Building Gets Easy. Selling Gets Brutal.

In 2012, any decent YouTube channel could monetize. Post consistently, grow an audience, collect AdSense. The formula worked. By 2024, the middle class of creators had been mostly squeezed out. Platforms captured the value, algorithms picked the winners, and everyone else competed on volume for shrinking per-view rates.

Most B2B builders watched this happen and thought: that’s a content problem. A creator economy problem. Not my problem.

It was never a content problem. It was a supply, distribution, and platform problem. And B2B builders are next — with less warning and faster timelines.

The Bilateral Squeeze

AI is collapsing the cost of building digital products. One person ships what used to take five. Time-to-MVP has gone from months to weeks. The barrier to entry for software, content, design, data products, and AI agents is approaching zero.

This feels like pure upside, and for a narrow window it is. Early movers in any market with cheap production capture the gains before competition arrives.

But Jevons tells us what happens next: cheap production means more production, not the same production more cheaply. The supply of digital goods is going to multiply several fold — probably an order of magnitude within five years. Apps, tools, templates, agents, APIs — everything.

More supply in a stable market means price compression. And here’s the structural problem: AI isn’t just compressing the cost of building. It’s compressing the cost of evaluating. A B2B buyer in 2024 could evaluate three or four vendors in a buying cycle. A B2B buyer in 2026 with AI assistance can evaluate fifteen. More options evaluated means more price pressure. Faster commoditization.

Supply multiplies. Buyer evaluation cost drops. Both ends of the margin compress simultaneously. That’s the bilateral squeeze — and it’s faster and more structural than anything the creator economy experienced, because creators couldn’t make their audiences better at consuming content faster.

The Pattern Has a Name

Every software platform cycle produces the same outcome. The standalone product becomes a free feature inside something bigger.

The timeline was 3-5 years. With AI, a platform with distribution can now build the equivalent feature in weeks. The window between “you found something that works” and “a larger player ships it as a free feature” is compressing from 3-5 years to 12-18 months.

The B2B AI tool lifecycle is already visible:

Founder identifies real pain point
    ↓
Builds AI-enabled point solution
    ↓
Gets early traction (the "hair on fire" buyers)
    ↓
Raises a round, scales GTM
    ↓
Salesforce / HubSpot / Microsoft / Google notices
    ↓
Ships native version inside existing product
    ↓
Buyer asks: "why am I paying $200/seat for this
             when my CRM does it for free?"
    ↓
Point solution gets acquired or dies

This isn’t speculation. It’s a named, documented pattern running at accelerated speed.

Diapers.com Had a Great Product

Diapers.com built a genuinely good business — fast delivery, loyal customers, real margins on baby products. Amazon noticed. Amazon dropped diaper prices below cost. Diapers.com couldn’t sustain the war. Amazon acquired them in 2010 for $545M, then shut them down in 2017 once the customer relationships were absorbed.

The lesson most people take: don’t compete with Amazon.

The more useful lesson: distribution wins when the product is replicable. Diapers.com had a better product and worse distribution. That was always going to resolve one way.

There is a counterargument worth taking seriously. Platforms only absorb things that reach a certain scale threshold. The “too small for the platform to care about” zone is actually getting larger as AI makes niche products viable with smaller teams. A tool serving 500 obsessive users in a specific industry vertical, priced accordingly, may never be worth absorbing. Going narrow and weird enough is a genuine escape hatch.

But for anything that reaches real traction in a large market? The playbook runs. And it runs faster now.

The Three Moats AI Is Eroding

B2B software companies spent the last twenty years building moats from three sources. AI is attacking all three simultaneously.

Data lock-in — your data lives here, migrating is painful. AI is making data portability and migration tools easier and cheaper to build. The moat is shallower than it was.

Workflow integration — we’re embedded in how your team works. A well-resourced platform with AI can now replicate a workflow integration in months that used to take years.

Network effects — your clients and partners are also on this platform. This still holds, but it requires genuine community, not just shared software. The “everyone uses it” moat from sheer inertia is eroding as switching costs drop.

What’s left: distribution, trust, and relationships.

Which is exactly what the creator economy survivors figured out. The creators who made it weren’t the ones who made the best content. They were the ones who owned the relationship with their audience directly — email lists, communities, Patreon — rather than renting it from the platform.

The B2B equivalent: the companies that survive aren’t the ones with the best AI features. They’re the ones whose buyers would miss them. Not the product. The team, the relationship, the domain expertise embedded in the service layer. That’s much harder to clone and much harder to absorb.

Distribution First, Build Second

This is the key flip that AI is forcing on every PM and builder.

The instinct, for most people building software, has been: build first, figure out distribution once the product exists. Ship the thing. Then find the people.

That instinct is now backwards.

The question has stopped being “can I build this?” — yes, faster than ever — and become “do I have distribution for this?” A product you build in a weekend that ships to zero people is worth less than a product that takes a month and goes to 10,000 people who trust you.

Before AI, building was hard enough that you could earn distribution time — by the time you shipped, the market hadn’t fully filled. Now the market fills before most products are finished. The distribution gap closes faster than the build gap. You can’t out-ship the supply explosion. You can only pre-empt it with audience.

The builders who win aren’t the ones who build fastest. They’re the ones who:

• Build for an audience they already own — the PM community, the ops team, the specific vertical where they have credibility
• Solve problems where relationships matter more than features — anything involving trust, context, or ongoing judgment
• Find the spaces too small or too weird for platform players to care about — the niche escape hatch
• Treat distribution as the product, not an afterthought — the list, the community, the brand, built before and alongside the thing

The Donut Machine, Again

This is the third piece in a sequence. The first — The Productivity Paradox — argued that technology grows efficiency, but platforms capture the gains. The second — The Speed Drug — argued that speed restructures thinking, and fast access stratifies who thinks well.

This one closes the loop at the market level. The same underlying dynamic operating at three scales: Jevons plus platform power equals gains concentrated at the top.

The donut machine metaphor from the Speed Drug lands here too, but differently. If a machine produces donuts at 100x the rate, it doesn’t make any one baker richer — it makes donuts cheap and abundant and it makes the question of who owns the corner where people are hungry the only question that matters.

Everyone will have a donut machine. The people who saw this coming built the corner first.

The builders win the sprint. The sellers lose the marathon. And the platforms win everything — unless you’ve already built something the platforms can’t easily absorb: an audience that trusts you, a relationship that matters, a corner that was yours before the machines arrived.

Data references: William Jevons, “The Coal Question” (1865); Barry Schwartz, “The Paradox of Choice” (2004); Amazon/Quidsi acquisition (2010), shutdown (2017); YouTube Partner Program earnings data (2012-2024); Barry Learmonth, B2B buying cycle research (2023); a16z, “The Future of AI Products” (2024).