The AI licensing gold rush has produced clear winners and many disappointed onlookers.

News Corp, the Financial Times, and a handful of others have announced deals worth tens of millions with AI companies. Meanwhile, most publishers have either been ignored or offered terms they found insulting.

What distinguishes the publishers who got deals from those who didn’t? And what can mid-size publishers do to improve their position?

I’ve been talking to executives on both sides of these negotiations. Here’s what I’ve learned.

What AI Companies Actually Want

Understanding AI company motivations is essential for negotiating with them.

Training data quality matters. AI companies want high-quality text that reflects clear thinking, good writing, and authoritative information. Publishers with strong editorial standards have more valuable archives than those with commodity content.

Freshness has value. Static training data gets outdated. AI companies value arrangements that provide ongoing access to new content—current news, updated analysis, fresh reporting.

Legal clarity reduces risk. AI companies face uncertain copyright litigation. Licensing agreements provide legal clarity that reduces litigation risk. Publishers who can offer clean rights to their content are more valuable partners.

Reputation provides cover. Partnering with respected publishers creates positive PR and signals responsible AI development. News Corp and The New York Times matter partly because of what their brands represent.

Specialized knowledge commands premium. General news is widely available. Specialized expertise—financial data, scientific reporting, industry-specific analysis—is harder to replicate and more valuable.

What Distinguished Winners

Publishers who secured significant deals shared certain characteristics:

Scale of archive. The deals involve massive content libraries. Publishers with decades of quality archive have something AI companies can’t easily replicate. Small publishers with limited history are less attractive.

Brand authority. Recognized, respected brands command premium pricing. The Financial Times name signals quality in a way that generic business coverage doesn’t.

Negotiating leverage. Some publishers used legal threats effectively—real or implied willingness to sue created negotiating pressure. Others leveraged exclusive content or unique capabilities.

Clear rights ownership. Publishers who could cleanly license their content—without complex freelancer agreements or uncertain ownership—were easier to deal with.

Realistic expectations. Publishers who understood the market and asked for reasonable terms closed deals. Those demanding outrageous sums got nothing.

Willingness to move quickly. AI companies want deals done fast. Publishers with agile decision-making closed while others dithered.

What Excluded Others

Conversely, some publishers were never realistic candidates:

Insufficient scale. AI companies are making a limited number of deals. Content libraries below a certain threshold aren’t worth the transaction costs.

Commodity content. If your content is interchangeable with freely available alternatives, there’s no reason to pay for it.

Complex rights situations. Publishers who don’t clearly own their content can’t cleanly license it. AI companies avoid legal ambiguity.

Unrealistic demands. Some publishers asked for compensation vastly exceeding what their content was worth. AI companies walked away.

Slow decision-making. Publishers that needed months of internal deliberation missed windows when deals were being made.

Options for Mid-Size Publishers

If you’re not News Corp, what can you do?

Collective bargaining. Several efforts are underway to aggregate smaller publishers for joint licensing. The logic: combined archives create enough value to warrant deals that individual publishers couldn’t get.

Specialized content plays. If you have genuinely unique content—specialized reporting, proprietary data, unique archives—that might be valuable even at smaller scale.

Direct integration partnerships. Some publishers are partnering directly with AI companies on specific products rather than broad licensing. This might mean providing real-time data feeds, fact-checking services, or specialized analysis.

Technical differentiation. Publishers who can provide content in AI-friendly formats with good metadata and clean structure make integration easier—and might be more attractive partners.

Blocking strategies. Some publishers are using technical measures to block AI training crawlers, aiming to establish leverage for future negotiations. The effectiveness of this approach is disputed.

The Uncomfortable Math

Here’s the reality most publishers don’t want to hear: AI licensing deals probably won’t save struggling publications.

Even the biggest deals—tens of millions of dollars—represent modest sums relative to publishing operations. For News Corp, the OpenAI deal is significant but not transformative. For smaller publishers, proportional deals would be rounding errors in their budgets.

The hope that AI licensing will become a major revenue stream for most publishers is probably misplaced. Some publishers will benefit meaningfully. Most will get nothing or close to it.

This doesn’t mean licensing isn’t worth pursuing—any revenue helps, and establishing a framework for AI compensation matters for the long term. But it’s not a business model solution.

The Longer View

The current licensing moment is unusual. AI companies are training foundation models and building content relationships. This creates a window where publishers have something AI companies need.

That window may not stay open forever. Once models are trained, the need for ongoing content becomes less clear. The leverage publishers have today may diminish.

Smart publishers are using this moment not just to secure deals but to build relationships that extend beyond one-time payments. Ongoing partnerships, integration into AI products, revenue-sharing arrangements—these might prove more valuable than upfront licensing fees.

The AI-publisher relationship is still evolving. The winners in the long term might look different from the winners today. Stay engaged, stay flexible, and don’t assume today’s dynamics are permanent.

Practical Next Steps

For publishers evaluating their AI licensing position:

Audit your content. What’s unique? What do you clearly own? What’s your archive scale? Honest assessment is essential.

Monitor collective efforts. Several industry groups are organizing joint licensing. Evaluate whether participation makes sense for you.

Develop direct relationships. Even if broad licensing isn’t available, specific partnerships might be. Explore what AI companies might need that you can provide.

Stay informed on litigation. Court cases will shape the legal landscape. Outcomes could change negotiating dynamics substantially.

Don’t wait for a windfall. AI licensing is unlikely to transform most publishers’ economics. Continue building sustainable business models independently.

The AI licensing story is still being written. Position yourself to benefit if opportunities emerge, but don’t bet the business on them.