Cultural compounding and the age of immortal IP
David Peterson

The release of Sora 2, and in particular its new “cameo” feature, which enables the licensed regeneration of a person’s likeness, marks the beginning of a new economic era for culture: the age of immortal IP.

This is a technical capability newly enabled by generative AI, yes, but I think this is much bigger than that. It’s a shift in the time horizon of culture itself. Stories will no longer merely endure. They will compound. Likenesses, voices, and movements will be endlessly regenerated and remixed into new works.

To understand the magnitude of this shift, it helps to remember how slowly culture once moved.

1. When stories lingered

The summer of 1975 belonged to Jaws. The film opened in June and stayed at number one at the U.S. box office for fourteen consecutive weeks, stretching through Labor Day and beyond. That kind of dominance is almost unimaginable today. A film could once define a season; a song could hold the airwaves for months. Now the half-life of attention is measured in hours. A meme peaks in days. A hit Netflix series fades in a week.

And yet even as everything accelerates, a few works seem to last longer than ever. The Beatles remain among Spotify’s most-streamed artists. The Office remains culturally ubiquitous 20 years after its premiere.

Some stories vanish instantly while others seem to have escaped decay altogether.

2. From distribution to discovery

For most of modern history, the bottleneck in culture was distribution. The scarce resources were physical: printing presses, theaters, radio stations, record shelves. Whoever controlled those channels controlled what reached an audience.

The internet erased that constraint. Anyone could publish, record, and broadcast. Distribution became free, attention became scarce.

Today, the bottleneck is discovery. Recommendation systems determine what we see, and in doing so, they decide what survives. Algorithms hunt for the next great meme, but also reward what already performs. They transform past attention into future attention, reinforcing the familiar and letting the unfamiliar fade.

This creates a new kind of feedback loop in which attention behaves like capital: it earns interest. Most works disappear almost immediately, while a small number compound indefinitely. 

Platforms like Spotify and YouTube are not archives in the traditional sense; they are memory systems that learn which memories to keep. Every listen or view increases the odds of being heard or seen again, creating an engine of cultural compounding.

3. The price of conviction

Venture capital has already lived through this shift.

In the early 2010s, many investors understood the power law intellectually but failed to act on it. They passed on companies that would define the decade because their valuations seemed too high. Meanwhile, firms like a16z were mocked for overpaying. But they realized something others missed: in a world of internet scale, the upside is practically uncapped, and the biggest mistake is one of omission, not commission. Paying $300 million instead of $200 million for a company that will eventually be worth $50 billion is a rounding error.

All of this makes me wonder: are today’s media incumbents making the same mistake? Are studios, streamers, and rights holders underestimating how large a single piece of IP can become once discovery, nostalgia, and generative tools begin compounding together? Are they still haggling over licensing fees in a world where value may compound indefinitely?

On the one hand, the power law of culture is already apparent. Disney is so protective of their IP for a reason. Star Wars wasn’t just a movie, it created a cultural movement that has been monetized over the course of generations. Pokémon hasn’t just stayed relevant, the revenue associated with the franchise has continued to increase with time.

On the other hand, if cultural compounding holds, the distribution of outcomes in culture may soon resemble that of software in the 2010s. There may be more winners, but the winners will be bigger than ever before, and everything else will round to zero. In that world, the rational strategy is not caution but conviction.

4. The business of immortal IP

This shift demands new business models and new ways of thinking about value. The owners of compounding IP may be sitting on massive generators of wealth, and the data around that IP may soon be even more valuable than the IP itself.

What if platforms that own the data exhaust of cultural engagement won’t just be able to predict demand, but shape it?

This is already happening. Netflix doesn’t just host old shows. It reanimates them. In 2023, Suits logged 57.7 billion viewing minutes, making it the most streamed acquired show in the United States that year and more popular on Netflix than it ever was on cable. That resurgence came years after its finale, driven by binge culture and recommendation loops that turned a mid-tier legal drama into a global hit.

Just as software monopolies consolidated around network effects, cultural monopolies are now consolidating around discovery loops. The more an audience interacts with a piece of content, the more likely it is to be rediscovered. Attention compounds like interest, creating algorithmic flywheels that favor the familiar.

This suggests two things. 

First, Meta (via Instagram), Bytedance (via TikTok), Google (via YouTube), and Netflix may be even more valuable than any of us currently imagine. 

And second, entire industries will undoubtedly be built around the management and regeneration of cultural assets. Will we see the rise of firms that specialize in revitalizing dormant franchises (perhaps by leveraging audience response data, or just making bets based on their personal taste)? What about financial instruments that treat cultural IP as infrastructure, i.e. assets with steady, predictable, algorithmically sustained cash flows?

And so we return to Sora 2. A model that can regenerate a face or a voice can also remake the economics of culture, because a century from now, new films may still feature the same stars we know today, their likenesses licensed, their voices reanimated, their performances extended into infinity.

Culture has entered the era of perpetual yield.

The AI-Native Enterprise Playbook
Ten real-time observations on a rapidly evolving playing field

No More Painting by Numbers
It’s the end of the “SaaS playbook.

The Age of Artisanal Software May Finally be Over
Every wave of technological innovation has been catalyzed by the cost of something expensive trending to zero. Now that’s happening to software.

Founders as Experiment Designers
David on why founders should run everything as an experiment.

ENTERPRISE/TECH NEWS

Small Language model are going to be big. Michael Parekh covers the increasing attention being paid to SLMs as they pose an ever more credible (and superior) alternative to LLMs in a growing number of use cases. ““The trend of AI researchers developing new, small open source generative models that outperform far larger, proprietary peers continued this week with yet another staggering advancement.” “Alexia Jolicoeur-Martineau, Senior AI Researcher at Samsung’s Advanced​ Institute of Technology (SAIT) in Montreal, Canada,​ has introduced the Tiny Recursion Model (TRM) — a neural network so small it contains just 7 million parameters (internal model settings), yet it competes with or surpasses cutting-edge language models 10,000 times larger in terms of their parameter count, including OpenAI’s o3-mini and Google’s Gemini 2.5 Pro, on some of the toughest reasoning benchmarks in AI research.””

The reinforcement gap. Russell Brandom writes that RL (or lack thereof) will shape which sectors see the most rapid benefit from AI. “As the industry relies increasingly on reinforcement learning to improve products, we’re seeing a real difference between capabilities that can be automatically graded and the ones that can’t. RL-friendly skills like bug-fixing and competitive math are getting better fast, while skills like writing make only incremental progress. In short, there’s a reinforcement gap — and it’s becoming one of the most important factors for what AI systems can and can’t do.”

The Cursor Resistance. Some engineers appear to be pushing back. “The technology has indeed become an inescapable part of Silicon Valley, but the rush to adopt it has brought a backlash among programmers. Partly that’s because the AI coding tools have some obvious technical limitations—sometimes producing error-ridden code, among other problems—and partly it’s because human coders worry any sort of adoption of the tools will hasten their own obsolescence. These frictions foreshadow the kinds of obstacles that will make it more difficult for AI to automate other kinds of knowledge work in the future.”

HOW TO STARTUP

The VC inversion. Roman Axelrod, CEO of XPANCEO, writes bravely in Fortune that raising $100 million is easier than raising $1 million, which he believes is a “tragedy for early-stage companies.” “The truth is that the most valuable companies of the 2030s won’t come from inside the system — they’ll come from the outsiders who endured long enough to earn belief. It is belief, after all, that has always driven venture capital. Every dollar raised is an act of faith in a version of the future. What has changed is the burden of proof. The first dollar is the hardest, and every subsequent milestone is a test of whether the future you describe is one the world is prepared to trust.”

HOW TO VENTURE

In defense of OG Venture Capital. Leslie Feinzaig does it again with an outstanding analysis of the current moment in Venture Capital. “When everyone chases the same thing, the consequences are predictable. Rounds move fast. FOMO replaces diligence. Metrics (real and made up) get stretched. The pressure to get in, to get in fast, to get in at all costs, is applied up and down the venture capital food chain. From the LPs who have to report to their investment committee, to the GPs who report to the LPs, to the VC partners, principals and associates who report to the GPs, to the founders who report to their own VCs and teams. We have all seen this movie before. We know how it ends. How quickly we forget the lessons.”

Firebolt joins the Connect with Confluent partner program.

Reco named a SINET16 innovator award winner.

Dualbird
HW Design Engineer (Tel Aviv, Hybrid)

FalkorDB
Core Developer (Tel Aviv, Hybrid)

Sourcix
Lead Product Manager (Tel Aviv)