Everyone in AI is building the wrong thing for the same reason
Every AI founder I talk to is on an accelerating treadmill, burdened by a nagging suspicion that the entire industry is moving too fast in a direction that doesn't quite make sense, with no idea about how to get off.
There is an overwhelming feeling that if everyone stopped and thought about it for five minutes, they'd build something completely different; but it's paired with a total inability to articulate why they specifically haven't stopped.
Well, the answer is simple.
Of course it is.
Nobody stops, because nobody can afford to, and that feeling has a name. It's called Moloch, and it's the most important thing happening in AI that isn't on your roadmap.
A race nobody chose
A new model drops. It's better at benchmarks. Every AI product company now has to integrate it (if they're a wrapper) or match it (if they're building a model) or explain to their investors why they didn't. Integration // parity takes engineering time, time that doesn't go toward the actual product differentiation that would create long-term value. But the company that doesn't integrate looks like it's falling behind, and looking like you're falling behind is fatal when your entire valuation rests on the assumption that you're on the frontier.
Everyone catches up, everyone shifts resources, everyone ships the update. The net competitive position of every company is exactly where it was before, except they all engineering time on esoteria and none of them made their end-product meaningfully better for end-users.
That's one cycle.
It happens every few weeks now.
This is Moloch; and it'scoordination failure as competition. Every company makes an individually rational decision (keep up with the frontier or die) and the aggregate outcome is an industry that's moving incredibly fast in the direction of model capability, and barely at all in the direction of product value.
The race is real // the destination is fake.
Convergence
Every AI product is converging on the same interface: a chat box. Maybe a chat box with some tool use. Maybe a chat box with some RAG. But at bottom, a text input and a text output, differentiated by branding and minor UX choices and which model is running underneath.
The chat box won because it ships fast, and Moloch rewards speed over design.
Building a novel interaction paradigm takes months or years of research, iteration, user testing. During those months, fifteen competitors will ship chat-box wrappers with the latest model and capture the market's attention. Your investors will get nervous. Your board will suggest you "ship something" while you figure out the long-term vision. You'll ship the chat box. The long-term vision will die in a Notion doc nobody opens.
Every company faces this exact choice and picks the same outcome. The competitive structure selects against creativity. The companies that take time to build something new get outrun by the companies that ship the obvious thing fast. And the obvious thing is always the same thing, because the obvious thing is, by definition, the thing that doesn't need original thinking.
Moloch doesn't want you to build something good.
Moloch wants you to build something now.
Fundraising as trap
The competitive dynamic is bad enough at the product level. At the fundraising level, it's catastrophic.
AI companies are raising at valuations that require them to grow at rates that are only achievable by chasing the broadest possible market with the most generic possible product. A company that raises at a $500M valuation needs to show a path to billions in revenue, which means it can't afford to be a niche tool that does one thing brilliantly for a specific audience. It has to be a platform, horizontal, aimed at enterprise, built for no one in particular.
Every AI company is building the same enterprise platform with the same features targeting the same buyers, because the fundraising math requires it, because the valuation requires it, because the competitive environment requires raising at that valuation to attract talent, because the talent market requires it.
Nobody in this chain chose this outcome. Every individual decision was rational. The aggregate result is an industry producing fifty identical products, none of which solve any specific problem particularly well, all of which are locked in a feature war that generates zero cumulative value.
The companies that would build something different (smaller, more focused, more opinionated) can't raise the capital to compete. The companies that raise the capital are structurally forced into sameness. Moloch eats the variance, and variance is where all the actual value lives.
Talent death spiral
It's in the talent market too.
Top AI researchers and engineers go to the companies that are on the frontier. The frontier is defined by model capability benchmarks. So companies invest disproportionately in model capabilty, because that's how you attract talent, even when the marginal return on model capability for their specific product is near zero.
Your users don't need GPT-5 when they can barely use GPT-4 effectively. They need better UX and better integration into how they actually work. But the engineer who could build that wants to work on the model, because model work is what gets you your next job, because the next company is also optimizing for model capability, because they need to attract talent too.
Every company overspends on model capability and underspends on product quality. The structure forces it. The talent market punishes you for having the right priorities.
Why "build something good" isn't a strategy
The startup advice I've heard for a decade+ is "Ignore the noise. Focus on users. Build something good."
Correct in theory, nearly impossible in practice, because Moloch specifically punishes this strategy in the medium term.
Building something good takes time, and time means you're not shipping. Not shipping means you look like you're losing, which means talent leaves and investors get nervous, and your competitors (who are shipping fast and bad) capture market position that's expensive to reclaim.
The window in which "build something good" pays off is long, two years, three years, maybe more. The window in which Moloch punishes you for it is immediate and continuous. Most companies don't survive the medium term to reach teh long term.
The founders who know this stay up at night.
They can probably see the right thing to build.
They can't survive building it.
Escape is im//possible
You can't outrun Moloch. You can only build a structure it can't reach.
In practice, this means: don't raise more than you need. Don't hire faster than your product vision can absorb. Don't compete on the frontier unless the frontier is where your users live. Find a niche that's too small for the Moloch-captured companies to notice and go so deep into it that by the time they arrive, your product is years ahead in the dimension that matters to those users.
Be small enough that the competitve pressure doesn't distort your priorities. Be opinionated enough that you can't be dragged into building the same thing as everyone else. Be capitally efficient enough that you don't need the growth rate that forces you onto the treadmill. Be weird enough that nobody can copy you without becoming you.