China’s collaborative office sector is undergoing a far-reaching reshaping of its landscape. Based on current growth trends, the industry judges that Feishu has a chance to challenge for the top position in revenue scale in China’s enterprise collaboration market within 2026. On the other side, its old rival DingTalk is in a period of management turbulence, having changed leaders twice in 15 months, and its strategy has entered a window of adjustment. Against this backdrop of one rising as the other falls, public opinion can easily reduce Feishu’s rapid growth to simply “seizing DingTalk’s existing customers.” But once the surface-level competition is set aside, the underlying logic actually driving the market shift has long since changed: with the arrival of AI, the yardstick by which enterprises choose an office platform has been completely rewritten. A group of medium and large enterprises with extremely high migration costs are proactively replacing their collaboration foundation, and what lies behind this is not a simple product price comparison, but enterprises searching for a unified vehicle for digitalization in the AI era.
Dual-Track Growth: Expanding the Circle Outward, Digging Deeper Inward
Feishu’s growth in this round is supported jointly by two curves: horizontal customer expansion and vertical deepening of procurement.
Looking back over its development, Feishu’s early labels were very distinct: rooted in the internet, hard tech, and new energy vehicle companies, it captured knowledge-intensive enterprises through its documents, meetings, and collaboration experience, and was long regarded by outside observers as better suited to digital-native companies. But over the past two years, the boundaries of its customer base have been rapidly widening. Enterprises in the consumer, manufacturing, and energy sectors — such as Luckin Coffee, Haitian Flavouring & Food, Nanjing Steel Group, Zijin Mining, Xtep, Jin Jiang Hotels, Lee Kum Kee, Guming, and Adopt A Cow — have successively moved onto Feishu, and the platform has formally crossed beyond the internet circle to penetrate deeply into the real economy. The industry phenomenon of the “Feishu content rate” (含飞率) is becoming increasingly prominent. When multiple leading enterprises in the same sector make the same choice, it sends a strong signal: Feishu has already completed validation at scale in complex organizational scenarios, and the selection risk for peer enterprises continues to decline.
A few years ago, when enterprises procured collaboration software, AI was merely an optional add-on function; today, enterprises plan their AI implementation schemes at the same time as they finalize their office foundation. The procurement logic has shifted from “first adopt collaboration tools, then pilot AI later” to “use a unified foundation to carry all of the enterprise’s AI business.” Horizontally broadening its industry map and vertically raising the depth of procurement per customer — these two forces superimposed have lifted up the strongest commercialization curve in Feishu’s history.
Public information shows that Feishu’s 2025 revenue exceeded RMB 3 billion (approx. USD 442 million), while DingTalk’s revenue over the same period exceeded RMB 4 billion (approx. USD 590 million), with the revenue gap between the two continuing to narrow. Competition over existing customers certainly exists, but relying solely on seizing customers can hardly support triple-digit growth rates. The incremental market brought by AI is the core key to Feishu breaking through its growth ceiling.
From Office Tool to the Operating Foundation for Enterprise Agents
In the mobile internet era, the criteria by which enterprises selected a collaboration platform were clear and convergent: put the organization online, connect attendance, approvals, and notifications, and move offline management processes online. This set of requirements cultivated the first round of the domestic collaborative office market, and also caused the basic capabilities of the major vendors to become homogenized.
After the AI wave arrived, new evaluation criteria surfaced. Enterprises gradually realized that simply embedding an AI chat box, generating document summaries and meeting minutes, can only improve individual employee efficiency. If AI cannot read the enterprise’s private knowledge base, cannot connect to business data, cannot comply with organizational permissions, and is difficult to embed into business processes, then in the end it can only remain at the “icing on the cake” pilot stage.
The difficulty does not lie in purchasing a sufficiently powerful large model, but in the lack of the organizational context to carry AI operations. Large models master general knowledge, yet they do not understand a given enterprise’s internal division of authority and responsibility, its historical projects, its supply chain information, or its store operating rules. To let AI evolve from a “chat tool” into an agent capable of solving business problems, it must have a unified system of documents, data, processes, and permissions as support.
This is precisely the foundation on which Feishu has formed its differentiated advantage. Messaging, cloud documents, knowledge bases, multidimensional tables (Base), approval processes, and organizational structure are natively connected, and all of the assets accumulated through an enterprise’s daily collaboration can naturally be opened up to AI. Agents can follow the enterprise’s original permission rules, retrieving information, initiating processes, and linking up business operations within security boundaries, avoiding data leakage and unauthorized operations.
On this basis, Feishu has built a dual-path agent system of “native + open.” Feishu aily supports enterprises in building dedicated industry Agents on the basis of their own business data; Feishu CLI opens up interfaces, allowing external agents such as OpenClaw to connect to the platform and call on capabilities such as documents, calendars, and messages within the scope of authorization. The platform is no longer limited to self-developed AI products; the goal is to build a unified environment that is compatible with all kinds of large models and agents and enables humans and AI to work collaboratively.
The yardstick for judging the value of AI is no longer how much text is generated or how many times interfaces are called, but operational metrics such as quality inspection accuracy, labor cost savings, and problem-handling efficiency. When AI can act directly on core business areas such as production, stores, and the supply chain, procurement budgets also begin to move beyond the IT and administrative departments and extend to the front lines of production, operations, and sales, opening up entirely new incremental space for the SaaS industry.
From Selling Seats to Embracing AaaS, the Industry’s Business Model Undergoes Transformation
For a long time, the collaborative office sector has followed a single logic: charging subscription fees according to employee seats. But as Agents are deployed at scale, this business model is beginning to show its limitations. Agents can execute tasks continuously across positions and across processes, and the number of employees cannot measure the value created by AI.
Overseas vendors have already taken the lead in exploring new models: Salesforce and Agentforce have introduced mechanisms that charge by agent action, and Microsoft Copilot supports pay-as-you-go. In the future, a hybrid billing model combining seat subscriptions with usage volume and business outcomes will gradually become mainstream.
This means that the upper limit of Feishu’s future growth will no longer depend purely on how many new employee seats are added, but more on how many business scenarios AI agents penetrate, how much repetitive work they take on, and how much quantifiable benefit they create. The industry is slowly transitioning from traditional SaaS toward AaaS (Agent as a Service). Of course, this transformation cannot be accomplished in one step, and seat subscriptions will still serve as a foundational revenue source for a long time to come.
Behind the opportunity there are likewise challenges. At present, most enterprise AI projects are still at the pilot stage, and the renewal rate of AI modules and the results of large-scale implementation remain variables yet to be validated. At the same time, Feishu’s past advantages have been concentrated in medium and large enterprises, and how to make the complex solutions accumulated with top-tier customers lighter and more standardized, and push them toward regional industrial clusters and small and medium-sized enterprises, is a problem that must be overcome in the next stage.
In the first half of the collaborative office game, the contest was over who could better connect people with people; in the second half of the AI era, the focus of competition becomes who can build a stable and reliable foundation that connects people with agents.
DingTalk is in a window of strategic adjustment, which leaves an expansion window open for its rival, but this is only a phased external factor. The underlying logic behind Feishu’s sustained rapid growth is that it has stepped precisely into a new round of the enterprise digitalization selection cycle. The high cost for enterprises of replacing their collaboration foundation is enough to show that choosing Feishu is not merely choosing a set of office software, but laying out in advance the infrastructure capable of carrying the enterprise’s long-term AI implementation.
Over the next several years, the key to winning or losing in the market will no longer be how many benchmark customers are harvested, but whether replicable industry AI solutions can be continuously accumulated, allowing agents to truly take root in business processes. When AI is no longer a standalone function but a part of organizational capability, the new landscape of the collaboration sector will only just have begun to be written.
[Disclaimer]: The above content reflects analysis of publicly available information, expert insights, and BCC research. It does not constitute investment advice. BCC is not responsible for any losses resulting from reliance on the views expressed herein. Investors should exercise caution.