The Frontier Isn't for Followers

In 2025, OpenAI spent $1.69 to generate every $1 of revenue. The company is projected to lose $14 billion in 2026 alone. By some estimates, more than 14,000 AI startups launched in 2024. Roughly 40% are already gone. Most were wrappers — copilots, assistants, content tools layered on foundation models. Margins collapsed every time API prices shifted.

This is what the front of the line looks like. The best-capitalized players haven't figured out how to make money. The companies trying to follow them are disappearing.

And yet: "If you're not doing AI now, you'll fall behind." The restaurant owner hears it. The factory operator hears it. The law firm hears it. People who were running their businesses just fine start to feel uneasy.

That unease runs on four emotions, not analysis.

Anxiety — "I'm falling behind." It gets confirmed daily in job postings, meeting rooms, LinkedIn. A business that was running fine yesterday feels inadequate after a single AI conversation.

Hope — "If I use AI right, I can leap ahead." Every week brings a new story about a solo founder pulling $1M ARR or a single designer replacing a ten-person team.

Inferiority — "The shop next door already built their own chatbot." Someone else's adoption story starts to read as a report card on your business.

Superiority — "We're already using AI." Whether it goes any deeper than that, or has any measurable effect, goes unchecked. Using it at all is enough to feel ahead.

The four cycle. Anxiety pushes you to act. Hope justifies the action. None of it checks whether the action makes business sense. And while the cycle runs, what disappears from view is the actual asset — the restaurant's customer list, the factory's twenty years of materials expertise, the law firm's case history. All you see is the comparison: you without AI, them with it.

The real reason a domain operator leaves their position isn't analysis. It's emotion.

Take the pressure at face value and here's what happens. Domain operators leave their positions and move capital into territory that hasn't been validated. They build their own chatbots. They try training their own models. They form internal AI teams. The customer data, the know-how, the case files — all of it gets set aside.

A domain operator can't solve, as a side project, problems that frontier companies burning $1.69 to make $1 haven't cracked. Building your own model means signing up for a game where 40% of players disappear in 24 months. A business with a clear domain has no reason to be in that game.

MIT research found that only 5% of corporate AI pilots produced measurable P&L impact. Deployments using validated third-party solutions had a 67% success rate — about 3x the rate of in-house builds. The most expensive position in AI right now is "build it yourself." The cheapest and most effective is "take something validated and attach it to your domain."

A restaurant's real asset isn't AI. It's the customer list, the recipe notebooks, the seasonal sales patterns, the feel for the neighborhood. A factory's asset is two decades on the shop floor, the supplier network, the warranty claim data. None of it can be replicated from outside.

You can layer AI on top of that. Analyze customer data to automate menu suggestions. Put a prediction model on top of materials ordering. Generate first drafts for warranty responses. This isn't adoption. It's augmentation — your domain is the core, AI is the tool you bolt on.

Adoption tries to change your business's identity — a restaurant deciding it's now an AI company. Augmentation keeps it intact and attaches AI to the parts where AI is genuinely good. The cost is different. The risk is different. And most of all, the recoverability is different.

But a domain asset doesn't last forever. And in the AI era, it can collapse without competition.

Not because the restaurant next door used AI to take your customers — but because your customers moved to an AI food recommendation app, and they only go where the app sends them. Not because a competing factory took your client — but because your client moved to an AI-driven procurement platform, and your orders stopped coming. Not because the law firm down the street won your case — but because clients now run their first consultation through an AI tool, and they only contact the firms it recommends.

Competition didn't disappear. The shape of competition changed. And the change is happening outside your domain, not inside it.

So a domain operator needs two things. Protect the asset. And augment it so it doesn't drift out of where the demand is moving.

Get pulled out by emotion and you lose the asset by leaving it behind. Tune out the noise so completely that you miss the shifts happening around you, and you lose the position itself.

Not being at the frontier isn't a weakness. It's an advantage — as long as you know what you're standing on.