Ant's AI Health App Has 150 Million Users. That's the Wrong Number.

The press release reads like a triumph. Ant Group's artificial intelligence health app, AQ — recently renamed Ant Afu — announced it had reached 150 million users and was handling roughly 20 million health inquiries a day. In a country of 1.4 billion people, that sounds like China's next default app.

Read the sentence again, and the two words that don't belong together. "Users" on one side. And on the other, a growth story whose biggest single push was a television comedy sketch.

Start with how the number got big, because the path explains the meaning. AQ launched in June 2025. It reached 100 million users around the start of this year, and during the Lunar New Year holiday it topped China's Apple App Store download chart for several consecutive days. By September the registered total stood at 150 million.

But "users" here means something specific, and something weak. In December 2025, the same period Ant put its monthly active users at about 15 million, it was also talking about a 100-million-user total. One in seven of everyone who had installed the app was still opening it a month later. The September milestone of 150 million is stated as total users, not monthly actives.

That ratio is the story sitting under the headline. When a consumer app reports total "users," it is counting how many people have ever tried it. It is not counting a product people use. Consumer apps keep both numbers; the banner one is almost always the broader.

The way AQ gathered its downloads explains why the two have drifted so far apart. The surge was a once-a-year event with a marketing budget behind it. During the holiday, young people installed the app for parents and grandparents — a gift, not a habit — and cities of the third tier and below produced 52 percent of the new users. Ant spent tens of millions of dollars on promotion, from subways to painted walls in the countryside. The downloads were subsidized twice over, financially and emotionally.

Health is a hard habit to build, and Ant knows it. The industry's own complaint is that health apps are low-frequency: people reach for them when sick, then forget they exist. The stated fix is to make the product feel like a companion rather than a tool, turning a rare health event into a daily check-in. That is a real theory. It is also, so far, mostly unproven — the monthly-active figure has never matched the download count.

There is genuine evidence the theory is working, and it's the sign to trust. Daily health inquiries roughly doubled, from about 10 million a day in February to nearly 20 million now. Query volume can be inflated, but a day-over-day count of people asking questions is closer to behavior than a download count. If that daily habit forms, the giant registered base was worth acquiring. Twenty million questions a day is real pull for a free product — the opposite of the subsidized downloads that produced the headline.

Now concede the part that matters to an investor: you cannot buy any of this. Ant Group is private. Its 2020 IPO was pulled by regulators and it has never returned to the public market. The health app exists to be a wedge, not a business that has to pay for itself. Hospital bookings and insurance reimbursements route through Alipay; doctor avatars and the health questions users type in feed the models Ant wants to train. That is why a giant can afford to give the product away. All of that is real, and none of it is a stock you can own.

So treat this announcement as instruction, not as an opportunity. The next AI-health headline — from a company you actually can own — will arrive with its own user count. Bring the split with you. Registered versus active. Bought versus earned. When a company leads with "reached N million users," it is quoting the number that flatters it; the number that reveals the truth is the active figure, and the number that predicts the future is whether people keep returning after the promotional push ends. Ant's own disclosures show the distance between 150 million downloads and a product people use. When the next one comes wrapped in a bigger-sounding total, that distance is the thing to measure.