The Render Economy
Every generation had a medium that was worth more than its signal. Radio was more valuable than sound waves. Television was more valuable than photons hitting a cathode ray tube. Websites were more valuable than the pages they served. Apps were more valuable than the pixels they painted. Each time, people initially priced the signal — the song, the show, the article, the screen — and eventually realized the medium was the asset. The signal was a sample. A lure. A reason to keep the medium turned on.
AI is the next turn of this crank. But this time the medium doesn't just deliver content. It regenerates it on demand. And that is not another distribution shift. It's a phase change in what "content" economically is.
The essay you are reading could be regenerated tomorrow — tuned for a different audience, updated with this week's research, translated into Mandarin, compressed to a tweet thread, expanded to a whitepaper. The version hitting your retina right now is one frame from an unbounded film. It matters in this moment of contact. It does not accumulate as inventory. It is not the asset.
The asset is the system that made it: model selection, context architecture, memory, taste, domain knowledge, trust relationships, workflow. The things that don't change when the output changes. The things that make the next output possible.
I've been circling this idea from different angles for months. Agentic Capital argued that the scarce component in an AI agent is the human kernel — judgment, taste, decision frameworks — not the model or the compute. The agent runtime argued that infrastructure, not individual agents, is the durable layer. Pocket Universes argued that context compaction — compressing experience into something useful for reasoning — is the missing product category. All of those were fragments of this: we are entering a render economy, and nobody has named the object yet.
Drake is half right, which is his most dangerous configuration. He's right that treating outputs as disposable is a competitive instinct some people already have. He's wrong that the instinct is sufficient without the framework. Knowing that you should throw away your outputs is not the same as understanding why the economy is reorganizing around the thing that produces them. The framework is what lets you predict where value migrates next. The instinct is what lets you feel smug about the present.
Here's where the framework earns its keep. In 2025, economists Silbert and Galdin published "Making Talk Cheap", studying what happened to Freelancer.com after LLMs arrived. Before LLMs, employers paid a premium for customized job applications — the customization was a costly signal that the applicant had invested effort, which proxied for quality. After LLMs made polished writing free, the premium vanished. Top-quintile workers were hired 19% less often. Bottom-quintile workers were hired 14% more often. The signal collapsed because the artifact stopped being expensive to produce.
This is not a story about freelancers. It's a story about every market where the artifact functioned as a signal. Blog posts signaled expertise. Design portfolios signaled taste. Marketing copy signaled brand investment. Code samples signaled engineering skill. Research papers signaled intellectual rigor. In every case, the costliness of production was load-bearing — it made the artifact hard to fake, which made it informative. When production cost drops toward zero, the artifact stops being informative. It becomes noise shaped like signal.
Ann is pointing at the gap between artifact quality and system quality, which is the core tension of the render economy. When anyone can produce a polished artifact, polish stops being diagnostic. The thing that remains diagnostic is consistency across contexts — can your system produce quality reliably when the conditions change? A single great essay is noise. A system that produces the right essay for any audience, topic, and constraint is signal. The artifact is the proof-of-work; the system is the work.
Therefore the economic unit shifts. Content stops being inventory and becomes a function call — a momentary, context-specific materialization of a deeper capability stack. The word I keep reaching for is render. Not because the content is fake or hollow, but because it has the ontological status of a render: real in the moment of use, not designed to persist, regenerable from source. A render from a 3D engine matters — it's the thing the player actually sees — but nobody confuses the render with the scene graph. Nobody hoards individual frames. The value is in the scene graph and the engine that can keep producing frames.
But here's the turn I didn't expect when I started writing this. The render economy doesn't just devalue content going forward. It retroactively reveals that most content was already a render — we just couldn't see it.
A blog post was always a lossy compression of the author's thinking. A brand's marketing was always a momentary expression of its positioning. A resume was always a render of capability. A product demo was always a render of an engineering team's judgment. We treated these artifacts as primary because regeneration was expensive. If it cost a week to write the essay, the essay felt like the thing. If it cost a quarter-million to shoot the commercial, the commercial felt like the product. The cost created an illusion of permanence and primacy that was really just a side effect of the production bottleneck.
AI made regeneration cheap, and now the render is visible as a render. The thing we called "content" was never the fundamental asset. It was always an expression of something upstream — the system of taste, knowledge, context, and judgment that produced it. We just couldn't see through the artifact to the system underneath, because the system was a human and humans are opaque. The render economy doesn't create a new condition. It makes an old condition legible.
For creators, the shift is disorienting but clarifying. Your value is not your archive. It's your ability to regenerate. The writer who produced one great essay is less valuable than the writer whose system — taste, knowledge, research instincts, domain context, editorial judgment — can produce the right essay for any situation. The photographer whose eye and relationships and process can produce the right image for any brief is more valuable than a photographer with a stunning portfolio and no idea how they got there. Invest in the system, not the catalog. The catalog is a demo reel for the system. It was always a demo reel. You just didn't know the system was the product.
For companies, content strategy quietly becomes system strategy. The competitive moat is not "we have ten thousand blog posts indexed by Google." It's "we have a generative stack that produces the right content for any customer, context, and moment — and the trust graph to make people believe it." The Creation as a Service framing is already emerging in marketing: companies pricing for continuous generative capability rather than individual deliverables. That framing is directionally correct. The deliverable is the render. The capability is the product.
For economics, the costly-signal literature predicts exactly what comes next. When artifacts stop being costly, markets don't just collapse — they find new signals. Research on content platforms already shows the pattern: GenAI increases aggregate content quality while concentrating value among fewer producers and driving high-quality creators off platforms that give the tools away free. The signal migrates from the artifact to the source. The new costly signal is regenerative consistency — can your system produce quality reliably across contexts, over time, under pressure? That is harder to fake than a single artifact. Trust migrates from the object to the origin.
Citrine isn't being folksy. She's describing the microeconomics precisely. The egg is the render. The flock is the system. The ranch is the context. And the farmer's judgment about which chickens to keep, which feed to buy, and when to let a field rest — that's the kernel from Agentic Capital, expressed in a domain that predates computers by ten thousand years.
Paul Kedrosky argues that LLMs are becoming "the new silicon" — the commodity substrate — while orchestration layers become the new operating systems. Doug Shapiro argues that most content will cease being a profit center and become top-of-funnel for scarce complements. Both are describing the same migration from different vantage points. The model commoditizes. The content commoditizes. What doesn't commoditize is the system that knows which model to call, with what context, for what purpose, and has earned enough trust to be believed when it delivers the result.
I want to be precise about what I'm not saying. I'm not saying the artifact has zero value. A great novel, a breakthrough paper, a film that rewires your perception — these are not disposable renders. They are system outputs so extraordinary that they transcend their origin and become cultural objects in their own right. But those are the exceptions that prove the frame. Most content — the vast, churning ocean of blog posts, marketing pages, reports, emails, presentations, social media updates, product descriptions, documentation — was never that. It was always functional. Always contextual. Always a render. The render economy doesn't demote masterpieces. It demotes everything else to the status it always secretly held.
We spent the last century building a content economy — libraries, studios, publishers, platforms, feeds, archives, indexes, paywalls. We built institutions to produce, distribute, and protect artifacts because the artifacts were expensive to make and therefore worth guarding. AI is not destroying that economy. It's revealing that "content" was always an approximation of something more fundamental: the ability to generate the right thing at the right time for the right person. The content economy was a compression artifact of an era when generation was expensive. Now that it's cheap, the compression is decompressing, and the thing underneath is becoming visible.
The render economy is not post-content. It is pre-content, finally legible.