August’s AI app rankings are out, and at first glance they look almost surreal. On one side, the chatbot track drums and cymbals — user counts in the hundreds of millions, with Doubao and DeepSeek practically new infrastructure inside Chinese phones. On the other, image and video apps quietly mint money, leading the revenue charts by a mile and leaving the traffic giants far behind. Half seawater, half flame. If the picture confuses you, you may not yet have grasped the brutal truth of AI application monetization.
‘Beauty’ Remains the Primary Productive Force — and It Got More Expensive
Start with the flame: the revenue list. The top four slots of China’s AI app revenue top-30 (excluding the domestic Android market) are locked down by image and video tools — Meixiu (BeautyCam), Meitu Camera, Jimeng AI, and AI Mate. The leader alone, Meixiu, posts an estimated monthly in-app revenue above $3.85 million on its app alone — a figure that would make the user-buying, cash-burning LLM companies blush.
Underneath sits a plain, unbreakable truth: users always pay for results — especially results that make them prettier, cooler, and worth posting. AI supercharged those results into something immediate and stunning: one-tap manga avatars, photos that talk, dreamlike scenes that never existed. The payment logic is simple — money in exchange for an instant visual outcome and social capital. This tool-to-monetization path was proven in the Meitu era; AI just poured gasoline on it.
‘Smart’ Is Becoming Cheap — Even Free
Now the seawater: the traffic lists. ByteDance’s Doubao swept both the download and active-user charts in a rout — 137 million monthly actives on the app alone, quietly making it a national-scale entry point. DeepSeek, Tencent Yuanbao, and Kimi follow, each holding tens of millions to hundreds of millions of users. Yet, eerily, these absolute traffic kings have all but vanished from the revenue rankings.
Here is the market’s core contradiction: users happily use AI at high frequency to improve efficiency — and refuse to pay for it. Today’s chatbots resemble an enhanced search engine, an on-call know-it-all: write my weekly report, look that up, translate this. The needs are frequent and real, but twenty years of internet free-education trained users to expect information and efficiency tools at zero cost. Charging for a chat or an answer is far harder than charging for a prettier photo. So the Doubao camp converged on the internet’s oldest playbook: don’t ask about revenue — take territory first. Free strategies vacuum up users, habit gets built, the entry point’s value gets maximized; monetization is a story for later.
The split deserves one more structural read, because it maps onto two different businesses wearing the same “AI app” label. The beauty-and-video camp sells a result: a finished image, a finished clip, delivered at the moment of demand, priced per outcome or per subscription the way games and filters always have been. The chatbot camp sells a capability: an always-available cognitive utility whose value compounds across thousands of small tasks — which is why per-use pricing feels wrong to users and why the camp’s monetization keeps slipping to later. History’s closest parallel is maps versus browsers: the result business monetized early and narrowly; the capability business monetized late but everywhere.
For builders and investors, the August numbers suggest a portfolio logic rather than a winner-picks-all. The result business funds itself today but faces ceiling questions — novelty decays, competition floods in, and “prettier” is an arms race with falling margins. The capability business loses money today but compounds distribution, data, and habit — assets that convert into enterprise APIs, agent upsells, and commerce routing the moment management’s patience for “land first” expires. And the role-play middle — entertainment with built-in willingness to pay — is the pragmatic hedge: emotional engagement monetizes like a result and retains like a capability. Watch the next three months for the first serious chatbot monetization move; the market’s temperature will be set by whoever blinks first between Doubao’s free empire and the finance teams behind it.
Between Ice and Fire, Who Wins?
So: one camp is acclaimed but unpaid, the other paid but unacclaimed — which is the future? Short-term, the image-video apps live the best life: a mature commercial loop, cash in hand. Long-term, the real battle probably stays on the traffic side. Whoever owns users’ highest-frequency scenarios owns the entry point of the future AI ecosystem — and once chatbots find their monetization model, whether B-end API pricing or innovative C-end value-added services, the explosive potential is phenomenon-scale.
The rankings also surfaced the bridge players: role-play apps like Maobox (Cat Box) carry entertainment traffic and monetize decently, climbing the revenue charts hard. They prove that between the extremes of “effect” and “efficiency” lies a wide territory of “experience” and “entertainment” — possibly the next gold mine — and the only tier where the ice and fire genuinely mix. In sum, August’s charts paint a vivid market picture: image tools are earning the money of the present; chatbots have bet on the imagination of the future. This song of ice and fire has just opened, and the real show starts the day the traffic giants finally get serious about monetization — until then, enjoy the spectacle of the most-used apps in the country being the ones making the least from it. That inversion is the single most important fact in China’s AI application market right now, and every strategy conversation in the industry quietly routes through it.
