Meitu: The Delivery Test After Imaging Subscriptions Hit Their Stride

Research Summary

Meitu is no longer the "photo-editing company that monetized through selfie phones" it once was, and it has not fully reached the position of high-stickiness enterprise software like Adobe or Figma either. The machine that actually makes it money today packs high-frequency image-editing demand into a product shell that is light enough, smooth enough, and mainstream enough, and then turns traffic into cash through VIP subscriptions and usage-based AI features. In the first half of 2025, the company posted revenue of RMB 1.821 billion, a gross margin of 73.6%, and adjusted net profit attributable to owners of RMB 467 million; for full-year 2025, on a continuing-operations basis, revenue was RMB 3.859 billion and adjusted net profit attributable to owners was RMB 965 million. More importantly, the company has moved "looking good" away from pure consumer photo editing and into e-commerce, design, video, and content production—scenarios that are more willing to pay. In Q1 2026, global paying subscription users rose further to 17.9 million, and productivity-application ARR was disclosed at roughly RMB 580 million, which shows this new path is not just slideware.

What the market trades today are two sharper questions, not "can a photo-editing app still survive." First, can vertical AI applications withstand the impact of general-purpose large models. Second, can Meitu turn "subscriptions" into a stable base and then make "usage-based fees" a second throttle. The company's own latest evidence is fairly clear: as of March 2026, user AI-credit consumption grew 59% versus December 2025, with Kaipai, RoboNeo, Meitu Design Studio, and Vmake growing even faster; in Q1 2026, imaging and design product revenue grew 34.3% year over year, and paying users of productivity applications grew 52.9% year over year. This shows users are increasingly treating AI as a tool rather than just trialing it. But the other side is just as clear: once AI capability becomes a platform default, the pricing power of vertical applications gets compressed immediately, and every upward revision in market optimism about general-model capability since last year has brought valuation pullback pressure on application-layer companies like Meitu.

The stock's past swings have basically followed a track where "the story runs faster than the business." When it listed in 2016, the capital markets saw a selfie-app company with enormous user scale, then found that 95.1% of its revenue in the first half of 2016 came from phone hardware. This mismatch of "internet-traffic asset, hardware-revenue structure" planted a valuation discount from the very beginning. In 2021, the company bought cryptocurrency and the share price spiked on thematic excitement; in 2024, liquidating crypto produced a one-time gain that lifted IFRS profit again. What truly changed the fundamentals was the sustained shift to subscriptions after 2020, profitability turning positive after 2022, the commercialization of AI-native products after 2023, and the 2025 partnership with Alibaba that connected channels, cloud, and e-commerce scenarios—not these noisy milestones. The market accordingly gave it a round of "AI-application re-rating" in 2025, with the share price touching a seven-year high on the day of the Alibaba partnership announcement; Google Finance shows a trailing-52-week high of HK$12.56. But by 2026-06-12 it had fallen back to HK$4.64, all but repricing the worry over "whether vertical applications can defend their value" all over again.

The most important bull-bear divergence right now looks like growth on the surface but is really about structure. Bulls look at the fact that imaging and design products have become the largest and fastest-growing segment, up 45.2% year over year in the first half of 2025 and up 41.6% for full-year 2025 on the core segment, with a rising share of new paying users coming from high-ARPU overseas regions, and with both the paying headcount and ARR of productivity tools accelerating—signs that the company is moving from "content beautification" into the more valuable budget pools of "commercial image output" and "video production." Bears focus on three other things: the advertising business has clearly slowed, growing only 5% in the first half of 2025 and even declining slightly for the full year versus 2024; AI compute and cloud costs, while not out of control for now, reached RMB 415 million in 2025, and if usage-based revenue grows faster than pricing optimization, gross margin will be continuously eroded; and one step further out, ByteDance's products, CapCut, Canva, Adobe, Kuaishou's Kling, and even stronger general-purpose models are all weakening the technical scarcity of vertical applications.

Putting fundamentals, competition, and valuation together, Meitu sits in a very typical and very hard-to-judge spot: the fundamentals are clearly better than a "theme stock," but the valuation has not yet become cheap enough to ignore execution risk. Google Finance gives a static P/E of 26.25x, while Yahoo Finance gives a trailing P/E of 33.42x—the two bases do not match. The root of the difference is that Meitu's 2025 IFRS profit was simultaneously affected by the high 2024 base from the crypto disposal, the one-time non-cash expense from the Alibaba convertible bond in 2025, and the reclassification between continuing and discontinued operations. Strip out this noise, and a rough owner-earnings estimate using 2025 operating cash flow minus capex implies the current share price corresponds to roughly 16x owner earnings, far less expensive than the headline P/E makes it look. The question is "whether it can prove this monetization structure for two more quarters," not "whether it is a bubble."

If I had to put one label on this company, I think it is closest to "a company in transition," and it has already passed through the ugliest stretch. It is no longer the old Meitu backstopped by hardware and traffic, but it has not yet become the typical SaaS with high retention, high seat penetration, and a high enterprise mix. It has found the entry point to a second growth curve, but the barrier still comes mainly from product understanding, brand, distribution, and the speed of scenario entry, not from a piece of underlying technology that others cannot replicate in the short term. The meaning of this label is clear: the company is worth studying seriously and worth continuing to track, but the research focus must be on subscription penetration, productivity ARR, usage-based monetization efficiency, overseas ARPU, and compute cost—not single-week price swings.

Vertical Development History

From Selfie Tool to Hardware Monetization

Meitu appeared early, seizing on the rigid demand among China's young mobile-internet users for "photo editing" and "selfies"—not AI. The company was founded in 2008 by Wu Zeyuan and Cai Wensheng. The annual report discloses that Wu Zeyuan began developing photo-editing software back in 2008; the prospectus shows that by October 2016, Meitu's apps had been activated on more than 1.1 billion unique devices worldwide, with monthly actives of roughly 456 million. What the product solved at first was simple: flatten Photoshop's complexity so ordinary people could edit a selfie into something "ready to post" in a few steps. This was Meitu's earliest product gift, and it determined why it could always package capability into mass-usable features faster than a pure-technology company.

The company's first large-scale commercialization relied on phones. The prospectus puts it bluntly: to meet user demand for high-quality selfies, Meitu launched a selfie-focused smartphone in 2013; in 2013, 2014, 2015, and the first half of 2016, smart-hardware revenue accounted for 59.7%, 87.8%, 89.9%, and 95.1% of total revenue respectively. This explains why many investors both loved and feared it at the IPO: it looked like an internet platform, but its revenue structure looked like a hardware company's. Hardware can amplify revenue, but it also brings inventory, channels, gross-margin volatility, and demand cycles—a structure that inherently cannot command the platform valuation it ideally wanted.

The IPO Narrative and Early Distortion

In December 2016, Meitu priced its Hong Kong listing at HK$8.50 per share, at the bottom of its offering range, raising roughly HK$4.88 billion, with an upper valuation cap of about US$4.5 billion during the roadshow. The story the company told the market at listing centered on enormous user scale, a strong brand, a social community, beauty e-commerce, and future diversified monetization, but what the market actually saw was continued losses and an excessive hardware share. In 2017, after its first annual report, the loss narrowed but the share price remained under pressure; that same year, MSCI removed Meitu from its indexes, dealing the stock another sentiment-driven blow. In those years, the market's deepest question about it came down to one line: are you a traffic company, or a company that sells selfie phones.

From Hardware Decline to the Subscription Inflection

The real turning point came around 2020. By then the company had clearly pulled back the hardware narrative and begun to put "subscriptions, in-app purchases, and advertising" back at the center of revenue. The 2020 results announcement stated explicitly that, despite pandemic disruption, revenue still grew 22.1% year over year to RMB 1.194 billion, and more importantly, premium subscription services and in-app purchases had become the second-largest monetization model after online advertising. In other words, Meitu finally began to get users to pay directly for the product itself, rather than relying on hardware spillover or pure ad traffic.

2022 was the second clear inflection. That year the company achieved its first full-year profit since the IPO. On an annual-report basis, revenue was RMB 2.085 billion, up 25.2% year over year; within that, VIP subscription revenue was RMB 782 million, up 57.4% year over year, and SaaS and related business was RMB 463 million, up a dramatic 1093.2% year over year. What this set of figures really shows is that "paid software finally worked at Meitu," not that "AI had already succeeded." That same year, total MAU was 243 million, still a large user base, showing it had found a more effective way to charge within its existing traffic pool rather than trading traffic for profit.

From AI Trial-and-Error to Core-Business Re-rating

Starting in 2023, Meitu entered a third stage: AI was no longer just a feature point but became the engine of a business re-rating. In 2023, revenue rose to RMB 2.696 billion and adjusted net profit was RMB 355 million, clearly above 2022's RMB 82 million. In 2024, the trend grew more solid: revenue was RMB 3.341 billion, up 23.9% year over year; within that, "imaging, video, and design products" revenue was RMB 2.086 billion, up 57.1% year over year, clearly rewriting the core business for the first time from the old "traffic + advertising" into "digital content tools + subscriptions." It was also in 2024 that the company invested more heavily in training large visual models, with R&D expenses rising to RMB 911 million, of which large-visual-model training cost was about RMB 140 million; that same year it disposed of all its cryptocurrency, bringing RMB 1.29 billion of cash inflow and a higher one-time profit base. The capital markets began to give it the new "AI application layer" label, but this round of relabeling was also mixed with a great deal of non-operating noise.

2025 brought this story into the stage of "from concept to delivery." In May, the company announced a strategic partnership with Alibaba across e-commerce, AI technology, and cloud capabilities, and issued a US$250 million three-year convertible bond to an Alibaba affiliate, with an initial conversion price of HK$6. By the 2025 annual report, the company's continuing-operations revenue was RMB 3.859 billion, up 28.8% year over year; adjusted net profit attributable to owners was RMB 965 million, up 64.7% year over year; paying subscription users exceeded 16.91 million, of which productivity-tool paying users were 2.16 million. The annual report also went into detail: as of the end of 2025, most of the company's products had integrated AI-agent capabilities, and the Agent feature of Meitu Design Studio rapidly became the main driver of billings growth from December 2025. The most important change at this stage is that AI is no longer just helping users edit photos but has begun to "do the work" for users in e-commerce, marketing, video scripts, product images, and similar scenarios.

Vertical Financial Review

Over the past few years, the driver of Meitu's revenue growth has shifted fundamentally twice. Before listing, growth came mainly from smart-hardware sales; in 2020–2022, the increment came from VIP subscriptions, in-app purchases, and SaaS; in 2023–2025, it concentrated further into imaging, video, and design products. The five-year financial summary in the 2024 annual report shows the company's revenue rising from RMB 1.194 billion in 2020 to RMB 3.341 billion in 2024, with adjusted net profit rising from RMB 49 million to RMB 589 million. In 2025, continuing-operations revenue rose further to RMB 3.859 billion and adjusted net profit attributable to owners rose to RMB 965 million. What this span really shows is that the company finally turned "massive users" into "sustained paying users."

Profit quality is better than the headline net profit. From 2022 to 2025, the company's cumulative operating cash flow was about RMB 2.65 billion, while cumulative net profit over the same period was about RMB 1.74 billion, so cash conversion is healthy. In 2025, operating cash flow was RMB 1.257 billion and capex was only about RMB 50 million, putting owner earnings roughly on the order of RMB 1.2 billion. The trouble comes mainly from IFRS noise: 2024 had a one-time gain from disposing of cryptocurrency, and 2025 had an approximately RMB 512 million one-time, non-cash IFRS 2 expense from the completion of the Alibaba convertible-bond issuance; meanwhile, fair-value changes in FVTPL long-term investments were a loss in 2024 and a gain in 2025, which also disturbs accounting profit. The conclusion is simple: looking at Meitu through the headline P/E will be misled by accounting items. Looking at cash flow and adjusted profit is closer to the real operations.

The balance sheet has actually grown steadier over these years. At the end of 2025, cash and equivalents were RMB 3.515 billion, and cash and other current financial resources totaled RMB 4.937 billion; total borrowings were only RMB 20 million, and the gearing ratio in the leverage measure was just 0.35%. Even counting the RMB 1.407 billion convertible bond as long-term debt, the company faces little funding pressure. More worth watching is the structural change: by the end of 2025, inventory had fallen from RMB 73.46 million at the end of 2024 to RMB 7.31 million, while contract liabilities rose from RMB 655 million to RMB 849 million. Read together, these two items mean the company is exiting low-quality supply-chain business and strengthening subscription and digital-service revenue that is more prepaid in nature.

Share Price and Valuation History

Meitu's price history is almost a history of "the market constantly rewriting its understanding." Early on, the market gave it a compromise valuation of "traffic-company vision, hardware-company reality"; after the 2017 MSCI event, the share price bore the pressure of passive money exiting. In 2021, the company bought about US$40 million of cryptocurrency and the share price jumped as much as 14.4% intraday, but this was more thematic trading than a fundamental inflection. In 2024, the company disposed of all its cryptocurrency, again making IFRS profit look unusually pretty. The 2025 Alibaba partnership and convertible-bond transaction made the market willing for the first time to re-rate it as an "AI-application monetization platform," and the share price surged to a seven-year high. By mid-2026, the market returned to calm: Google Finance shows a 52-week high of HK$12.56 and a low of HK$4.10, against the current HK$4.64. The high volatility is because the market has not decided whether to price it as a growth application, a light SaaS, or a theme stock—not because the company turns upside down every quarter.

The current headline valuation has differences in basis. Google Finance shows Meitu's static P/E as 26.25x on 2026-06-12, while Yahoo Finance gives a trailing P/E of 33.42x on the same day; what lies behind this is a difference in basis, not a market distortion. In the 2025 annual report, the company had both continuing/discontinued reclassification and one-time non-cash expenses and investment fair-value swings. Strip those out, and a rough owner-earnings estimate using 2025 figures implies the current share price corresponds to roughly 16x owner earnings, a valuation markedly below the headline P/E. For this company, whether the valuation midpoint can lift does not depend on "whether the market loves AI more," but on whether the paying intensity of productivity tools can hold steady across several quarters.

Business Model and Industry Analysis

Revenue Structure and Cost Leverage

Meitu's current revenue engine is already fairly clear. In 2025, on a continuing-operations basis, imaging, video, and design product revenue was RMB 2.954 billion, the absolute core; advertising revenue was RMB 843 million, already relegated to second place; other revenue was RMB 62 million, mostly non-core tail business. By Q1 2026, the company broke it down further in its official business update: imaging and design product revenue was RMB 852 million, up 34.3% year over year; within that, casual-scenario applications contributed 82% and productivity applications contributed about 18%, with the latter's revenue growing faster, at 45.4%. This means Meitu still rests on mass users as its base, but its profit elasticity is pushing upward from the more willing-to-pay productivity crowd.

The profit source of this engine is mainly subscriptions. In the first half of 2025, paying subscription users were about 15.4 million and the subscription penetration rate was 5.5%; by the end of 2025, paying subscription users exceeded 16.91 million; by March 2026, global paying subscription users further exceeded 17.9 million. More interesting is where the increment came from. The company's 2025 annual report states that the bulk of new paying subscription users in 2025 came from high-ARPU regions such as Europe, the Americas, and East Asia; in Q1 2026, paying users of productivity applications again grew 52.9% year over year, reaching 2.34 million. In other words, Meitu is migrating toward regions and scenarios that can pay more, rather than hard-pulling subscription headcount with rock-bottom unit prices.

The cost structure means this company has operating leverage but also two very hard ceilings. In 2025, the largest cost item was the revenue-share fees paid to payment channels such as Apple and Google, at RMB 618 million; the second key cost was compute, bandwidth, storage, and API-related costs of RMB 415 million, of which nearly half corresponds to inference compute. In the first half of 2025, the company also gave more intuitive data: payment-channel revenue share was RMB 281 million, up 45.1% year over year, while compute and cloud-related costs were RMB 107 million, up 15.0% year over year, clearly slower than paying-user growth. This means that, so far, the revenue increment from user growth runs faster than compute costs, and operating leverage is positive. But this positive leverage has a precondition: rising AI usage must be accompanied by simultaneous increases in pricing, conversion, and retention. As soon as compute usage grows faster than charging capability, gross margin will turn down.

Another improvement that often gets overlooked at Meitu is actively discarding low-quality revenue. In 2025, the company terminated the cosmetics supply-chain management business that dominated its former "beauty-industry solutions," with related revenue falling from RMB 345 million in 2024 to RMB 33 million in 2025, while the remaining AI skin-analysis SaaS was reclassified into "other." This will worsen statement comparability in the short run but is right in the long run, because the supply-chain business devours inventory, has low gross margins, and does not fit the "productivity and globalization" main line. After exiting this block, the company's revenue is lighter, and inventory fell sharply with it. Many investors who see revenue shrink mistake it for a growth problem; in reality, it is more like deliberate slimming.

Moat and Governance

Meitu's moat truly holds in three ways.

The first is brand and product mindshare. The company's prospectus once disclosed that an iResearch survey in June 2016 showed about 53.5% of photos on China's mainstream social networks had been processed by Meitu's apps; by the first half of 2025, the company still cited QuestMobile data stating that the Meitu app ranked first by MAU in China's mobile image-editing market. For a vertical tool to still hold the "default photo-editing entry" a decade later is itself a barrier. It is not unassailable, but it is certainly not something "any model with a wrapper" can take away.

The second is product understanding and workflow-packaging ability. Meitu's most valuable trait is knowing what users want to turn a given photo, video, or product asset into, rather than possessing a unique large model. It packs underlying models into clearly defined scenarios such as AI Removal, AI product images, smart editing, AI voiceover video, and batch image output. By the end of 2025, the company had rolled AI-agent capabilities across most product lines; in Q1 2026, AI-credit consumption rose 59% versus December 2025, with Kaipai, RoboNeo, and Meitu Design Studio growing faster. This barrier lies in scenario abstraction, interaction design, and aesthetic tuning, not in the underlying algorithms. It is shallower than a technology barrier but deeper than a pure-traffic barrier.

The third is the cross-product distribution network. Meitu is not just one app but a product matrix covering casual photo editing, video beautification, e-commerce assets, AI design, and AI video production. It can funnel the audiences accumulated in Meitu, BeautyCam, and Wink toward productivity tools such as Meitu Design Studio, Kaipai, WHEE, and RoboNeo. This ability to "screen heavy paying users out of light users" is a more valuable asset than a single hit product. The 52.9% year-over-year growth in productivity-application paying users in Q1 2026 reflects this funneling mechanism at work.

Its weakest link is the technology moat. Meitu itself is building vertical models and fine-tuning open-source models, and the 2025 annual report notes that third-party API costs remained at a mid-single-digit share of total costs, showing it does not fully depend on external closed-source models. The issue is that what is easiest to compress in this industry has never been "whether you can expose model capabilities," but "whether others can do the same thing at lower cost, with more traffic, across broader scenarios." This is exactly where Adobe, Canva, CapCut, Kuaishou's Kling, and more platform-style AI applications in the future will keep applying pressure. The moat exists, but it is not strong enough to ignore industry evolution.

On governance, Meitu has strengths as well as sources of discount. The strength is that founder Wu Zeyuan is still chairman and CEO, so business decisions and product judgment have not been outsourced to professional managers; in 2026 the company also launched a buyback plan of up to HK$300 million, and after the 2025 annual report proposed a final dividend of HK$0.05 per share, signaling that management is starting to take shareholder returns more seriously. The risks are threefold: first, the company uses a Cayman structure and controls its domestic business through contractual arrangements, so VIE risk has not disappeared; second, the Alibaba convertible bond carries director-nomination rights, allowing it to nominate one non-executive director as long as its holding size or shareholding ratio meets the conditions; third, Cai Wensheng remains a major shareholder, with the 2024 annual report showing his interest at about 23.45%, so the company remains a typical founder-led structure. Companies like this often decide quickly and inherently carry a governance discount.

Industry Structure and Cyclical Position

Meitu sits at the intersection of three industries rather than a single one: creative software, content-platform tools, and AI applications. This position has upsides and downsides. The upside is that the demand side is large enough—photo editing, design, video, e-commerce assets, and marketing content are long-standing high-frequency tasks; AI merely replaces "manual creation" with "semi-automatic generation" without creating demand out of thin air. The downside is that comparable peers are very fragmented: Adobe, Figma, and Canva look more like SaaS; CapCut and Kuaishou's Kling look more like platform extensions; Meitu is a light SaaS that grew out of consumer-grade imaging tools. As a result it is rarely fully understood by the market, but it also struggles to command the stable valuation midpoint of a pure-software leader.

The real sources of industry growth are fairly clear. First is rising subscription penetration, turning a small portion of originally free users into long-term paying users; second is AI bringing people who "can't make images, can't edit video, can't write scripts" to the table, expanding the paying pool; third is the persistent need from e-commerce, short video, and cross-border sellers for lower-cost visual assets; fourth is globalization, especially the natural amplifier of high-ARPU overseas markets. Canva disclosed 260 million monthly actives and US$3.5 billion in revenue in 2025, with ARR exceeding US$3 billion, essentially riding the same few winds. Meitu's difference from Canva is that it entered from mobile imaging first and then moved toward productivity; Canva grew out of templated design and team collaboration.

The most obvious cycles in this industry are the technology-iteration cycle, the advertising cycle, and the traffic-distribution cycle—not the inventory cycle of manufacturing. For Meitu, the variables that benefit most in an upcycle are paying-user count, penetration, ARPU, and usage-based consumption; the most fragile variables in a downcycle are advertising demand, model costs, and the App Store revenue share. Full-year 2025 advertising revenue declined slightly versus 2024, which already reminds the market of one thing: advertising can no longer be seen as a growth engine; it is more like a profit cushion. The real cyclical judgment has to watch the growth of imaging and design products, productivity ARR, and the conversion efficiency of AI credits.

On regulation, Meitu is not as heavily regulated as finance, gaming, or healthcare, but it is not light either. The annual report lists China's requirements on personal-information protection, telecom and internet-user-information protection, and online protection of children's personal information. For a company spanning images, video, cloud processing, and overseas users, this means content security, cross-border data transfer, privacy compliance, and generated-content compliance all cannot go wrong. Geopolitics affects it more indirectly, concentrated on the App Store ecosystem, overseas ad placement, cloud-service supply, and the pace of global expansion. The Alibaba partnership eases uncertainty over cloud and e-commerce channels on one hand, while making the market more attentive on the other to whether ecosystem binding will evolve into a new dependency.

Horizontal Competitor Analysis

Competitive Landscape and Niche

This is a track with "ample competitors but no perfect comparable." Compare Meitu with Adobe and you find the unit price and enterprise attributes are too far apart; compare it with Kuaishou's Kling and you find different platform attributes and traffic entry points; compare it with Wondershare and the business model is closer, but brand strength and domestic traffic base are not in the same league. The real situation is that Meitu stands at the intersection of consumer imaging applications, AI design tools, and lightweight productivity SaaS. It is neither a first-tier enterprise-software vendor nor a platform ad company; it is more like "a niche platform that grew from the consumer side and is moving toward the enterprise budget pool."

I therefore prefer to view comparables in three groups. The first is high-end global software companies with strong seats and deep workflows, represented by Adobe and Figma; the second is templated, collaborative, AI-enabled visual design platforms, represented by Canva; the third is players carrying platform traffic and video ecosystems into AIGC, represented by Kuaishou's Kling and ByteDance's products. What Meitu really competes for is part of the profit pool of the third and second groups: the budgets in the hands of e-commerce sellers, content creators, small and medium merchants, and lightweight professional users. It is further from Adobe's profit pool and closer to Kuaishou's and Canva's budget pools.

A Group Portrait Comparison

Adobe represents the endgame form of mature creative software: a vast subscription revenue base, deep workflows, strong enterprise-customer binding, and extremely high cash generation. Adobe's fiscal Q1 2026 revenue was US$6.40 billion, up 12% year over year; in June, with fiscal Q2, it also raised full-year revenue guidance to US$26.5 billion to US$26.6 billion, with AI-first ARR exceeding US$500 million. But the market's worry about it is sharp too: AI lowers the barrier to creative tools, and Adobe no longer enjoys the almost natural high valuation it once did—Google Finance shows its P/E at about 11.89x, and the share price has fallen to near an 8-year low. Comparing it with Meitu shows two things: first, even the leader in creative software gets repriced by AI; second, as long as workflows are deep enough, platform-grade software can still turn AI into a paid upgrade rather than a free giveaway. Meitu is still some distance from the second point, but it has already felt the pressure of the first.

Figma is a more instructive SaaS comparison. It builds a collaborative design and product-development platform, not photo editing. In Q4 2025, revenue was US$303.8 million, up 40% year over year, and net revenue retention rose to 136%; fiscal 2025 operating cash flow was US$250.7 million. Figma's most important reference for Meitu lies in revenue quality, not product form: when a design tool truly enters team collaboration and organizational workflows, growth can be steadier, retention higher, and cash flow significantly better than consumer-style subscriptions. Meitu has not reached this point yet; it is more "individual creators and small merchants pay first," with enterprise seat penetration only a distant prospect. Precisely for this reason, Meitu's growth narrative is sexier than Figma's, but its revenue quality is more fragile.

Canva is Meitu's closest strategic mirror image. Its monthly actives reached 260 million in 2025, with revenue of US$3.5 billion, and mid-2025 official disclosure put ARR above US$3 billion. What did Canva become? It blended design templates, team collaboration, brand assets, AI generation, and office documents into a visual workbench that "everyone can use." Users choose it because it shortens the path to "a usable finished product" to the minimum, not because it is the most professional. What Meitu did on mobile photo editing over the past decade-plus was, in fact, the same thing. The fork between the two companies is that Canva leans more toward teams and organizations while Meitu leans more toward individuals and merchants; Canva's collaboration attribute is stronger, while Meitu's accumulation in aesthetics and imaging processing is deeper. If Meitu can truly turn its productivity tools into a "light Canva," the valuation midpoint will lift markedly. If it cannot, it will stay long-term in the valuation range of "a high-quality application company."

Kuaishou's Kling is the most concrete AI-video rival. Kuaishou disclosed in Q4 2025 that Kling AI's single-quarter revenue reached RMB 340 million, and its single-month revenue in December 2025 exceeded US$20 million, corresponding to ARR of US$240 million. Kling's strength lies not only in model capability but in the short-video platform, marketing customers, and content-distribution scenarios standing behind it. For Meitu, this kind of rival is the hardest to fight, because its product capability and commercial entry are bound together. Meitu's advantage lies in being lightweight, cross-platform, able to do both images and video, with an aesthetic style better suited to consumer imaging and e-commerce assets; Kling's advantage lies in the explosive power of video generation, content-distribution linkage, and a stronger platform budget pool. As long as AI video keeps heating up, the budget contest between Kaipai, Wink, and Kling will become increasingly head-on.

Wondershare is a comparison closer to the "global creative-software company" path. It is a software vendor going global with tools such as Filmora, PDF, and mind mapping. Public secondary-market information shows Wondershare's 2025 revenue at about RMB 1.53 billion, with the loss narrowing, AI-native application annual revenue exceeding RMB 130 million, and AI invocation volume growing rapidly. The significance of this comparison is twofold: first, going global with creative tools is itself still viable; second, without a strong platform and brand mindshare as a foundation, AI commercialization easily falls into the trap of "features but no pricing." Meitu is stronger than it in user brand and imaging mindshare; weaker in that it is more easily treated by the market as a traffic stock rather than a software stock. The data here is less comparable than the companies above, so I treat it only as a path reference, not a strict valuation anchor.

Current Fundamentals and the Bull-Bear Divergence

Looking only at the past year's public disclosures, Meitu's operating state is in fact quite solid. In 2024, the company's revenue was RMB 3.341 billion, with imaging, video, and design product revenue of RMB 2.086 billion, up 57.1% year over year, and adjusted net profit of RMB 589 million; in the first half of 2025, revenue was RMB 1.821 billion, adjusted net profit attributable to owners was RMB 467 million, and gross margin was pulled up to 73.6%; for full-year 2025, continuing-operations revenue was RMB 3.859 billion and adjusted net profit attributable to owners was RMB 965 million; in Q1 2026, the company again updated 17.9 million paying subscription users, RMB 580 million productivity ARR, and RMB 852 million in imaging and design product revenue. Read together, this string of data does not look like a company "riding the AI concept," but more like a company actually putting AI into scenarios that can really charge.

The bulls' strongest evidence is fourfold. First, the core business grows far faster than total revenue, showing the structural improvement is real, not an accounting trick. Second, paying subscription users keep hitting record highs, and productivity users grow faster, showing the migration of consumer traffic to enterprise-type scenarios is not empty talk. Third, the growth of compute and cloud-related costs is temporarily below the growth of paying users, and gross margin is still climbing. Fourth, cash flow and the balance sheet are strong; in 2025, cash and other current financial resources approached RMB 4.94 billion, and the company also has an HK$300 million buyback plan, giving it ample financial latitude.

The bears' strongest evidence is also fourfold. First, the advertising business has slowed, with full-year 2025 advertising revenue declining slightly versus 2024, so the second growth curve must add increment soon. Second, growth in AI-credit consumption is good, but it also means inference costs will keep climbing—2025 compute, bandwidth, storage, and API costs already reached RMB 415 million. Third, the Alibaba partnership is both a channel positive and a dilution-and-governance variable. Fourth, industry competition is brutal; since 2025 the market has repeatedly hammered such stocks with the worry that "general-purpose models may replace vertical applications," and Meitu's price pullback from its one-year high is a textbook expression of that worry.

Valuation, Risk, and Catalysts

Valuation Analysis

First see through the cash flow. In 2023–2025, the company's operating cash flow was RMB 413 million, RMB 746 million, and RMB 1.257 billion respectively; over the same period, capex was about RMB 52 million, RMB 53 million, and RMB 50 million respectively. That is, over the past three years owner earnings rose roughly from RMB 361 million to RMB 1.207 billion. On a rough 2025 owner-earnings basis, the current share price corresponds to an owner-earnings multiple of about 16x; looking at 2025 adjusted net profit attributable to owners, the multiple is about 20x; looking at headline trailing P/E, Google Finance gives 26.25x and Yahoo Finance gives 33.42x. Three sets of valuations give three feelings: on owner earnings it does not look like a bubble; on adjusted profit the valuation looks neutral; on headline P/E it looks not cheap. I trust the first and second sets more, because this company's 2024–2025 accounting profit was too obviously disturbed by one-time items.

On historical position, the current price is at least not expensive on a "past-year" basis. Google Finance shows the 2026-06-12 share price of HK$4.64, close to the 52-week low of HK$4.10 and far from the 52-week high of HK$12.56; pulled back to the listing starting point, it also remains below the 2016 IPO price of HK$8.50. This shows the market has already disgorged a large stretch of AI optimism, and that the current price is not pricing in very aggressive long-term success. It is certainly not deeply undervalued, but compared with the market's most euphoric moment in mid-2025, its safety is much higher.

Horizontally, Meitu's pricing is sandwiched between two types of company. Compared with mature platforms like Adobe and Kuaishou, Meitu's headline P/E is higher; but compared with more standard design SaaS like Figma and Canva, it is clearly cheaper, because its revenue quality has not reached that level. Adobe's current static P/E is about 11.89x, Kuaishou's about 9.89x; Figma's revenue quality is better, but its GAAP profit is more distorted by the large post-IPO stock-based-compensation expense. The market's compromise pricing for Meitu largely expresses this judgment: you are no longer a theme stock, but you are not yet a high-certainty SaaS either.

The table below is the three-tier absolute-valuation framework I use. Its purpose is to discuss the delivery speed of "subscription + usage-based fees," gross-margin changes, and valuation multiples within the same framework—not to provide investment advice. Here I prefer the owner-earnings basis over headline net profit. The RMB-to-HKD conversion used in the calculation is a rough estimate, so the ranges are better treated as price bands than as targets accurate to the decimal. The core data relied upon comes from the company's 2025 annual report, the Q1 2026 business update, and the current share price.

Under this framework, I place the conservative value band at HK$3.4–4.0, the neutral value band at HK$4.5–6.0, and the optimistic value band at HK$7.4–9.0. The current price of HK$4.64 is already close to the lower edge of the neutral range but still above the upper edge of the conservative value band, meaning it is no longer significantly overvalued but does not yet offer a full margin of safety. It is more in the position of "you can hold it and wait for the answer, but don't treat it as a cigar butt."

My conclusion on the margin-of-safety review is "not obvious." The reasons are fourfold. First, the current price is a premium, not a discount, relative to the conservative scenario. Second, the most fragile assumption in the neutral scenario is "compute costs won't swallow the incremental profit," not revenue growth; if you take 70% of the owner-earnings assumption in the neutral scenario, the valuation falls back to around HK$3.1–3.7. Third, if earnings grow at zero over the next three years, shareholder returns will come mainly from the roughly 2% dividend and the slight share shrink from buybacks, and the return is not generous. Fourth, this looks much like a company where "a good business is taking shape, but the price has not become cheap enough to buy unconditionally." For new money, I lean toward waiting for a lower price, or waiting for the next results to further confirm "usage-based fees + overseas ARPU."

Risk Analysis

The business risk that truly warrants vigilance is that pricing power gets taken away, not the empty line of "intensifying competition." I rate the probability medium and the impact high. The trigger path is concrete: if ByteDance, Kuaishou's Kling, Canva, Adobe, or a stronger general-purpose model turns capabilities such as product images, short video, voiceover video, and image generation into low-priced or even free defaults, Meitu will be forced to make concessions on conversion and ARPU. At that point, paying-user count may still grow, but revenue elasticity and gross margin will be pinned down, and the capital markets will re-rate it from "AI-application platform" to "feature-type tool." For this risk, what to watch most is whether the growth of productivity ARR, paid penetration, and imaging and design product revenue declines for two straight quarters.

The second risk is compute-cost erosion. Probability medium, impact medium-high. The company explicitly disclosed in 2025 that compute, bandwidth, storage, and API costs were RMB 415 million, and the compute and cloud-related expenses recognized as cost of sales were RMB 232 million, nearly half from inference compute; in Q1 2026, AI-credit consumption again grew 59% versus December 2025. The transmission path of this risk is "the more usage, the faster the bill grows"—once usage-based fees and conversion improvements fail to keep up with usage growth, gross margin falls first, then profit margin. By that point, the market will not wait for the income statement to deteriorate before reacting; the valuation will shrink first. The best indicators to watch are the ratio of compute and cloud-related costs to imaging and design revenue, and whether gross margin falls below 70% for two consecutive periods.

The third risk is governance and ecosystem binding. Probability medium, impact medium. The Alibaba partnership gave Meitu e-commerce scenarios, cloud resources, and potential distribution help, but it also brought the convertible bond, potential dilution, and director-nomination rights. If the partnership goes smoothly, the market will see it as "backed by an ecosystem"; if the partnership advances slowly and the commercialization effect is not obvious, the market will in turn question whether the convertible bond bought enough strategic value. This kind of risk will not hit revenue first but will hit market expectations first. Hard indicators to observe are few, mainly two things: first, whether the Design Studio's commercialization in e-commerce scenarios keeps expanding; second, whether the Alibaba-related cooperation has gone from "signing" and "integration" to a sustained increment reflected in revenue and paying users.

The fourth risk comes from statement comparability and valuation misjudgment. Probability high, impact medium. Over the past two years Meitu has had multiple basis disturbances such as crypto disposal, convertible-bond IFRS 2 expense, continuing/discontinued reclassification, and FVTPL fair-value swings in long-term investments. They do not necessarily change the business, but they change headline profit. If investors judge by static P/E, they easily mistake cheap for expensive when it is cheap, and expensive for cheap when it is dangerous. The best solution for this risk is to consistently look at the same set of owner-earnings and adjusted-profit bases, rather than hunting for "a more accurate P/E."

Catalysts and Tracking Indicators

The most weighty positive catalysts over the next year are three things. First, productivity-application ARR keeps accelerating, ideally pushing from the current RMB 580 million toward RMB 800 million to RMB 1 billion. Second, usage-based fees not only grow AI-credit consumption but also grow revenue and gross margin in step, proving that "subscription + usage" is not a high-cost boom. Third, new paying users from high-ARPU overseas regions continue to be the main force, showing globalization relies on products that truly sell rather than low-price buying of users. The sustained execution of buybacks and the implementation of Alibaba e-commerce scenarios will also provide marginal positives.

Negative catalysts are equally concentrated. First, the advertising business keeps declining, showing the base business no longer provides a cushion. Second, core-business growth looks like it is still there, but gross margin starts to turn down. Third, the market again treats general-model upgrades as a negative for vertical applications, and growth-stock valuations fall collectively. Fourth, if the company can only prove "usage is rising" but cannot prove "cash is rising" for two consecutive disclosure cycles, this round of re-rating will stop.

The table below is the dashboard I think is truly worth watching. The data cited is the company's most recent disclosure, and the warning thresholds are research discipline, not company guidance.

Among these indicators, only three are truly critical: paying users, productivity ARR, and gross margin. The first two decide whether it can still grow, and the third decides whether that growth is worth anything. Advertising, overseas MAU, and contract liabilities are more like verification items. Cash and buybacks decide whether the company can prop up a floor itself when market sentiment is at its worst.

Key Data Table

The table below sums up the core data I cite repeatedly in the main text, for easier side-by-side reading. 2025 revenue has a comparable-basis difference due to the discontinued-operations reclassification, noted in the table's footnotes.

† 2025 is on a continuing-operations basis; the company's disclosed comparable continuing-operations revenue for 2024 was RMB 2.996 billion, putting the 2025 year-over-year growth on that basis at 28.8%.

Looking at a few key reference companies' public trading and operating snapshots makes Meitu's position more intuitive. The table below shows only the most core market snapshots and does not try to force different business models onto one ruler.

What this set of numbers reveals is not as simple as "who is expensive and who is cheap." Adobe and Kuaishou are both larger and steadier, and both bear valuation pullbacks because "AI is both an opportunity and a threat"; Wondershare is smaller, showing creative software is not a naturally high-valuation track. The reason Meitu can still maintain an above-average growth valuation is that its operating delivery over the past two years has indeed been decent, not because the market is generous. If it stops delivering, it will quickly fall back to the pricing logic of a tool stock.

Horizontal-Vertical Synthesis

The Bull and Bear Cases

Bull case:

• Core imaging and design products have become the largest and fastest revenue engine, up 45.2% year over year in the first half of 2025 and up 41.6% for full-year 2025 on a continuing-operations basis.

• Paying subscription users rose from about 15.4 million in mid-2025 to over 17.9 million in Q1 2026, and growth is still continuing.

• Productivity tools have begun to show second-curve characteristics, with ARR of about RMB 580 million in Q1 2026 and paying-user growth of 52.9%.

• Cash flow and the balance sheet are strong, with 2025 operating cash flow of RMB 1.257 billion and current financial resources of RMB 4.937 billion, and the company has already launched an HK$300 million buyback plan.

• The current price has pulled back sharply from its 52-week high, the headline valuation is distorted by one-time items, and the real owner-earnings multiple is much lower than the apparent P/E.

Bear case:

• The advertising business has shown fatigue, declining slightly for full-year 2025 versus 2024, indicating the old base's support for growth is weakening.

• AI usage is rapidly amplifying, and compute and cloud-related costs will follow upward over the long run, with 2025 total related costs already reaching RMB 415 million.

• The moat leans more toward product and scenarios than irreplaceable underlying technology, and it lacks an overwhelming advantage against Adobe, Canva, Kuaishou's Kling, and general-purpose models.

• The reporting basis is complex and the headline profit easily misleads valuation judgment, so investors who look only at static P/E easily get it wrong.

• The Alibaba partnership brings channel and cloud support but also introduces the convertible bond, potential dilution, and governance variables.

Pre-mortem

If this investment loses 50% three years from now, I think the two most likely scripts are as follows.

The first script plays out on the competition side. Around 2027, platform-style products and general-purpose models turn product images, short-video scripts, voiceover video, and image generation into lower-priced "platform features," and CapCut, Kling, Canva, and Adobe Express together compress the willingness of light-to-medium creators to pay. To defend its users, Meitu has to stuff more and more AI features into the existing subscription package, usage-based-fee growth slows, gross margin falls from the current 73% area back to around 65%, and owner earnings stall at RMB 700 million to RMB 800 million. The market simultaneously compresses the valuation from 17–20x owner earnings to 10–12x, and the share price falls to around HK$2, nearly halving from the current price. The key to this script is that per-user value is taken by the platform, not that users disappear.

The second script plays out on the cost and execution side. In 2026–2028, AI-credit consumption keeps surging, but the company fails to smoothly convert usage into revenue, or the revenue it does convert is insufficient to cover inference cost, payment-channel cuts, and continued R&D. On the surface, paying users and usage are both still high, but the operating cash flow in the statements no longer improves in step; advertising struggles to recover, and the Alibaba partnership fails to materially open up e-commerce monetization. The market finally admits that "this is a company with decent growth but limited pricing power" and assigns a lower software multiple. This script may not crash the company's fundamentals, but it is enough to make the stock fall a lot first.

Final Research Conclusion

Along the way, Meitu has truly proven two abilities. First, it has always deeply understood how ordinary users want to make "photos, videos, and product assets" look better, and it has always been better than most technology companies at packaging complex capabilities into simple, smooth, chargeable products. Second, it finally turned this product understanding into real money after 2020: subscription headcount keeps growing, imaging and design revenue has become the main force, and cash flow looks more and more like a mature software company's than an internet company's with a traffic illusion. Its past success owes something to the era's dividend and something to management's long-term grasp of user needs. Today, the era's dividend is still here and management's ability is still here, but the competitive barrier has fallen, and you can no longer infer "Meitu will surely stay strong" from "Meitu was once strong."

Horizontally, Meitu's most real advantage is that it stands between consumer imaging and light productivity workflows, with both large-scale user mindshare and proven partial software pricing ability. Its most real weakness is that this advantage has not yet gone deep into enterprise workflows—retention, seat expansion, and team-collaboration attributes are all still insufficient. The market's misjudgment of it was, in the past, treating it as a traffic stock inside a hardware shell, and may now be overtreating it as "a fragile application layer that models could eat at any time." I think both views are incomplete: Meitu is neither a pure theme nor a worry-free software platform. It is more like a company turning from a "traffic-aesthetics tool" into an "AI visual-production tool," and this transformation is already half complete.

At today's price, my conclusion is: the company is worth continuing to hold while waiting for the answer, and new money need not rush to load up. The reason is simple. The current price is much safer than the 2025 high, and the valuation on an owner-earnings basis is not high; yet it has not fallen low enough to ignore competition, compute, and execution risk. The most critical variables over the next year are productivity ARR, overseas high-ARPU paying users, and the revenue conversion of AI credits; over the next three years, the most critical variable is whether "subscription + usage-based fees" can move from experiment to stable institution; over the next five years, the key is whether Meitu can truly raise itself from a mobile imaging tool into a higher-stickiness visual workbench. As long as the answer is still "in the process of proving," its stock is better suited to discipline by range than to chasing the price by imagination.

【Company Profile Scores】

• Fundamental quality: Medium

• Growth: High

• Moat: Medium

• Financial soundness: Strong

• Management credibility: Medium

• Valuation attractiveness: Medium

• Risk level: Medium

• Suitable investor type: Long-term growth

【Investment Rating】

• Rating: Hold

• One-line investment thesis: the core subscription business is already working; usage-based fees and global expansion still have to keep delivering.

• 【Ideal Buy Price】3.4–4.0 HKD Rationale: corresponds to roughly 15–16x owner earnings under the conservative scenario, leaving a fuller buffer for rising compute costs and intensifying competition.

• Acceptable holding price: 4.5–6.0 HKD

• Clearly overvalued price: 7.4–9.0 HKD and above

• Current price classification: acceptable to hold

• Worth waiting for a better price: yes. For new money, below HK$4.0 is more attractive; or wait until the next results further prove productivity ARR, overseas paid penetration, and compute-cost control before accepting a slightly higher price. The opportunity cost of waiting is that if the company proves "subscription + usage" is delivering smoothly in two consecutive updates, the share price may revise up ahead of time.

• Target holding period: 1–3 years

• Expected annualized return: conservative about -20%; neutral about +13%; optimistic about +77%

• Maximum loss risk: about 50%–60%; the trigger is general-purpose models and platform-style competitors compressing pricing power, with gross margin and valuation multiple declining together.

• Signals that trigger a reassessment: if imaging and design revenue growth falls below 15% for two consecutive disclosure periods; if gross margin stays below 70% for two consecutive disclosure periods; if productivity ARR growth drops below 25%; if compute and cloud-related cost growth persistently exceeds core-business revenue; if new overseas paying users slow while advertising weakens at the same time.

【Valuation Range】

• current: 4.64 (as of the 2026-06-12 close)

• bear (conservative · ideal buy zone): [3.4, 4.0]

• base (fair · acceptable holding zone): [4.5, 6.0]

• bull (optimistic · above the clearly-overvalued line): [7.4, 9.0]

Research Uncertainty

This report still has four blind spots. First, Meitu does not, like U.S. SaaS, continuously disclose full quarterly financials—Q1 2026 is a business update, not a complete income statement. Second, the company does not publicly disclose retention, ARPU, and CAC for each major product, so the unit economics of productivity tools can only be judged indirectly. Third, Canva and ByteDance's products are not full listed comparables, so the horizontal comparison is naturally asymmetric on the financials. Fourth, retail data sources differ on the basis of Meitu's headline P/E, so this report places more weight on the owner-earnings and cash-flow framework and does not treat a single static P/E as the core anchor.

Reference Sources

This report mainly uses the following public sources: Company and exchange filings, including Meitu's 2025 results announcement, 2025 interim report, 2024 annual report, Alibaba partnership and convertible-bond announcements, buyback-plan announcement, and founder share-purchase announcement. The company's official business updates and media pages, including the Q1 2026 business update and product and company introduction pages. Authoritative financial media and quote pages, including Reuters, Google Finance, and Yahoo Finance. Official materials of comparable companies, including official financials, investor-relations pages, and official press releases of Adobe, Figma, Kuaishou, Canva, and others. Industry and product research references, including a16z's 2026 Top 100 Gen AI Consumer Apps report.

Other Tickers Mentioned in This Report

• ADBE.US — the global creative-software leader, used to benchmark Meitu's pricing power and valuation-compression risk in the AI era.

• FIG.US — a collaborative-design SaaS benchmark, used to gauge how far Meitu is from high-retention, high-seat-penetration software.

• 01024.HK — Kling AI directly competes with Kaipai and Wink for content and marketing budgets in video generation.

• 300624.SHE — Wondershare is a listed reference for the global-expansion path of creative software, used to compare the overseas-commercialization quality of AI tools.

• 09988.HK — Alibaba is both a strategic partner and a convertible-bond investor, affecting Meitu's cloud costs, channels, and governance structure.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.