Original video in Chinese.
Key Takeaway
- Mid-sized companies are being hollowed out by AI: they have neither the traffic and infrastructure of big platforms nor the extreme flexibility of small teams/individuals. AI is lowering internal management costs at the same time it lowers external transaction costs, which benefits big companies and individuals/small teams, and makes the “organizational arbitrage” of mid-sized companies disappear quickly. They are the most uncomfortable.
- The essence of why companies exist is to reduce transaction costs: Coase’s theory points out that a company substitutes management costs for external transaction costs. But AI makes management cheaper, and also makes external collaboration (finding people, signing contracts, delivering work) simpler. A company is no longer the naturally optimal organizational form, and project-based collaboration (the Hollywood model) will rise on a large scale.
- The future is a dynamic collaboration network: individuals become independent nodes, each with their own AI system and judgment. When a project comes along, people form a temporary team, split the money after the work is done, and disband. The goal of the newtype community is to gather such super individuals amplified by AI and form a high-quality project-based collaboration network.
In the AI era, there is one kind of company that will have a miserable time, and that is the mid-sized company.
They are not big enough, and not small enough.
Not big enough to control platform distribution.
Not small enough to have the flexibility of individuals and small teams.
Mid-sized companies are stuck in the middle, belonging to neither side. This is not because the boss is not trying hard enough, and it is not because the employees are not grinding enough. It is because AI is changing something more fundamental:
the reason companies exist.
Think about it: what exactly is a company for?
Many people think a company is just a group of people getting together to do things. But that understanding is too superficial.
The economist Coase explained long ago that the reason firms exist is that outside transactions are too troublesome.
You have to find people, negotiate prices, sign contracts, and supervise whether the other side is actually doing the work properly.
When you add all that up, the cost is too high.
So companies came into being.
They turn external transactions into internal management. There is no need to negotiate every time; the boss gives an order, and the employee executes.
Put simply, a company substitutes management costs for transaction costs.
As long as management costs are lower than transaction costs, a company has value.
But after AI arrived, this equation changed.
For large companies and big platforms, AI will make internal collaboration cheaper and cheaper.
Work that used to rely on lots of processes and meetings for coordination will be absorbed by AI going forward.
So big companies and big platforms will keep getting bigger.
On the one hand, they already have the ability to grow because they control traffic, payments, data, and credit;
on the other hand, the internal coordination factors that once constrained their growth have already been solved by AI. They can be both big and flexible.
Now let’s look at the other side: super individuals and small teams.
AI rapidly lowers external transaction costs.
In the past, if you wanted to build a product, you might have needed to hire front-end, back-end, design, and operations people.
Now, one person with a few AIs can build a prototype, put together the pages, publish the content, and run through user feedback.
In the past, if you wanted to find people to collaborate with, it was a hassle.
Now, AI can help you do all of that.
So being small is no longer a problem.
A team of three to five people can do the work of a team of three to five dozen.
A super individual can operate like a company.
At that point, the most uncomfortable ones are the mid-sized companies.
They are not platforms, so they do not have distribution or infrastructure; nor are they small teams, so they are not low-cost and lightning-fast to pivot.
These mid-sized companies have offices, lots of middle managers, regular meetings, management processes, and fixed salaries. They have a bunch of people they have to keep on the payroll.
In the past, that was called organizational capability.
Now, it has become a burden.
So you can see that in the past, the greatest value of a company was to organize a group of people stably. But now the situation is different: AI makes organization cheap, so the company’s original advantages are no longer as valuable.
Those mid-sized service companies, consulting firms, and outsourcing companies fundamentally earn their money from organizational arbitrage. For example, the PR company I worked at for more than ten years before. They were basically matching, managing, and delivering in the middle, then taking a layer of profit.
Now that there is AI, clients can go directly to small teams. Employees can become small teams themselves. That middle layer of “organizational arbitrage” will get thinner and thinner.
When you get to this point, you realize that we may need to rethink what a company actually is.
A company is not a naturally correct organizational form. It is just a tool used in a certain era to reduce transaction costs and distribute benefits.
Joining a company is, at its core, joining an agreement about how money is split. In other words, who gets a salary, who gets equity, who bears risk, who gets the residual return — those are the core of a company.
So the truly important question in the future is not which company should I join? It is which collaboration structure should I enter?
Is it a platform structure? Or a project structure?
A platform structure is something everyone understands. So what is a project structure?
Actually, this is not new; Hollywood has already perfected it.
When Hollywood makes movies, it does not keep all the directors, screenwriters, actors, cinematographers, and editors on a single company payroll for the long term.
A project comes along, and everyone comes together temporarily. You write the script, he shoots, that person understands special effects. Everyone collaborates around a specific work.
When the project is over, they split the money and disband. When the next project comes, they form a new team again.
In the past, this model was only suitable for the film and television industry. That is because it requires a very high level of trust, contracts, reputation, and revenue-sharing systems. Ordinary industries could not afford to play this game.
But AI will lower the threshold for this whole system. For example, communication can be assisted by AI; contracts can be assisted by AI; project management can be assisted by AI; knowledge accumulation can also be assisted by AI.
So project-based collaboration will spread from a few industries to more knowledge work industries.
Note: not all industries. Industries like manufacturing and basic research still require long-term organization. But content, software, consulting, design, education, and marketing will show this very clearly.
So in the future, more and more organizations like this will appear:
Usually, everyone is an independent node. They have their own judgment, their own work, and their own AI workflow.
When a project comes along, they form a temporary team.
When the work is done, they split the money.
People do not maintain order through a superior-subordinate relationship. They distribute benefits based on work, credit, and contribution.
That is also why I built the newtype community.
The ultimate form of newtype must be a collaboration network. It will gather many individuals amplified by AI. Each person will have their own judgment, their own skill stack, and their own small system.
They will grow separately in normal times. When a good project comes along, they will team up like Hollywood. Whoever can contribute something will get something.
After the work is done, they split the money, review what happened, and move on. Next time, they reorganize.
This sounds a bit idealistic. But I think it is actually very realistic.
Because in the AI era, what is truly scarce is this: who has judgment; who can deliver; who is worth collaborating with.
So for now, the goal of the newtype community is still selection. I want to select people who can judge independently, deliver independently, and also collaborate at a high level with others.
If you understood this episode, welcome to join through Knowledge Planet.
OK, that is the content for this episode. If you want to understand AI, want to become a super individual, and want to find like-minded people, come join our newtype community. See you next episode!