Following the release of MEITU's 2025 financial report, market opinions on the company remain divided. Is AI weakening or strengthening MEITU? As large language models continue to evolve with multimodal capabilities, is the space for image processing applications shrinking or expanding?
In an exclusive dialogue following the earnings call, MEITU CEO Wu Xinhong offered a direct response: if forced to choose a single operational metric to answer this question, he would select profit growth. Users vote with their wallets, with the majority of this growth coming from subscriptions and a portion from token consumption. Both point to the same conclusion: product recognition.
However, the logic behind profit growth requires deeper analysis. AI's impact on MEITU's value chain is uneven. Upstream, the data and imaging platform built over years by thousands of engineers cannot be easily replicated. Downstream, the channel reach advantage, supported by 280 million monthly active users, is also difficult for AI to replace. The area significantly transformed by AI is the middle layer: the product creation process itself.
AI has reduced costs while simultaneously increasing MEITU's production capacity. Previously, users faced high learning curves in certain scenarios. AI has substantially lowered these barriers, enabling MEITU to offer more products to the market, thus expanding its potential. Wu Xinhong cited an internal example: a single designer can now independently complete a full AI product, a task that previously required a team of twenty to thirty people.
This reality raises a sharper question: if one person can create a product generating annual revenue in the millions, why would they remain with a company? Wu Xinhong's answer points to scalability. The reuse of platform capabilities and the low-cost customer acquisition afforded by channel advantages are things an individual starting from scratch externally cannot replicate.
This combination of an upstream platform and downstream channels also dictates MEITU's approach to new products. The demand for each new MEITU product emerges from the previous one. For instance, MEITU Design Studio arose from MEITU Xiuxiu users' needs for e-commerce materials, while KAIPAI originated from prompter users' demands for talking-head videos within BeautyCam. Identifying new needs within the existing user base and leveraging channel advantages for low-cost acquisition is the growth methodology Wu Xinhong repeatedly emphasized.
Nearly 80% of MEITU's users are lifestyle scenario users, forming the core user base. The company's strategy is to extend upwards from this base to reach professional consumers and small-to-medium businesses (SMBs) with stronger payment willingness and higher ARPU. Wu Xinhong clearly stated that MEITU will not compete in the professional user and large enterprise market served by Adobe, citing a gap, high customer acquisition costs, and entirely different product requirements.
Regarding overseas growth, the explosive popularity of RoboNeo in 15 countries, including Brazil, has drawn market attention. Wu Xinhong stated the company has developed a replicable methodology. The evaluation criteria are straightforward: whether user growth and revenue increase in sync, and whether retention rates remain strong. If user numbers grow without corresponding revenue growth, it is unhealthy. In European and American markets, where payment willingness is inherently strong, user growth typically leads to proportional revenue growth.
On the commercialization path for productivity tools, the e-commerce design scenario has been the first to prove successful. The highly competitive e-commerce industry creates an urgent need for cost-reduction and efficiency tools. Furthermore, e-commerce design workflows are highly digital, allowing AI teams to fully replicate them. Wu Xinhong stressed that both standardization and digitization are essential. If a task is standardized but relies on offline human labor, AI intervention is difficult. MEITU Design Studio is far from reaching its user growth ceiling, with significant room remaining in global markets. User ARPU in productivity scenarios is rising rapidly, and advancements in model capabilities are improving output quality and stability, making previously conceptual ideas truly usable.
The industry's revenue model is undergoing structural change, evolving from early advertising and traffic diversion, to adding subscriptions in 2018, to the current rapid expansion of token consumption. Wu Xinhong expressed that token consumption is becoming one of MEITU's primary future revenue sources. This trend is clearer in the strategy around OpenClaw. MEITU launched Meitu CLI, integrating eight AI Skills into OpenClaw, positioning itself as the imaging infrastructure for the AI era. Wu Xinhong defines OpenClaw as a channel for capability distribution and a revenue source.
Addressing concerns that widespread external API access might weaken the entry point mindset for MEITU's own products, Wu Xinhong used smartphone manufacturers as an analogy: regardless of their investment in imaging, they can only meet general needs with limited product depth. MEITU Xiuxiu's market share consistently grew during the smartphone boom. OpenClaw provides general capabilities, while MEITU's own products offer depth; the two are complementary.
Looking ahead two years, Wu Xinhong believes the greatest uncertainty comes from agile startup teams, not large corporations. In the AI era, the biggest bottleneck for all companies may lie within the organization itself, ranking above technology and computing power. If an organization becomes redundant, inefficient, or bureaucratic, no strategy can be effectively executed. As AI drastically compresses the middle stages of product creation, those who can continuously identify needs and reach users at low cost will gain a competitive advantage.