Meitu’s Subscriber Math Has Outrun Its Headline Earnings

Start with the figure that frames the whole position: 17.9 million global paying subscribers, up 30.2% year over year as of the Q1 2026 operating update. That is the number the market spent three months ignoring and then repriced in a matter of days. At 5.50 HKD against a consensus average target of 9.37, the stock sits roughly 70% below where the analyst range centers, and the gap is not built on a hope about some future product. It is built on subscription revenue that already exists and is already compounding.

The core photo, video, and design segment grew 34.3% to RMB 852 million in the most recent quarter, and that growth is subscription-led rather than ad-cycle-led. This matters because of the order in which the variables move. Engagement across a 276 million monthly active user base feeds paid conversion. Paid conversion lifts ARPU. ARPU, not raw user count, is what monetizes. Management has described higher ARPU trends “driven by productivity tools and token/agent consumption,” with further upside expected from premium production offerings. The AI productivity line carries the same signature: annual recurring revenue up 56.2% to RMB 580 million, with AI credit consumption rising sharply. When a company shifts its mix from impression-based advertising toward recurring subscription and token consumption, the multiple it deserves changes, because the revenue quality changes.

Now the part that creates the opportunity. Full-year 2025 revenue rose 28.8% to RMB 3,859 million, but operating income fell from RMB 490 million to RMB 319 million, and operating margin compressed from 16.37% to 8.27%. Net income dropped about 34.9%. On the surface, that is an earnings decline, and the stock was treated like one through late March. But free cash flow over the same period went the other way: RMB 1,233 million against RMB 700 million the year before, a 76% increase. A business whose cash generation nearly doubles while its reported operating margin halves is not a business in decline. It is a business spending into a transition, and the spending shows up in the income statement before the return does.

This is the setup that gets missed repeatedly in platform names that pivot their monetization model. The pattern is familiar: a company moves from a mature, higher-stated-margin revenue base toward a faster-growing one that carries heavy upfront investment, and the market anchors on the margin print rather than the trajectory underneath it. The earlier error is almost always the same. Investors read margin compression as deterioration when it is actually mix shift plus growth investment, and they sell the multiple at exactly the point where the new revenue line is inflecting. The 28.5% rally over a single month, triggered when the subscriber and ARR figures landed, is the market beginning to correct that read. The 11% to 16% intraday move on the release tells you how much was not in the price.

The balance sheet removes the usual objection to paying for a transition. Net cash sits at RMB 3,221 million against total debt of RMB 1,645 million, and book value per share is RMB 1.24. This is not a company funding its AI build with leverage it cannot service. It can absorb the margin compression for several more quarters without touching its capital structure, which means the silent question worth watching is not solvency but substitution: whether newer AI deployments are displacing older, higher-margin maintenance and legacy contracts or genuinely adding on top of them. If they are mostly displacing, the margin recovery I expect is slower and shallower.

The environment is the one real complication, and it is regulatory rather than financial. China’s Cyberspace Administration has advanced security assessments for generative AI, and draft Interim Measures on human-like interactive AI services, drafted in spring 2026 and slated to take effect mid-year, raise compliance costs on exactly the data and content handling Meitu’s apps depend on, with estimates running around 20% higher year over year. Abroad, the recalibrated EU AI Act and tightening data-localization rules lift the operating and capital cost of international expansion.

None of this breaks the subscription thesis. It does mean the margin recovery has a regulatory tax embedded in it that was not there two years ago, and it slows the path back toward the prior 16% operating margin.

So the risks worth naming are specific. If compliance costs land at the high end of the 20% range and stick, the operating margin recovery I am underwriting stalls below 12% even as revenue keeps growing, and the multiple re-rating loses its anchor. If the substitution effect dominates, the 34.3% core growth flatters a revenue base that is cannibalizing its own higher-margin legacy work, and ARPU expansion slows faster than subscriber count suggests. And the rally itself raises the bar: having moved 28% in a month, the stock now needs the next subscriber print to confirm rather than merely not disappoint.

Figures reflect the most recent available data and may differ slightly from live market prices. · © Mathstock