Why public-domain classics may be the first target of the multimodal AI wave — and what could rise in response
They sell the highest: Tolstoy, Premchand, Manto, and the great epics share one trait that makes them uniquely exposed to generative AI: they’re free of copyright. Anyone can already feed the full text of Crime and Punishment or Godan into an AI model without licensing a single word. As multimodal generation matures, the plausible next step isn’t just AI narration or a summary, it’s on-demand, fully produced films, audiobooks, and illustrated editions generated instantly from a reader’s prompt, no studio or narrator required.
That possibility points to a real economic question publishers and authors are already grappling with: if a digital adaptation can be generated for free, what happens to everything currently sold as a digital adaptation? The framing below explores one influential argument, that digital content and physical objects could pull apart into two very different markets, while being upfront about how speculative parts of it are.
The Digital Flood
The logic is straightforward: once an AI system can turn public-domain text into a narrated audiobook, an animated adaptation, or illustrated art on demand, the marginal cost of producing another one approaches zero. Standard audiobooks, generic e-books, stock illustration, and basic video adaptations built on well-known, copyright-free texts are the most exposed, because the underlying content is freely available and the production step is exactly what AI is improving at fastest.
It’s worth being honest about the timeline here: how fast this actually happens, and how good “Hollywood-grade” AI-generated film really gets, is genuinely uncertain. Multimodal generation is advancing quickly, but predicting a specific year for a specific capability threshold is inherently speculative. Treat any date, including 2028, as a plausible scenario rather than a forecast.
The Physical Surge
The flip side of the argument is more durable: scarcity that can’t be digitized tends to hold or gain value precisely because it can’t be copied. A section-sewn book on heavy, textured paper, a limited print run, an author’s signature, none of that is reproducible by a generative model. As digital content becomes abundant and cheap, some publishers are already betting that physical craftsmanship becomes the differentiator, positioning a book less as a delivery mechanism for text and more as a tactile, ownable object.
This isn’t a new pattern… Vinyl records, film photography, and letterpress printing all saw similar revivals once their functional digital replacements became ubiquitous and cheap. Whether books follow the same curve at the same pace is a reasonable bet, not a certainty.
What This Could Mean for Authors and Publishers in an AI dominated handheld devices
It’s worth pausing here on something important: none of what follows is financial advice, and we are not financial advisors. This is a summary of one plausible thesis, not a recommendation to buy, sell, or divest anything. That kind of decision deserves independent research and, ideally, a licensed advisor who knows your actual financial situation.
The thesis itself is intuitive: companies whose value depends entirely on distributing digital text or standard audio narration face real disruption risk if AI generation gets cheap and good enough, particularly for content built on public-domain source material. Businesses anchored in physical production, specialty binding, or high-touch, low-volume craftsmanship face a structurally different risk profile, since their value isn’t just the content but the object and the process. That’s a reasonable framework for thinking about exposure. It isn’t a guarantee about how any specific company or asset class will perform.
Human-to-Human Networks
For authors and small publishers, the more actionable version of this idea isn’t a stock-picking strategy, it’s a positioning one. A hand-bound, premium edition doesn’t have to compete with a free AI-generated audiobook because it isn’t selling the same thing: it’s selling texture, scarcity, and the story of a specific physical object. Author events, signed copies, and direct relationships with readers lean into something AI genuinely can’t replicate. The fact that a human being made this particular object and is standing in front of you.
None of that guarantees outcomes for any individual publisher or investor. But as a framework for thinking about where AI pressure concentrates and where it doesn’t, the digital-versus-physical split is a useful lens, treated as one scenario among several, not a settled prediction.