The Attention Recession

The Attention Recession

Content supply is racing toward infinity while the human attention it competes for has stopped growing. When making things becomes free, the scarce skill is no longer creation. It is distribution.

Two curves describe the media economy of the mid-2020s, and they are moving in opposite directions. The first is the supply of content, which is approaching infinity. The second is the supply of human attention, which has effectively stopped growing. For most of the internet's history these lines rose together — more to watch, and more time to watch it. They have now decoupled. The space between them is the defining economic fact of this decade, and we think it deserves a name: the attention recession.

A recession is not an absence of activity. It is a contraction in the thing that actually matters while the surrounding noise stays loud. By that definition the analogy is exact. Production has never been more frenetic. But the underlying asset — sustained human attention — is shrinking per unit of content even as the aggregate stays flat. Understanding why changes what you should spend your scarce hours building.

The conclusion runs against a decade of received wisdom. The 2010s rewarded those who simply showed up and posted consistently. The late 2020s reward something else entirely. When the cost of making a thing collapses to zero, making it stops being the constraint. Getting it in front of a human being who chooses to care becomes the whole game.

The ceiling: attention is finite, and we have hit it

Start with the wall. EMARKETER projects that US adults' total daily time spent with media will grow by just two minutes in 2025 and a single minute in 2026, with daily digital media time already sitting north of eight hours.¹ Its own framing is blunt: peak media saturation is almost here as top-line growth fizzles out. The audience day cannot get meaningfully longer. There are no more hours to capture.

This is the part most strategy decks quietly ignore. Every business plan that assumes "we'll grow by capturing more attention" is implicitly assuming a market that no longer expands. The total pool is fixed at roughly the waking, screen-facing day. From here, every minute one platform, creator, or brand wins is a minute taken directly from another. Attention has become explicitly zero-sum — a fixed budget, not a growing one.

The flood: creation has been fully commoditized

Now the other curve. While attention flatlines, the machinery of production has gone vertical. Ahrefs analyzed 900,000 newly created English-language web pages first detected in April 2025 and found that 74.2% contained AI-generated content — most a human-AI blend, a small slice entirely machine-made.² Nearly three out of four new pages on the web now carry a machine fingerprint, and on one image platform's accounting people produce an average of 34 million AI images a day.³

Single-platform supply tells the same story at a different scale. Creators upload more than 500 hours of video to YouTube every minute — roughly 720,000 hours a day, against an estimated 800 million videos already accumulated on the platform.⁴ No human being, and no plausible audience, will ever consume a measurable fraction of that. The constraint was never the making. It is, and will only ever be, the watching.

The marginal cost of a competent piece of content has collapsed toward zero. That is genuinely good news for anyone who wants to make something. It is genuinely bad news for anyone who assumed that making something was their edge. Commodities do not command premiums. And content, in the aggregate, is now a commodity.

The saturation point: more machine than human

The two curves meet at an unsettling place. In 2024, for the first time in a decade, automated traffic surpassed human activity, accounting for 51% of all web traffic — with malicious bots alone making up 37%, up from 32% the year before.⁵ Imperva attributes the surge largely to AI and large language models lowering the barrier to entry. The "dead internet" moved from a fringe theory to a line in a security report. The audience for online content is now, literally, more machine than human.

Set that against the eight-hour human ceiling and the picture sharpens. An exploding share of what gets produced is made by machines, distributed through channels increasingly trafficked by machines, and competing for a slice of attention that — on the human side — is not growing at all. Volume has stopped being a signal of anything. It is just weather.

The fragmentation: attention isn't only fixed, it's shattering

If the total were merely fixed, the problem would be hard but legible. It is worse than that, because the fixed pool is breaking into ever-smaller pieces. Reported attention metrics, while imperfect, all point the same direction: average focus on a single social post fell from 12.1 seconds in 2015 to 8.25 seconds in 2025, and average Facebook session length dropped from 2.7 minutes in 2013 to about 54 seconds in 2025.⁶ Attention is not just capped; it is shattering into smaller and smaller fragments.

It is also splintering across surfaces. EMARKETER's mid-2025 breakdown of where the fixed budget actually goes shows US adults spending roughly 2 hours 29 minutes a day on traditional TV in 2025 — still more than any other media activity — with connected TV, subscription streaming, and YouTube together claiming a comparable share of the digital day.⁷ Distribution no longer means publishing into a single void. It means winning a specific placement on a specific surface at a specific moment, over and over, against everyone else doing the same.

The stampede: why distribution became the moat

Here is where capital makes the squeeze acute. Goldman Sachs estimates the creator economy's addressable market could roughly double to $480 billion by 2027, with the number of people who identify as creators reaching 107 million by 2030 — even as professional creators stay around 2.5% of the pool.⁸ More money, more entrants, more uploads, all chasing the same flat, fracturing pool of human attention.

That is the textbook setup for a commodity glut on the supply side and a bidding war on the demand side. When everyone can produce and almost no one can be heard, the scarce, defensible skill is the one that closes the gap between a finished thing and a human who actually receives it. Distribution — the craft of earning finite, fragmenting attention on the surfaces where it lives — has quietly become the moat. Creation is table stakes. Reach is the asset.

The Theodyx Perspective

If creation is free and distribution is scarce, the natural reflex is to optimize distribution itself: more posts, more channels, more algorithmic gaming. That race has a ceiling too, and it is lower than it looks, because the channels are filling with the same machine-made sameness everyone else is pumping out. You cannot out-volume a system that produces 720,000 hours of video a day. You will lose.

The way through is not to compete on the axis that has been commoditized. It is to make the work itself the reason attention gets spent. When the feed is 74% machine-touched and indistinguishable, the rarest signal a human can send is unmistakable authorship — a point of view, a taste, a lived specificity that no model can synthesize because it was never in the training data. That is the one input still genuinely scarce. It is why we build on a single belief: express what only you can. It is the supply-side answer to a demand-side ceiling. Distribution gets your work to the door; only irreplaceable expression gets a human to stay. In an attention recession, the enduring advantage belongs to those who own something a machine can replicate the form of but never the source.

‍

Sources

1. EMARKETER, "US Time Spent With Media 2025," Feb 27, 2025. https://www.emarketer.com/content/us-time-spent-with-media-2025

2. Ahrefs, "74% of New Webpages Include AI Content (Study of 900k Pages)," May 2025. https://ahrefs.com/blog/what-percentage-of-new-content-is-ai-generated/

3. Everypixel Journal, "AI Image Statistics for 2024: How Much Content Was Created by AI," 2024. https://journal.everypixel.com/ai-image-statistics

4. Teleprompter.com, "2025 YouTube Statistics: Global Overview and Key Trends," 2025. https://www.teleprompter.com/blog/2025-youtube-statistics

5. Imperva (Thales), "2025 Imperva Bad Bot Report: How AI is Supercharging the Bot Threat," Apr 15, 2025. https://www.imperva.com/blog/2025-imperva-bad-bot-report-how-ai-is-supercharging-the-bot-threat/

6. SQ Magazine, "Social Media Attention Span Statistics 2026: Viral Content Data," 2026. https://sqmagazine.co.uk/social-media-attention-span-statistics/

7. EMARKETER, "Traditional TV still dominates daily media time," 2025. https://www.emarketer.com/content/traditional-tv-still-dominates-daily-media-time

8. Goldman Sachs, "The creator economy could approach half-a-trillion dollars by 2027," 2024. https://www.goldmansachs.com/insights/articles/the-creator-economy-could-approach-half-a-trillion-dollars-by-2027