The AI Monetisation Crisis: How OpenAI and Anthropic Are Changing the Economics of Intelligence

The AI industry is entering a very important stage, and technological progress is no longer the main problem. Instead, the real question is about money: can AI really make a lot of money?

This question is no longer just a theory for companies like OpenAI and Anthropic. It is immediate, structural, and a matter of life and death.

The economics behind AI are starting to fall apart after years of rapid progress, huge amounts of money, and widespread use. What once seemed like a simple plan—build strong models, get users, and make money later—is now showing deep cracks.

The Myth of Inexpensive AI

In the beginning, it seemed like access to generative AI was almost free.

Users could use powerful models for little or no cost. Developers could use APIs to build things without having to worry too much about limits on how many times they could be used. Startups quickly added AI to their products, thinking that costs would go down over time, just like with traditional software infrastructure.

But that idea isn't working out.

The truth is that AI, especially modern systems that rely on reasoning, costs a lot of money to run. Training models costs a lot of money up front, but the real cost driver has moved to inference, or the ongoing use of these systems.

Every question, every answer that is generated, and every automated workflow uses up computer resources. And as AI systems get better, those costs go up a lot.

AI Agents Are Messing Up the Cost Model

This problem has gotten worse because of the rise of AI agents.

AI agents are made to do complicated, multi-step tasks, unlike simple chatbots. They think, repeat, check outputs, and often start new processes to finish one request. This makes them much more useful, but it also makes them much more expensive.

These systems use a lot more tokens than older models, and they often do hidden calculations that users can't see.

The outcome is a basic mismatch:

• People want AI to work like software

• Providers are dealing with infrastructure that needs a lot of computing power

This is where the monetisation crisis starts.

The Change from Subscription to Pay-Per-Use Pricing

AI companies are quickly changing how they charge for their services to deal with rising costs.

Flat-rate subscriptions used to be the norm, but they are no longer possible. Heavy users, especially those who run AI agents, use a lot more resources than pricing models thought they would.

So, businesses are starting to do:

• Pricing based on usage linked to tokens

• Plans with stricter limits in tiers

• Limitations on third-party integrations

For instance, Anthropic recently limited how its models could be used in some agent frameworks, which pushed users toward more expensive pricing plans.

This isn't just a change in price; it's a sign that the business model isn't working well.

The Change Is Caused by Investor Pressure

There is a huge financial reality behind these changes.

Hundreds of billions of dollars have been invested in the AI industry, and even more money has been promised for future infrastructure like data centers, GPUs, and energy systems.

Investors won't keep giving money to this ecosystem forever. They want returns.

AI companies need to make trillions of dollars in sales over the next few years just to get even small returns on their investments.

This puts a lot of pressure on:

• Make more money

• Make operations more efficient

• Put high-value use cases first

In real life, this means that the time of cheap or free access to AI is coming to an end.

The Cost-Quality Trade-Off

This change makes it harder for businesses that use AI to find the right balance between cost and performance.

Companies now have to think carefully about when to use expensive models and when to switch to cheaper ones. Some people are already using a mix of different methods, such as:

• High-end models for important jobs

• Open-source or smaller models for everyday tasks

Even small drops in the quality of output can have a big effect on business, which makes these choices hard and high-stakes.

This is different from traditional SaaS, where costs on the margin are pretty stable and easy to predict.

The End of the "Infinite Scale" Idea

A lot of people thought that scale would fix everything during the AI boom.

More users means more data, which means better models, which means more money.

But AI doesn't grow like regular software.

AI systems don't benefit from almost no extra costs; instead, they have to pay for every interaction. This changes the economy in a big way.

In fact, using it more often can sometimes make the business less profitable, not more so.

This inversion is making businesses completely rethink how they plan to grow.

A Time for the Industry to Come Together

The AI industry is likely to consolidate as these pressures grow.

Not all businesses can afford the infrastructure costs that come with competing at the highest level. A small number of players with:

• A lot of capital reserves

• Infrastructure that works well

• Strong business ties

will be able to live for a long time.

For some people, the way forward may be:

• Focus on specific use cases

• Using open-source models

• Buyout by bigger platforms

What This Means for SaaS and Business Tech

The effects go far beyond the AI companies themselves.

The move toward token-based economics is a big change for SaaS companies in how they price and deliver software.

Companies may start to pay based on the following instead of per user:

• Use of computers

• How hard the task is

• Delivery of results

This makes software prices more in line with how much infrastructure they use, which makes it harder to tell the difference between SaaS and cloud services.

The Big Picture: AI as a Way to Make Money

The monetisation challenge ultimately uncovers a more profound aspect of AI.

It's not just a product; it's a whole economy.

There is a cost to every interaction. More computing power is needed for every improvement. Every time you scale up, you need more infrastructure.

This is what makes AI different from other waves of software innovation.

Last Point of View

The AI field is at a turning point.

The thrill of fast innovation is now running into the facts of cost, infrastructure, and making money. Companies like OpenAI and Anthropic are being forced to make difficult decisions—raising prices, limiting access, and restructuring their business models.

These changes might seem limiting at first, but they are part of a necessary change.

The future of AI will not be determined by the development of the most powerful models, but rather by the ability to render them economically sustainable.

And that is a much more difficult problem to solve.