces most of a monthly financial review in about 30 minutes, according to an Andreessen Horowitz report dubbed “You Need a New CFO.” The venture firm says the leanest finance teams it sees are targeting 2% of company headcount, compared with a longstanding rule of thumb of about 5%. A junior analyst who once learned by putting that review together might now start by checking AI’s work.
As companies weigh which layers of management they can do without, Lee would have people in the same role agree on how they handle recurring problems so that know-how survives their departure. An accounting team could record what evidence to check before clearing an unusual transaction, then look at what happens when AI applies the same rules. “AI should learn by applying those skills and seeing the results as a positive or negative example,” he said.
Some decisions that reach a CFO’s desk will come with little data to go on, however much faster AI can run the numbers. “Taste is innately human,” Lee said. He pointed to the iPhone as an example of foresight, the ability to “predict what people will like, even though they can’t tell you they like it.”
No Comps: Lee sees a place for that kind of foresight when a CFO is weighing which markets to enter or which businesses to start. Those calls also take courage, which he described as “the willingness to be accountable in doing something, and take the fall if it doesn’t work.” Asked what the best leaders he’s met have in common, Lee said, “I would put conviction first.”











