Y Combinator and Andreessen Horowitz have just published their "what to build" theses for 2026.

These documents are roadmaps for founders: where to put your energy, which markets are about to emerge, what ruptures are quietly forming. By the time they're published, it's usually too late to be first into any specific space they name. But they're still the most useful single signal you can read about where the next two years are heading.

I read both lists carefully, and then I read them again next to my own portfolio and the founders I'm meeting weekly. Two things stand out.

The central message is clear

We're no longer just building AI tools. We're rebuilding entire industries as AI-native.

That's the thesis underneath both lists. It's not "add a copilot to an existing job." It's: rethink the organisation itself.

The disruption is technological, of course. But the deeper rupture is organisational. The org chart of a 2026 AI-native company looks fundamentally different from the org chart of a 2022 SaaS company at the same revenue. That's the thing both YC and a16z are telling founders to internalise.

What's in the lists

For reference, the a16z 2026 build categories are:

1. Infrastructure

  • Multimodal data: extract structure from documents, images, and videos for the enterprise
  • Agent-native infrastructure: support massive "agent-speed" workloads
  • Multimodal creative tools
  • AI-native data stack: how AI agents navigate context problems + the evolution of BI tools
  • Interactive video

2. Consumer

  • AI-native university: professors as architects of learning
  • Healthy MAUs: consumers who aren't sick but want to monitor their health
  • ChatGPT becomes the AI app store
  • World models in storytelling: full 3D environments from text prompts

3. Crypto

  • Privacy creates network effects
  • Prediction markets scale
  • "Know your agent": cryptographically signed credentials for agents to transact
  • Staked media: skin in the game, with proof

4. Enterprise + Fintech

  • Systems of record lose primacy: the interface becomes a dynamic agent layer
  • Vertical AI goes multiplayer: collaboration becomes the moat
  • Voice agents handle entire workflows and customer relationship cycles
  • AI-native banking infrastructure

5. American Dynamism

  • AI-native industrial base: simulation, automated design, day-one operations
  • Factory mindset: modular AI deployment with skilled workers, assembly-line style
  • Physical observability: real-time understanding of cities, power grids
  • Autonomous scientific labs: hypothesis to experiment design, closed loop

Read this carefully. The recurring word is native. Not augmented. Not enabled. Native.

How I read it as an investor

Three things to take from this convergence — and one big risk.

The convergence itself is the signal

YC and a16z don't usually align this tightly. When they do, it's because the underlying movement is too large to be branded as one fund's thesis. We saw the same convergence around mobile in 2010, around SaaS in 2014, around fintech in 2018. Each of those was followed by a five-to-seven year category-defining wave.

Treat the lists with healthy scepticism — VC theses are also marketing — but when this much capital agrees on the shape of the next era, the prudent default is to assume they're directionally right.

The locus of value moves from "tool" to "outcome"

Look at the recurring pattern across both lists: agencies that sell results, not software. Voice agents that own the customer relationship, not just the call. Vertical AI where collaboration is the moat. The implication for founders is heavy: the buyer no longer wants a tool to do the job. The buyer wants the job done.

That changes pricing models (outcome-based, not seat-based). It changes go-to-market (you sell to the budget owner, not the IT department). It changes what a "product" even is. Founders who keep selling tools into a world that's switching to outcomes will compete on a margin profile that won't sustain a venture-scale business.

The ChatGPT app-store thesis is the quiet bombshell

Of all the bullets on a16z's list, "ChatGPT becomes the AI app store" is the one I'd weigh most heavily. If even half of that thesis plays out, it reshapes consumer distribution the way the iPhone did in 2008. The founders who position early will compound advantage faster than anyone else in the cycle.

The bigger thing nobody is saying clearly enough

These new players will be built with far fewer employees. When an agent does the work of X people, the cost structure changes. Speed changes. Barriers to entry change.

I wrote a separate piece about three people doing the work of twenty, and the YC and a16z lists are entirely consistent with what I've been observing on the ground.

The implication for venture is significant: the unit economics of an AI-native company at €10M ARR look unlike anything we modelled in the last cycle. Smaller teams, higher margins, faster iteration, lower CAC if the distribution is right. It's not just a new wave of companies — it's a different kind of company at every stage of growth.

The implication for society is bigger and not for me to solve in this article: productivity step-up, employment dislocation, retraining at scale, value redistribution. These are real consequences. They deserve to be discussed honestly, not waved away.

What I'm actually doing with this

Two practical positions at Honey Lab heading into 2026.

Backing AI-native operators, not AI-enabled retrofits. The founders we'll back over the next year are the ones who can articulate, clearly and specifically, what their company is structurally able to do that an AI-enabled incumbent cannot. If the answer is hand-wavy, it's a no.

Co-investing across geographies. The shifts described in these lists won't happen only in San Francisco. The European and LatAm operators who internalise this thesis early will compound just as fast. That's where Honey Lab's geography — France, Greece, Miami, Buenos Aires — becomes a real edge.

We're not in a new tech cycle. We're at the start of a deep reconfiguration of how organisations and work are arranged. That's a longer story than 2026, but 2026 is when the founders who'll define it are placing their bets.