Having worked in MCNs for years, I hear the same claim every so often: ‘MCNs are about to become obsolete.’ When short-form video took off, some said creators could make content on their own and no longer needed institutions. When livestreaming grew, others said streamers could face users directly, making MCNs unnecessary. As platform capabilities improved, people said platforms would take over all of an institution’s work. Now it is AI’s turn. And this time, the argument sounds more plausible than before. One person with a few AI tools can already write scripts, create images, generate video, produce voiceovers, edit, translate, and even operate several accounts at once. Work that once required ten people may eventually be handled by two. The production capabilities that MCNs once took pride in are rapidly becoming accessible to ordinary creators as well. So this time, I do not particularly want to defend MCNs. If an MCN’s value is limited to helping creators get production done, that value is indeed becoming less distinctive. The real question is: if production is no longer scarce, why do institutions need to exist?
When we helped found OST in 2017, MCN was still a term far less familiar than it is today. Over the years, we have moved through short-form video, creator commercialization, livestreaming, group livestreaming, and now AIGC. Looking back, every shift in platforms has redefined the value of institutions. Early on, one essential capability was finding people. Later it was developing them. Then came traffic operations, commercial monetization, livestream operations, supply chains, and team management. As the business grew, so did the headcount. I, too, once naturally saw that scale as a moat. But after AI emerged, I began to reconsider: if a team once needed 100 people simply because a great deal of work had to be completed manually by 100 people, were those 100 people truly an asset—or a cost left behind by insufficient technology? As AI enters organizations, it may be the organization itself—not content—that is repriced first.
When people discuss AI’s impact on the content industry, they often begin with whether it will replace writers, directors, editors, or streamers. From an operator’s perspective, though, I am more interested in another category of work: the coordination, handoffs, and repetitive execution that fill the middle of the production chain. Organizing materials, synchronizing information, generating basic scripts, breaking down data, producing variants, adapting content across languages, answering recurring questions, checking standard processes—people once had to do all of this, while AI Agents and automation tools are particularly suited to it. AI may not remove the most exceptional creators first. It may first reduce the work that exists inside organizations because information does not flow efficiently enough.
Suppose that, in the future, a mature AIGC creator can independently handle topic selection, scripts, visuals, video, sound, editing, multilingual versions, publishing, data analysis, and even basic commercial communication. Why would that person still join an MCN? ‘We can help you make content’ is not compelling. ‘We have editors’ is not compelling when the creator has AI. ‘We can help operate your account’ is not compelling when an Agent may be operating it around the clock. The real issue is not whether AI will eliminate MCNs. It is what MCNs can still offer that individuals may find difficult to possess on their own once production capability is released from institutions into the hands of individuals.
In the past, the basic business unit of an MCN was the creator. Operations, editorial staff, and commercial teams were assigned around one person to keep producing content and monetizing it. AIGC is introducing another possible business unit: the content system. A system may contain several AI characters, a dozen accounts, a shared world, multiple language versions, multiple platforms, a group of Agents, and a production-and-review process. It need not correspond to a single human creator—or even to a single account. Where we once asked, ‘What should this creator post next?’ we may instead ask, ‘What should this content system generate today?’
Traditional MCNs like to calculate labor efficiency: how many streamers can one operator manage, and how many accounts can one editor or producer support? Because human time is fixed, expanding scale has usually meant adding people. AI may change that curve. What may be worth measuring in the future is not ‘How many accounts can one person manage?’ but ‘With AI, how many stable content systems can one person manage?’ One video demonstrates model capability. A hundred pieces of consistently strong content begin to demonstrate production capability. Maintaining quality across a hundred accounts begins to test organizational capability.
If production becomes increasingly inexpensive, the capabilities MCNs may truly need to build are judgment, organization, asset management, commercialization, and distribution. Ten thousand pieces of content can be generated; which ones are worth making? A hundred characters can be created; which ones deserve continued investment? Those questions require judgment. Coordinating people, models, Agents, accounts, data, and commercial resources within a single process requires organization. Who a character is, who holds the likeness rights, whether a voice can be used, and in which countries it can be used all require asset and rights management. Being able to generate content does not mean someone will pay for it. Why a brand would buy, how to price it, how to deliver it, and how to earn repeat business all require commercialization. Bringing the same content to different platforms and countries is not simply a matter of clicking to publish it simultaneously. AI lowers the barrier to content production, but it does not automatically lower the barrier to operating content as a business.
A rather unusual kind of company may emerge: one with fewer and fewer people managing more and more content. ‘Large’ may no longer mean having many employees. It may mean managing more content assets, controlling more IP, reaching more markets, holding more user relationships, and being able to mobilize stronger production capabilities.
But a large quantity does not equal industrialization. Producing 100 pieces a day, launching 100 accounts at once, or having one person control 50 characters may all sound very AI. But if most of those 100 pieces receive no attention, and if none of the 100 accounts has a clear reason to exist, that is merely excess generation capacity. True industrialization should at least mean that standards can be repeated, quality can be controlled, costs can be calculated, problems can be traced, experience can be reused, and results can be reviewed. I would rather ask: Is the 1,000th piece smarter than the first?
The MCN of the future may increasingly resemble less a traditional talent agency and more a Content Operating System. At the base, it connects to different models. Above that, it manages human creators, AI characters, and IP. In between, Agents perform a large share of production and operational tasks. Further up, it connects platforms, brands, commercial clients, and global markets. Human work gradually shifts from repetitive execution toward setting objectives, defining rules, training systems, judging results, handling exceptions, building relationships, and making final decisions.
When discussing AI transformation, it is easy for owners to say, ‘Employees need to learn how to use AI.’ But perhaps the owners are the ones who need to learn AI first. If the organization remains unchanged and every employee is merely asked to become 20% more efficient, the likely result is just an old company that runs faster. The harder questions are: Which processes should not continue at all? Which lessons should enter the system? Where should the labor and time that are freed up be reinvested? AI’s greatest challenge for MCNs may not be ‘How do you use AI?’ but ‘Do you dare redesign your company?’
I do not believe AI will rescue every MCN. On the contrary, it could accelerate the loss of relevance for some MCNs, especially institutions whose core capabilities rest on information asymmetry, low-cost execution, repetitive labor, simple matchmaking, and platform tailwinds. Your competitor may not be a larger MCN. It could be three people who understand content deeply, working with a very strong AI system.
Yet I believe in organizations more than I used to. The stronger the tools, the greater individual capability becomes; but as possibilities multiply, organizational complexity may rise as well. Inefficient institutions may fall behind, while highly capable individuals and new forms of organization may grow stronger at the same time. Individuals gain production capabilities that once belonged only to institutions, and institutions must develop system capabilities they did not previously need.
Why does an institution exist? Not because it has 500 people, not because it has signed 5,000 people, and not because it can produce 50,000 pieces of content a day. Those are measures of scale, not value. The real question is: without you, could these creators, characters, content assets, and commercial resources work better?
That is why I would rather call it ‘reinventing the MCN.’ In the past, we used people to scale content. Next, we may need to learn something else: to manage more content with fewer people; to retain more assets with stronger systems; and to let people stop trying to produce everything, so they can decide what is worth producing. AI has not automatically eliminated MCNs. It has simply pushed older MCNs to demonstrate why they exist—and to reinvent what a content company can be.
Copy this essay for someone thinking about AIGC content, IP, organisation or digital identity.
