Who Really Profits From OpenAI and Anthropic IPOs?
AI may change the world.
That does not automatically mean that shareholders of AI model companies will be the biggest winners.
As IPO speculation around OpenAI and Anthropic grows, the market is beginning to shift from asking “How powerful can AI become?” to “Who can actually capture the profits from AI?”
That distinction matters.
AI can make individuals more productive. It can help companies write code, summarize documents, handle customer support, translate text, analyze data, and automate parts of knowledge work.
But productivity gains do not always become corporate profits.
A company may save time with AI, but that saved time does not automatically become higher revenue or lower costs. It may simply become more meetings, more internal documents, more checks, and more administrative work.
This is especially true in countries like Japan, where AI-driven productivity gains may not quickly translate into layoffs or direct labor cost reductions.
AI can be useful without immediately becoming highly profitable for the companies providing it.
AI’s Value May Be Real, but the Profit Pool Is Complicated
The core issue is not whether AI has value.
It clearly does.
The issue is where that value goes.
Some of it may go to AI model companies such as OpenAI and Anthropic.
Some of it may go to cloud providers.
Some may go to semiconductor and memory companies.
Some may go to data center operators, electricity providers, and networking infrastructure.
Some may go to companies that successfully redesign their workflows around AI.
And some may simply go to consumers who receive powerful AI tools at low prices.
In other words, AI’s social value may be large while its profits are distributed across many different players.
This is one reason why the AI investment boom is more complicated than it first appears.
The First Clear Winners May Be Infrastructure Companies
If AI companies continue to compete, they need more compute.
That means more GPUs, more high-bandwidth memory, more cloud capacity, more data centers, more electricity, more cooling, and more network infrastructure.
In that sense, the clearest early winners may not be the AI model companies themselves.
They may be the companies that provide the infrastructure behind the AI boom.
This resembles a classic pattern in technology booms.
During a gold rush, the people selling tools, transportation, and infrastructure can sometimes capture profits more reliably than the people digging for gold.
AI may follow a similar pattern.
The model companies are the most visible.
But the infrastructure providers may be the first to monetize the boom in a more direct and measurable way.
OpenAI and Anthropic Still Have a Real Path to Profit
This does not mean OpenAI or Anthropic cannot become huge winners.
They can.
But their path is narrower than the simple phrase “AI will change the world” suggests.
To justify very high valuations, model companies need pricing power.
They need to become deeply embedded in enterprise workflows.
They need to become hard to replace.
If AI becomes a core operating layer for companies, supporting research, coding, sales, customer service, legal work, finance, cybersecurity, and internal decision-making, then companies may be willing to pay high recurring fees.
In that scenario, OpenAI or Anthropic could become something closer to an enterprise operating system than a chatbot provider.
Strategic AI models also create another path.
Advanced AI systems used in cybersecurity, critical infrastructure, defense-related work, drug discovery, and national security may command high prices because they are not ordinary consumer products.
But those markets also come with restrictions.
The more powerful the model, the more likely it is to face access controls, audits, regulation, liability concerns, and government oversight.
That can support high pricing, but it can also limit scale.
AI Adoption Can Become a Cost of Competition
There is another risk.
AI may become so widely adopted that using it stops being an advantage.
At first, early adopters may gain a competitive edge.
They can produce faster, automate tasks, reduce friction, and improve productivity.
But once every competitor uses similar AI tools, the advantage disappears.
AI then becomes a basic requirement for staying in the game.
At that point, AI spending may start to look less like a source of extraordinary profit and more like a necessary operating cost.
This is similar to websites.
Having a website once gave companies an edge.
Today, having one is simply expected.
AI may move in the same direction.
The Real Question
The question around OpenAI and Anthropic IPOs is not whether AI is impressive.
It is.
The better question is this:
Who can turn AI’s value into stable, durable profit?
Right now, the most obvious near-term answer may be infrastructure companies.
The next strongest group may be large platform companies that can integrate AI into existing businesses such as cloud, office software, search, advertising, operating systems, and enterprise services.
OpenAI and Anthropic may become major winners if they can become deeply embedded in enterprise workflows or dominate high-value strategic AI markets.
But if AI becomes widely available, increasingly substitutable, and expensive to operate, model companies may face a harder path than the current excitement suggests.
AI may change the world.
But that does not mean the shareholders of AI model companies will automatically capture most of the value.
The full Japanese version of this article explores the historical comparison behind this argument and looks more deeply at what may happen after AI company IPOs.
You can read the full version here:
https://note.com/sekahan_0623/n/ndf7f5b4002e3?sub_rt=share_pw
Even if you do not read Japanese, note and browser translation tools should make the article accessible in English.
