Bloomberg Tech

Kai-Fu Lee on Why the Money Is Above the Model

Kai-Fu Lee· CEO of 01.AI at 01.AI
·~13 min·English·Bloomberg
LLMOpen SourceBusiness StrategyAI Company
TL;DR

Kai-Fu Lee argues that as models grow smarter than all humans and open source narrows the gap, the value in AI moves off the model and onto applications that change a company's bottom line, and 01.AI is betting on hands-on enterprise deployment in markets its US rivals are unlikely to enter.

01The Trajectory

Smarter Than All Humans

In three years, Kai-Fu Lee says the models went from average-human ability to smarter than every human, with no end in sight.

three years ago, were like an average human, but now they're smarter than all humans.

— Kai-Fu Lee, Bloomberg Tech
Key Insight
If capability is a steep, still-rising curve, any strategy that treats the model itself as the scarce asset is betting against the trend line. The scarcity has to move somewhere else.
02Open Source

Open Weights Close the Gap

Kimi K3 signals open models narrowing the gap on closed ones, and Lee says the bigger deal is on-premise deployment, not the benchmark score.

having an open source model actually is great for enterprises because they want on premise deployment. They don't wanna send their data to the cloud.

— Kai-Fu Lee, Bloomberg Tech
Key Insight
The unlock isn't parity on a leaderboard. On-prem open weights sidestep the data-governance objection that stalls enterprise cloud AI, so distribution, not raw capability, becomes the wedge.
03The Pivot

Don't Waste Genius on Simple Tasks

Rather than keep training its own foundation model, 01.AI pivoted about a year ago to building applications on the best available models, and its costs collapsed.

we're better off building applications on the very, very best model and using their intelligence to change the financial statement of companies.

— Kai-Fu Lee, Bloomberg Tech
Key Insight
It is a public admission that the model layer is commoditizing faster than a mid-size lab can monetize it. The defensible margin has moved up to the application that changes a customer's bottom line, and Lee calls it 'a pity' to spend such models on simple tasks.
04The Product

BOSS AI: A God's-Eye View for the CEO

01.AI's flagship gives a CEO one god's-eye view over company data to make the decisions that move the financial statement, instead of another sugar-coated demo.

It is a product designed to give the CEO the god's eye view over all the data to make all the right decisions that will move the dial in the financial statement of the company.

— Kai-Fu Lee, Bloomberg Tech
Key Insight
By selling to the CEO rather than the CIO, 01.AI reframes AI from an IT project full of demos into a P&L lever, the same top-down wedge Palantir uses. Bookings up three-to-four times suggests the framing lands.
05Growth Shape

Not a Token Factory

01.AI expects hundreds of percent growth rather than the 50x of a token vendor, because its ceiling is how fast it can hire multilingual consultants to fly out and deploy.

we are not a token factory, so you're not gonna see a 50 times growth.

— Kai-Fu Lee, Bloomberg Tech
Key Insight
A services-shaped AI business scales with headcount, not GPUs, which caps the multiple but also builds a moat a pure-inference rival can't erase by cutting token prices. The bottleneck is deliberately human.
06Geopolitics

Export Controls Failed

Kai-Fu Lee calls US export controls an imaginative but failed attempt to contain China, arguing the real GPU shortage is global demand and frugality, not the ban.

exports control was an imaginative but unsuccessful effort to contain China. It has failed.

— Kai-Fu Lee, Bloomberg Tech
Key Insight
Accept the verdict or not, the strategic point stands: a supply-side lever loses its bite once open weights let Chinese firms compete on software and deployment instead of on frontier training runs.
07US vs. China

The Four-Front Split

Kai-Fu Lee maps the race front by front: the US leads on model profits and the most profitable enterprise apps, while China leads on model market share, consumer software, hardware, and hands-on enterprise transformation.

in models, US will make a lot more money, but China will have a larger market share because of open source being a powerful distribution and on prem premise deployment being such an important thing to enterprises.

— Kai-Fu Lee, Bloomberg Tech
Key Insight
The framing rejects a single 'who's winning' scoreboard for a segmented one. Profit and share decouple, and China's edge concentrates where distribution and operational grind matter more than a benchmark lead.
08The Moat

A Two-Horse Race the Other Horse Won't Run

Kai-Fu Lee frames enterprise AI transformation as a two-horse race with Palantir, but says Palantir is unlikely to enter the Asia, Middle East, and Belt-and-Road markets 01.AI is claiming first.

we have a two player horse race, a two horse race where the other horse chooses not to run-in the tracks that we're entering.

— Kai-Fu Lee, Bloomberg Tech
Key Insight
The moat isn't a better product; Lee concedes Palantir's parity. It's geographic: first-mover presence in markets a US defense-linked vendor is structurally unwilling to serve. Competition avoided is more durable than competition won.