Who Is Paying for AI? The Monetization Problem No One Talks About

Generative AI is full of promise. OpenAI’s Sam Altman talks about GPT-5’s “PhD-level” reasoning and lightning-fast responses. According to IDC, generative AI is expected to contribute $19.9 trillion to the global economy by 2030. Billions of people are using popular chatbots to write, code or design in seconds.

But here’s the truth: every single one of the answers you get from a chatbot costs real money to generate. Unlike software or games that can be built once and sold a million times, AI doesn’t scale that way. Each response is custom-made for one user at a high marginal cost. Someone has to pay for it.

Right now, the free prompts people type into ChatGPT are subsidized by investors burning cash to keep the lights on. But it can’t go on like that much longer. The math is brutal: give it away and you go broke, put the service behind a paywall and your reach shrinks immediately.

Why AI is different from software

Traditional software has an almost magical property: once it’s built, it can be replicated at nearly zero marginal cost. Excel, Photoshop or Candy Crush can all be downloaded millions of times without incurring more expenses.

Generative AI is structurally different. Each query requires real energy and processing power. A search query on a large AI model can cost up to ten times more than a traditional Google search. This is why the stakes are so high for AI monetization. Without sustainable revenue models, companies will drown in infrastructure costs long before they realize IDC’s multi-trillion-dollar prediction.

Why ads won’t save AI

Advertising permeates the modern internet. Google Search works because you type in a query, browse results and see ads along the way.

Generative AI makes this model obsolete. AI is about precision. You ask a question, you get an answer. That’s it. There’s no incentive to stick around and browse, which means there’s no room for traditional ads.

That doesn’t mean advertising will vanish. We may see more targeted, high-value placements emerge, but there’s simply not enough user journey to monetize at scale.

Why subscriptions won’t save it either

What about subscriptions? After all, services like Netflix and Spotify thrive on them.

Here’s the problem: generative AI is a vast market with thousands of specialized services. There are AI tools that help with resume writing, AI that takes meeting notes, AI that creates images of people with six fingers. People may use them occasionally, but not enough to justify subscribing.

That’s why I coined what I call Cosmin’s Law: 98% of users will never subscribe. We’re already seeing this play out. OpenAI has surpassed $1 billion in annual revenue, which is remarkable. Yet only about 2% to 4% of ChatGPT users pay $20 per month for premium access. The other 96-98% just generate expenses with every prompt.

The copyright battle

There’s another issue that needs immediate action: copyright. Actors, writers and media companies are already raising their voices. Disney is aggressively protecting its intellectual property from being used by AI. The New York Times has sued OpenAI over alleged copyright infringement. Writers Guild strikes in Hollywood are a prime example of creators’ anxieties about ownership in the age of AI.

This isn’t just about ownership. It’s also about getting fairly compensated for your work. If AI-generated answers rely on copyrighted data, who deserves compensation? Legal uncertainty is yet another block on the path to monetization.

What users really want

So, if traditional monetization models are powerless, what is the way forward for AI?

People crave choice. They don’t want another mandatory subscription or invasive ad model. They want prices that fit into their budgets. They want access the moment they get an idea. Simplicity matters too. People are less likely to engage if there is an eight-step sign-up process. Above all, users want privacy and the assurance that their data is not being mishandled.

If AI providers ignore these criteria, users will walk. If they take them seriously, there is still a chance to succeed.

An iTunes moment for AI

The breakthrough may come from microtransactions. Think back when iTunes broke out. Before them, you had to buy an entire CD even if you wanted to hear just one song. Apple let you get the track you wanted for 99 cents. Affordable, fast, simple. This move paved the way for streaming services and changed the music industry forever.

AI needs its iTunes moment. Instead of locking users into monthly fees or forcing them into watching ads, providers should let them pay for what they use, when they use it. Picture a running tab at a bar: you add drinks or food to your tab, you get them immediately, and only pay when you’re done.

The same could work for AI. Instead of pushing subscriptions and commitments, companies can offer individual items, like a single generated picture or text, or small bundles. It’s simple and accessible for the user and sustainable for providers.

This flips the model: access and value first, payment later. Lower barriers, build trust, and get the 98% to pay for what they use.

Why This Matters

AI monetization is an issue that deserves a lot more attention than it’s been getting. It has the power to make or break the technology. If compute costs stay high, ads underperform, and subscriptions plateau, we will have another “dot-com” moment on our hands.

But if we get monetization right, if we make it affordable, fast, simple and private, there is a chance for a sustainable ecosystem. Not just for the AI giants, but for the thousands of startups building AI tools that can’t rely on venture dollars forever.

The AI revolution is here. The only question is: who is paying for it?