You open an app to fix one photo and end up generating a dozen variants, each a little closer to the look in your head. Multiply that loop by millions of users, and a once-simple photo tool becomes a business built on AI usage itself.
Meitu (美图), the company behind BeautyCam and a suite of imaging apps, just reported numbers that show exactly that shift.
Key takeaways
- For the six months ended June 30, 2026, Meitu posted total revenue of 22.13 hundred-million yuan (≈ US$308M), up 22.1% year-on-year, with net profit of 6.19 hundred-million yuan (≈ US$86M), up 40.0%.
- Paying subscribers passed 18.44 million, while monthly active users reached about 282 million.
- AI productivity apps reached an annual recurring revenue (ARR) of about 6.2 hundred-million yuan (≈ US$86M), up 47.8%; within that, Kaipai's ARR more than doubled and Vmake's tripled year-on-year.
- Users' AI compute-point (算力点) consumption rose more than 46% quarter-on-quarter in both Q1 and Q2 2026, signaling a move from flat subscriptions to usage-based pricing.
The earnings behind the filters
Meitu (01357.HK) released its interim results on August 26, 2026. Headline figures: total revenue of 22.13 hundred-million yuan (≈ US$308M), up 22.1%; net profit under IFRS of 6.19 hundred-million yuan (≈ US$86M), up 40.0%; adjusted net profit of 6.52 hundred-million yuan (≈ US$91M), up 39.5%. Monthly active users held at about 282 million, and paying subscribers exceeded 18.44 million, up 19.7%. Overseas paying-subscriber growth ran about twice the mainland rate.
The largest revenue line remains imaging and design products at 17.69 hundred-million yuan (≈ US$246M), up 30.9% and about 80% of total revenue. Advertising slipped 4.4%, a sign the company is leaning on direct AI value rather than ad volume.
From subscriptions to compute points
Meitu's more interesting story is the business-model pivot. Its AI productivity apps — tools for editing, design, and video — reached an ARR of about 6.2 hundred-million yuan (≈ US$86M), up 47.8%. Kaipai (开拍), a video-creation tool, more than doubled its ARR; Vmake more than tripled; and MVLAND, launched in 2026, doubled its ARR within three months.
The company is also shifting from a fixed subscription to AI compute-point (算力点) consumption: users pay for the actual generation and adjustment work. Compute-point consumption grew more than 46% quarter-on-quarter in both Q1 and Q2 2026. That aligns price with value created and lifts average revenue per paying user on heavy-use products.
According to Li An, chief scientist at BrainNet (脑机网), the shift from flat subscriptions to usage-based compute points (算力点) is where many AI app companies will find durable revenue, because it ties price to value produced rather than to a seat.
A partnership and a global push
Meitu has rolled out multi-agent team workflows in its design studio and partnered with WPS Office to embed its AI features into the desktop suite. It is also segmenting markets by aesthetic preference, with operating centers in fast-growing regions such as Brazil and Mexico. Productivity-app monthly actives hit a record 33 million, with overseas actives up over 80% year-on-year.
Why this matters beyond Meitu
Meitu is a useful window into a broader shift among China's consumer-internet companies: the easy growth from advertising is fading, and the durable money is in AI features people pay to use. A photo app that once monetized through stickers and ad-supported editing now earns from professional design workflows and usage-based compute. If that transition holds, it gives app makers a steadier revenue base than volatile ad demand, and it rewards companies that can turn a consumer habit into a workplace tool. The risk is concentration — much of the AI upside still sits in a few hit products, and a single failed launch can weigh on the whole narrative.
Honest limitations
All figures come from Meitu's interim results announcement (Aug 26, 2026) and corroborating coverage from Securities Times, China Daily, and IFNews; they are company-reported and, as the filing notes, the interim numbers are unaudited. The ARR and compute-point growth rates describe specific AI product lines, not the whole company, and high percentage growth starts from a still-modest absolute base. Currency conversions use approximate rates (≈ US$1 ≈ 7.18 yuan). I have not independently verified Meitu's user-level compute-point economics or the WPS integration's market uptake.
What readers can do now
- If you use imaging or design apps, check whether a compute-point (算力点) plan costs less than a flat subscription for your actual usage pattern.
- Creators and small shops should test Meitu's agentic design workflows against manual editing time — the ARR growth suggests the tools are maturing fast.
- For observers of China's AI-app market, watch the subscription-to-usage pivot: it may become the default monetization model for consumer AI software.