Most AI companies are renting their position. These 4 prompts tell you if yours is one of them.

On February 25, Perplexity shipped the best agentic product of the month. And it might not matter.

Perplexity Computer launched to genuine excitement. It’s a cloud-native, multi-model orchestration system that routes work across 19 frontier models, spawns sub-agents, persists for months, and delivers finished artifacts while you sleep. It runs Claude Opus 4.6 as its reasoning core, Gemini for deep research, Grok for speed, GPT-5.2 for long-context recall. It is available today for $200 a month and it is almost certainly worth it for heavy research and ops workflows.

It is also, structurally, a cautionary tale about where most of the AI industry is building right now, and why good execution on the wrong layer of the stack doesn’t save you.

I should say this upfront: this is not a Perplexity hit piece. Perplexity is one of the best-run AI companies in the world. They read the market correctly. They made a genuinely bold call to kill their own ad business in February because they understood that trust is the new distribution. They are targeting $656 million in 2026 revenue, and their search API already has major enterprise customers running it in production.

And yet. Their core reasoning engine runs on a direct competitor’s model. Their deep research runs on another competitor’s model. Even their speed layer depends on a third provider who is building the same product Computer competes with. The week Perplexity launched Computer, Anthropic shipped the enterprise expansion of Claude Cowork with deep connectors, private plugin marketplaces, and the ability to pass context seamlessly across tools. Cowork doesn’t need 19 models. It has one. And it owns it.

That asymmetry between the quality of Perplexity’s execution and the fragility of its structural position is the thing I can’t stop thinking about. Because it’s not just Perplexity’s problem. It’s the position of almost every AI company that isn’t Anthropic, Google, OpenAI, or Meta. Which means it’s probably the position of the company you work for, invest in, or advise.

Here’s what’s inside:

• Perplexity Computer, honestly reviewed. Who should pay $200/month, who should skip, and the five use cases where it actually over-delivers.

• The middleware trap, and the math behind it. Why good execution on the wrong layer of the stack won’t save you, and how the hyperscalers’ $690 billion infrastructure bet makes it worse by the month.

• Four positions that survive. The specific structural plays where the hyperscalers’ incentives align with your existence rather than your replacement.

• The diagnostic. A five-step test you can run today on your own company, your portfolio company, or your client’s company to know if you’re building a durable position or renting it.

We’ll start with how February 2026 drew the map for everything that follows.