AI Has a Supplier Problem

The real AI shortage isn't computing. It's the knowledge ecosystem that feeds the models — and nobody's getting paid for it.

There is a debate raging over AI companies’ use of copyrighted books, articles, videos, music, and websites to train their models. Some argue that everything on the public internet is fair game. Others argue that every creator should be compensated. Lawyers will have a field day.

However, I'm less interested in the legal argument than the business argument.

After decades of working with software companies, I've learned that products rarely fail because of technology. They fail because somebody misunderstands the economics of the ecosystem. The technology works. The business model doesn't.

AI is heading toward a similar problem.

The real AI shortage isn't computing, it’s content

Most discussions about artificial intelligence focus on algorithms, data centers, chips, and model performance. Those things matter. But they are not the entire supply chain.

AI companies often talk as if computing were the scarce resource.

I don't think it is.

The scarce resource is content. Knowledge. Experience. Judgement.

The internet did not appear by magic. Every article was written by someone. Every book required months or years of effort. Every tutorial, video, podcast, research paper, product review, and discussion forum exists because a human person decided to invest time creating it.

Many of those people were not compensated directly for their work. They wrote blog posts to build a reputation. They answered questions to help a community. They published books to establish authority. They shared ideas because they enjoyed teaching. The important point is not that they were paid — it's that they had a reason to create.

I’m a creator. I started writing because customers often asked the same questions, so I wanted to create solid responses to share. I have been blogging since 1999. I have created multiple courses on product management topics. I have two books on Amazon and my website: Turn Ideas Into Products and Customer Conversations. I have recorded dozens of videos for on-demand learning programs.

Today, AI systems are extracting enormous value from the world’s accumulated body of knowledge. In many cases, they are building businesses worth billions of dollars by learning from content that was created for entirely different purposes.

Yes, every innovation builds on previous innovations. In 1675, Isaac Newton wrote, “If I have seen further [than others], it is by standing on the shoulders of giants.” Today’s products are built on the experience of others.

That alone doesn't bother me. What concerns me is something else.

What happens when AI removes the incentive to create?

“Free” contributions have value

I've seen this problem before — not with AI, but with customer communities.

Think of the online communities built around companies and their products. Customers contribute ideas. They answer support questions. They create templates, integrations, and best practices. The community often becomes one of the company’s greatest assets.

Customers were doing work that otherwise would have required employees. Management loves this. Why wouldn't they? After all, the community software isn’t very expensive.

Over time, however, some companies began viewing those contributions as free labor rather than valuable participation. They focused on extracting value from the community while investing very little back into it. The result was predictable, even if few predicted it.

Participation declined. The most knowledgeable contributors disappeared and moved over to Substack and Medium. The quality of discussions dropped. The community became less valuable.

Management was surprised… but they shouldn't have been. The contributors had figured out something basic: they were creating value, but they were no longer receiving enough value in return. So they left.

AI faces a similar challenge, only at a much larger scale.

Creators are suppliers 

The creators of books, articles, videos, tutorials, music, and research are not merely content producers; they are suppliers.

Most product managers understand the importance of suppliers. When a key component becomes unavailable, your product doesn't ship. When a critical vendor exits the market, your whole roadmap changes. You don't have to love your suppliers. But you have to understand that the relationship is mutual.

Imagine telling your manufacturing partner that you plan to use their components indefinitely but have no intention of paying them. Imagine telling a channel partner that you need them to keep selling your product while you systematically reduce their margins.

That relationship doesn't last long.

Yet that is effectively the conversation happening around AI and content creation. The bet many AI companies seem to be making — implicitly, if not explicitly — is that the supply of knowledge is infinite and self-replenishing. It is neither.

The Case for Licensing Over Taxation

The debate often pivots to taxation at this point. Since AI may eliminate some jobs while benefiting from publicly available content, perhaps governments should tax AI providers and redistribute the proceeds.

I don’t think that is going to work.

Taxation solves only half the problem. Collecting money is relatively easy. Allocating it fairly is much harder.

Who deserves compensation? The author of a bestselling book? The professor who published research that four hundred people read? The blogger who wrote a useful article a decade ago? The software developer who answered a question on a technical forum? The photographer whose images were used in training? The musician whose recordings influenced a model?

The challenge is not determining whether value was created. It clearly was. The challenge is determining how much value was created, and by whom, and in what proportion.

I suspect the answer will eventually look less like a tax and more like a licensing model.

Spotify solved a version of this problem for music. Kindle Unlimited solved a version of this problem for books. Neither system is perfect — creators complain about both, often with good reason. Yet both systems acknowledge this: if a platform generates revenue from creative work, some portion of that revenue should flow back to the people who created it. The alternative is eventually running out of creators.

That may sound dramatic. But consider what happens if AI succeeds beyond our expectations.

Suppose junior developers become unnecessary. Suppose junior writers become unnecessary. Suppose junior analysts and designers become unnecessary. Suppose entry-level knowledge work is automated.

Where do future senior experts come from?

Nobody begins their career as a vice president. Nobody starts as a distinguished engineer. Nobody starts as a chief product officer. People develop expertise by doing the work — the early, unglamorous, frequently inefficient work that eventually produces someone who knows what they're doing.

The entry-level jobs that seem easy to automate are often where future experts are created. If we eliminate enough of those roles without building new pathways for people to learn, we are not simply reducing labor costs. We are weakening the system that produces expertise. And the people who benefit most from AI-generated knowledge today will eventually need new human experts to train future models, to interpret AI output, to catch what AI gets wrong.

AI and the supply chain 

From a product management perspective, this is not primarily a workforce issue or an ethics issue. It is a supply chain issue.

Many organizations focus almost entirely on customers. The best organizations understand ecosystems. Customers matter. Partners matter. Suppliers matter. Developers matter. Creators matter. Communities matter.

Whenever one group creates value while another group captures all the value, instability follows. We’ve seen it in channel programs. We’ve seen it in partner ecosystems. We’ve seen it in customer communities. The pattern is always the same: the extracting party believes the relationship is more durable than it is, right up until it isn't.

The same lesson may apply to AI. The companies that ultimately win this market may not be the ones with the largest models or the fastest chips. They may be the companies that build the most sustainable ecosystem around content creation — the ones that understand where their supply comes from and invest in keeping it healthy.

At some point, someone has to keep planting the crops. And somebody has to pay the farmer.