For an AI application company, surviving matters more than innovating: Liblib turns cash-flow positive
An AI application company accused of "only copying + buying traffic" has been cash-flow positive since May, with paid advertising revenue at just 4%. Founder Chen Mian says: the era of applications hasn't arrived yet, and the most important goal is to stay alive; between being first to find PMF and ultimately winning the market lies a vast distance.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Chasing high gross margin at the application layer actually suppresses demand
The outside world says we blew up, but in May and June the company's cash on hand kept growing and cash flow was positive; they say we buy traffic, but we spend less than 1 million RMB per month on traffic, and paid advertising revenue is only 4%, because mainstream video media can't open accounts for AI tools — we couldn't spend the money even if we wanted to. Chen Mian attributes the outside attention to AI being too hot and the application layer being a topic, and says he "doesn't deserve" so many comments. On gross margin, he states his short-term target clearly: no more than 30% and positive — the application layer doesn't control the model, its greatest value is users, and chasing high gross margin would instead suppress demand.
— Chen MianAI membership pricing bets that users won't use up their credits
The 3.9 discount on LibTV memberships at launch was understood by outsiders as a 3.9 discount on Chat API, but Chen Mian says this is a misunderstanding: it was a 3.9 discount on the monthly-annual package. The essence of pricing is balancing two assumptions — renewal rate and consumption rate. If a user buys 1 million credits and only uses 200,000, the remaining 800,000 is profit; the bet is that "the vast majority of users won't use up their credits over the year," the same logic as a gym betting that annual-card holders won't come often. He admits that if users used all their tokens at the original pricing, the company would definitely lose money; but this was the norm in the previous era of tool software, except that this era adds the token scale.
— Chen MianThe real panic wasn't at 4,000 RMB left, it was at 6 million
The story of having only 4,000 RMB left in the early days is true, but Chen Mian says the real panic began when there was only 6 million left on the books. At the time he had just finished a vicious fight with a wealthier rival: because he put all his energy into competition, he turned down all fundraising, didn't even add investors on WeChat, and burned 3 million to 4 million USD in a few months. The first mistake he identifies in retrospect is "founders don't look at the books, thinking that once the business takes off there will be money." Only after the battle ended did he frantically meet investors, and found no one willing to invest in a company about to go under. This experience shaped the "trade time for space" undertone of his later strategy, and explains why he is extremely sensitive to cash flow.
— Chen MianBeing first to find PMF is a world away from winning the market
Facing the originality controversy, Chen Mian admits LibTV was not the first product to find PMF, but stresses that "being first to find PMF and ultimately winning this market are a world apart." He believes it's hard for an interface to have 0-to-1 original innovation, that the optimal interface for a vertical will converge, and that engineering innovation has a shelf life of only days to a week. He is criticized harshly because "it wasn't the first PMF, yet it reached relatively the largest scale at extremely fast speed." The real moat isn't the interface, but marketing rhythm, word-of-mouth touchpoints, and the ability to scale quickly. He also responds to acquisition rumors: initially he wanted to invest a little, but the other side ignored him; later he judged this market to be bigger than his main business, so he entered it himself.
— Chen MianManagement matters least, but ignoring it is already costing you
Chen Mian quotes "management doesn't matter but organization and culture do," then adds his own line: management is relatively the least important in this era, but the cost of ignoring it is now showing. He breaks out two reward models: the first is business-oriented — whoever fits goes up, whoever doesn't gets pulled — which he did well, and which is why the company grew; the second he didn't do well — what happens to the people pulled down? The company didn't give them enough respect or feedback to find a new place. He admits he is sometimes hasty in hiring, having casually introduced 10 co-founders externally in the early days because he "didn't take titles seriously," but the other party treats it as a promise. Only one CTO truly left, because the company was about to go under, but 4 technical leads reporting directly to him left.
— Chen MianThe near-death moment wasn't running out of money, it was the day Sora launched
The moment that truly approached death wasn't having only 4,000 RMB left, but seeing Sora launch in January. The team had spent a year exploring the "workflow control" direction, but a new model directly internalized the workflow — "it just hit the next stage directly." Chen Mian was in Shanghai at the time; the moment he opened his laptop and saw the page he was in tears, and called a shareholder who was fundraising to say "we're finished." Later, on the flight from Shanghai back to Beijing, he figured out the direction — shifting from asset production to providing context, a new product combining agent and workflow. After landing he excitedly found the marketing lead and said "do it now." This was also the first time he cried in front of a shareholder, but he later said that cry was closer to a breakdown than the one after a startup failure.
— Chen MianWhile the giants fight for the Central Plains, you can first unify Jiangdong
Chen Mian uses the Three Kingdoms metaphor to explain the independent survival path: when the giants are contending for the Central Plains, they have a priority order, and Jiangdong is not their highest priority — that's the startup's time gap. "When he enters the Central Plains, your unifying Jiangdong has nothing to do with his entering the Central Plains." Trade time for space, then space for resources, and finally pour resources into building a moat — a real moat needs time to be watered, "any moat built overnight is not a real moat." Geographic advantage is also dynamic: at first it might be Bashu, and only by building the Yangtze River natural barrier does it become Jiangdong. Today's Bashu is the "Shu Road is hard" between multimodality and text — general products can't easily become professional creation tools, but he also admits this path "isn't that thick."
— Chen MianOnly by surviving first do you earn the right to talk about innovation and moats
In closing, Chen Mian sums up his three years of entrepreneurship as "driving a super-fast car, fixing it while driving, people getting on and off, scenery passing by outside the window," but with no time to stop and say hello and say goodbye. He admits his biggest regret is not having time to say goodbye, and can only grip the steering wheel and look ahead. Asked about the scenario "suppose the model company is a train and you might just be a small car attached outside it," he says he too fears being thrown off, but wants even more to build the car well in the process. This echoes the opening — the survive he repeatedly stresses, above all specific tactics, is ultimately a philosophy about time: only by surviving first do you earn the right to talk about anything else.
— Chen MianIn their own words · checked verbatim
If I'm not aggressive, how can I be formidable? If I'm not formidable, how can I survive?
不侵略性强，我怎么厉害呢？我不厉害，我怎么活下来呢？
Chen Mian20:03
Please, innovation has a cost, it has a price. If you innovate something and you can't defend it, can't do it well, what's the point of innovating it?
拜托创新是有成本的，是有代价的。你如果创新了一个东西，你都把它守不住，把它做不好，你创新它的意义是啥。
Chen Mian58:25
When he enters the Central Plains, your unifying Jiangdong has nothing to do with his entering the Central Plains.
当他逐入中原的时候，你一统江东跟他逐入中原没有关系。
Chen Mian1:31:40
So essentially it's whether you believe that path exists or not. I believe that path exists.
所以本质上是你相信那条路存在，还是不相信那条路存在。我相信那条路存在
Chen Mian1:38:03
I think entrepreneurship is driving a super-fast car, and you fix it while driving, people get on and off, scenery passes by outside the window, but you have no time to stop and say hello and say goodbye to these people, and no time to properly look at the scenery outside. All you can do is grip the steering wheel.
我觉得创业就是开了一台。超级快在跑的车，然后你边开车边修车，然后车上的人来来去去，窗外的风景也来来去去，但你没有时间去停下来跟这些人say hello和say goodbye，然后也没有时间好好看窗外的风景。你能做的事情就是握紧方向盘。
Chen Mian2:06:34
Figures
| Company funding valuation and amount | 2 billion USD valuation, raised 300 million USD | 1:00 |
| May-June cash flow | Positive, cash on hand kept growing | 5:14 |
| Paid advertising revenue share | About 4% | 5:14 |
| Monthly traffic-buying spend | Less than 1 million RMB | 5:14 |
| Short-term gross margin target | Greater than 0% and no more than 30% | 8:28 |
| Lowest cash on hand | Less than 4,000 RMB | 22:13 |
| Number of technical leads who left | 4 left, only 1 CTO truly left | 48:35 |
| Lib's contribution to revenue | More than half | 47:22 |
| LibTV peak daily revenue | 1 million USD | 1:26:23 |
| Global annual video creation revenue ceiling | About 100 billion USD | 1:27:26 |
Glossary
- PMF / product-market fit
- The state in which a product exactly meets a real and scalable market need.
- Reward model
- The source borrows the reinforcement-learning term to metaphorically describe incentive and feedback mechanisms in management.
- LTV / lifetime value
- The total revenue a user contributes over their entire usage lifetime.
- GTM / go-to-market
- The playbook and rhythm from definition to launch for a new product or new market.
- AI native
- Talent naturally adapted to the AI-era way of working, with high-density capabilities.
- ARR / annual recurring revenue
- Recurring revenue of a subscription business annualized, a core metric in valuation narratives.
How to listen
AI application founders, investors watching application-layer valuations, and product leads doing growth and management in the shadow of model vendors.
The 1:46-1:54 segment about being an internet-addicted teenager and theater days can be skipped; if you only care about the strategic survival path, go straight to the Jiangdong metaphor at 1:28.