Behind the $300 million financing and $2 billion valuation, China's AI applications have for the first time achieved a full-fledged product matrix.

In the past year, the narrative focus of the AI industry has been centered around models.

From OpenAI, Anthropic to DeepSeek, large model companies have attracted the majority of attention. However, at the same time, another commercialization path is rapidly emerging - AI applications.

Recently, Evoken, the parent company of LiblibAI, completed a nearly $300 million Series B+ financing round, with a post - investment valuation exceeding $2 billion. This is also the highest single - round financing for a Chinese AI application company to date.

This round of financing was jointly led by Granite Asia, Tencent, and Shunwei Capital, with HT Investment and Jeneration Capital participating in the investment. Existing shareholders such as GaoRong Capital, Ant Group, Yingce Capital, MingShi Venture Capital, Source Code Capital, Sequoia China, and several other well - known investment institutions continued to increase their support.

More worthy of attention than the financing amount is another set of data: As of May 2026, Evoken's ARR has reached $300 million, nearly tripling compared to when this round of financing was completed.

In China, it is one of the few AI application startups outside of large tech companies to enter the "hundred - million - level ARR" club.

This means that Evoken's revenue - increasing leverage is not a single product with accidental success but is based on the successful implementation of the Product - Market Fit (PMF) of multiple AI products.

Looking back at the evolution of the AI industry in the past three years, three of the most important application waves can be clearly seen: image generation, Agents, and video generation. Evoken has presented its representative works in almost every cycle:

LiblibAI, an AI creator community launched in 2023, has accumulated over 30 million users. One out of every three designers in China is an active user of LiblibAI. The AI design Agent, Xingliu, launched in July 2025, has served a cumulative user base of tens of millions.

LibTV, an AI video creation platform launched in February 2026, has refreshed the speed of self - sufficiency for domestic AI applications. In its first month of launch, LibTV's daily revenue exceeded one million US dollars. Two months after its launch, the product's revenue increased by more than 13 times.

While the industry is still discussing how to find the AI PMF, this company has started to answer a more practical question: How can AI truly become a profitable business?

From LiblibAI to Xingliu and then to LibTV, what this company is trying to build may not just be a single hit application but the first AI content matrix in China.

Using three products to become one of the most profitable AI companies in China

The AI industry is never short of hit products but lacks companies that can continuously create them.

A large number of star products have emerged in the AI industry in the past three years, but the vast majority of companies still remain at the stage of "having only one super - product." For example, Cursor is labeled as AI Coding, and Suno is labeled as AI Music. Undoubtedly, these products have achieved great success, but as of now, these super - products still bear the main burden of generating revenue for their parent companies.

This actually stems from a dilemma: most AI companies can find one PMF but do not have the ability to continuously replicate it. For instance, Character.AI has tried multiple directions such as communities, Agents, and games, but the label that is often remembered is still "AI character companionship," and it is not easy to establish a second growth curve.

In contrast, from LiblibAI to Xingliu and then to LibTV, Evoken has almost fully experienced the three technological cycles of image generation, Agents, and video generation:

In 2023, as Midjourney and Stable Diffusion brought AI image generation into the public eye, a large number of startups began to pour into this field. Evoken's LiblibAI chose to enter the market from the creator community and model ecosystem, filling the gap of the "tool - seller" for domestic multi - modal models.

Subsequently, the AI application entered the Agent era. Represented by Manus, the industry began to explore how to make AI move from "generating content" to "completing tasks." At this stage, Evoken launched the AI design Agent, Xingliu.

This year, with the release of high - performance video models such as Seedance 2.0 and Kling 3.1, and the rapid growth of the downstream short comic drama market, Evoken quickly launched LibTV. In the video generation track that emphasizes "single - shot" generation, it was the first to establish the concept of "delivering finished videos" among downstream customers.

For the AI industry, achieving PMF once proves product capabilities; achieving it three times in a row proves organizational capabilities.

One of Evoken's team's methodologies is to discover opportunities brought by changes in model capabilities earlier than others. Chen Mian, the founder of Evoken, summarized it as two things in an interview: First, closely follow model iterations; second, the internal team has aligned on an assumption that models are getting stronger, but in the short term, they are more like humans and have not yet surpassed them.

△ Chen Mian, the founder of Evoken

In his view, the compulsory course for application - layer companies is "how to leverage cutting - edge models," that is, to make the best use of the latest models in the shortest time. Compared with model companies that focus on the boundaries of capabilities, application companies are more concerned about the inflection points of capabilities: When a model develops a new capability, what problems that couldn't be solved before can be solved? What new interaction methods will be spawned? Which workflows will be re - engineered?

The birth of Xingliu is a typical example.

Before the launch of high - performance image generation models such as GPT - Image - 1, the Evoken team predicted that model manufacturers were focusing on solving problems such as complex multi - round instruction understanding, consistency control, and editing capabilities.

If these problems are solved, the most core interaction method of design software may change - users no longer need to learn complex toolchains but can continuously collaborate with AI through natural language to complete design. Based on this prediction, the team bet on the "ChatCanvas" product form for Xingliu in advance.

However, simply understanding technological changes is not enough. In the past few years, many AI startups have been able to keenly sense the progress of models but may not be able to convert technological advantages into real - world needs. Compared with discovering technological opportunities, identifying market opportunities is often more difficult.

Evoken's second ability is to disassemble, re - engineer, and ultimately productize downstream needs.

LibTV is an application born at the intersection of model capabilities and downstream needs. In the eyes of the outside world, the core issues in the video generation track are the beauty of the shots and the accuracy of understanding. However, after communicating with a large number of customers, the team found that what short comic drama teams, MCN agencies, and advertising companies really lack is not the ability to generate single shots but the ability to produce complete content.

Only by being able to integrate into the entire production chain and help customers complete the delivery of works can real commercial value be created. Therefore, from the very beginning, LibTV aimed not at the video generation model itself but at the video production workflow.

This concept actually runs through Evoken's product development path in the past few years: LiblibAI solves the problem of creators obtaining and managing AI materials; Xingliu solves the problem of the design workflow of human - AI collaboration; LibTV solves the problem of delivering finished videos.

On the surface, they belong to different tracks, but they follow the same logic: Instead of looking for the strongest aspects of the model, look for the most in - need - of - re - engineering links in the industrial chain.

And this may also be the most important rule for AI application startups: In a rapidly growing incremental market, the most important thing is not to be different but to do the right thing at the right time.

Chinese AI applications are entering the "group war" era

At the beginning of 2026, the popularity of AI remained high, but several "hit" products declared "death." According to the statistics of AI Graveyard, 392 AI tools around the world stopped service in 2025. This means that in the past year, an AI product died on average every day.

The most shocking "sudden death" came from AI giant OpenAI. On March 25, 2026, OpenAI announced the removal of the Sora application from the market. This hit product, which had reached 10 million downloads upon release, only lasted for 25 months.

As Bryan Kim, a partner at a16z, said, "There is no moat in the consumer - level AI field." A significant signal is that the narrative of single - hit products is becoming obsolete.

The rapid iteration of model capabilities is swallowing up AI applications. At the same time, the iteration of coding capabilities has rapidly reduced the cost of replicating hit products. As a result, the lifecycle of hit AI applications has shortened, competition at the product level has intensified, and customer acquisition costs have also increased. According to feedback from some industry practitioners, in 2024, the average customer acquisition cost (CAC) of AI products was still between $20 and $30; now, the figure has risen to over $100.

In this context, "grouping" has begun to become a new way for AI enterprises to build barriers.

Compared with a single product, a multi - product matrix has stronger business risk - resistance capabilities. More importantly, grouping means that enterprises are no longer competing for a single tool track but for the ecological niche of the industry. For latecomers, it may not be difficult to replicate a single product, but it is much more difficult to replicate an ecosystem composed of multiple products, tens of millions of users, and a complete business system.

Evoken is one of the first batch of Chinese AI application companies to enter the "group - warfare" stage.

First, look at the horizontal product matrix. From LiblibAI to Xingliu and then to LibTV, there is a clear main line in Evoken's product evolution: the delivery of creative content across AI technology cycles.

This means that the users of the three products highly overlap. Compared with single - point products scattered in different scenarios, they can share users, data, and commercialization capabilities.

For example, image creators in the LiblibAI community may need the AI design Agent, Xingliu, to further assist in creation; image creators may also transform into video creators. Different products serve as traffic entrances for each other, naturally extending the user lifecycle.

Then, look at the vertical content industry ecosystem. From LiblibAI providing creative inspiration and material generation to Xingliu and LibTV delivering visual design, Evoken's multiple products together form a complete content production chain.

Especially LibTV, which was the first to propose the "dual - entry of users and Agents," is not just a product designed for humans but more like building infrastructure for the Agent era. As AI moves from "answering questions" to "completing work," more and more content production links will be encapsulated into callable capability modules, and video generation is one of the most core links.

In other words, today's LibTV serves creators, and in the future, it may serve Agents. As more and more Agents start to participate in creative and content workflows, whoever masters the key production capabilities such as image, video, and design will have the opportunity to become an important entrance to the next - generation content ecosystem.

In business history, it is a common story that companies become bloated and their actions become distorted as their business expands. When a three - year - old enterprise rapidly grows into a "group," the organization faces increasingly severe tests: How to improve organizational efficiency? How to ensure the accuracy of decision - making?

Evoken's core answer is: speed.

In a public interview, Chen Mian once said, "Speed is the rarest barrier in an era of frequent model updates and short product lifecycles." For example, 36Kr learned that LibTV took only one month from project initiation, user interviews, R & D, to final launch.

Behind the speed is an organization built around content - creative products. Chen Mian described Evoken's organization as "having no product managers, only designers" and "only people who teach AI."

The logic behind this employee profile is that "when tools are intelligent enough, 'people who manage requirements' are no longer needed, but 'people who define requirements' are even more important." In simple terms, "industry know - how" will become the core asset of the team.

In the context of global AI application companies, the valuation logic of Evoken may need to be re - examined.

From the simplest perspective of PS (Price - to - Sales ratio), Evoken is in an obvious undervalued position globally. A typical comparison is: Suno, an AI music creation tool founded in the United States, reached an ARR of $300 million in March 2026, corresponding to a valuation of $5.4 billion; while Evoken, with the same ARR volume, has a post - financing valuation of only about $2 billion, less than half of Suno's.

In terms of revenue scale, growth rate, and commercialization capabilities, the two have reached the same level, but there is a significant gap in the market's pricing.

However, what may be more worthy of attention is not the PS itself but the changing valuation system of AI application companies.

In the past two decades, the capital market has been used to measuring software enterprises with the logic of SaaS companies: software is just a tool, and the real value is created by the people using the tool. Therefore, the valuation of an enterprise ultimately depends on indicators such as subscription revenue, the number of customers, and the renewal rate.

However, in the AI era, the role of tools is undergoing a fundamental change. As Chen Mian said, "We cannot use the thinking of the tool era to understand the tools in the AI era. The essence of SaaS is that services are provided by people, and people use tools. Now, AI has become the main provider of services."

Therefore, the value anchor of AI applications is shifting from "software seats" to "labor seats."

In the past, enterprises bought software; in the future, enterprises will buy digital employees that can continuously deliver results. The standard for measuring an AI company will gradually shift from how many tools it sells to how much work it undertakes and how much productivity it creates.

In this sense, Evoken's value should not be simply regarded as that of an AI tool company. The AI content creation matrix it has built is essentially re - engineering the production mode of the content industry.

When the market starts to measure AI applications with "digital labor" instead of "software tools," Evoken's $2 - billion valuation may just be the beginning.