The devil is, of course, in the detail: $188 billion of that cash went to four companies – OpenAI, Anthropic, xAI, and Waymo – accounting for 60% of all venture capital deployed in Q1.

VCs generally revel in change, but even by our standards what it takes to raise has shifted at an unprecedented rate over the past 12 months: expectations of early-stage founders – in terms of defensibility, scaling, and pricing – have been completely rewritten.

Fearing the ‘AI-pocalypse’, early-stage founders could be forgiven for feeling despondent seeing these figures. The outlook looks daunting, but it doesn’t need to be. The concentration of capital around a handful of AI giants is changing the rules, but the game goes on. The question for founders right now is not whether Big AI wins, but how can smaller companies survive and thrive alongside it.

A Rising Tide Is Lifting Most Boats

Investment in the US rose 190% YoY in Q1 but deal count fell 26%, meaning that capital is concentrating into fewer, larger cheques. However, the remaining $112 billion of funding sits in line with recent quarterly highs, and the environment for smaller rounds remains robust.

Frontier lab mega-rounds do not directly compete with early-stage cheques: indeed, pre-Seed to Series A companies are growing faster than ever before. Stripe recently revealed data showing that the top 100 AI-native companies are scaling from $1 million to $30 million ARR five times faster than previous software generations.

It’s still a very good time to be an early-stage AI company, especially one that can move quickly and provide genuine agentic solutions rather than simple GPT wrappers. That distinction matters because the startups most under threat are those with no meaningful moat – i.e. businesses that lack sufficient brand, switching costs or economies of scale. These companies will be squeezed not only by platform owners from above them, but also by falling prices below them.

Learnings from the SaaSpocalypse

It wasn’t so long ago that Anthropic’s Claude Cowork launched and within hours wiped hundreds of millions off the valuations of some of the world’s leading software companies. Investors and founders alike feared that a new army of AI agents was about to eat everybody’s lunch.

As of June 2026, however, debate on the death of SaaS is ongoing. Enterprise customers could theoretically vibe code much of their own software, but very few are actually going to do so. At least not yet. At the same time, the opposing view – that concerns about model companies are entirely overblown – has become equally unhelpful. The phrase “what if model companies do this?” has become a catch-all objection that often substitutes for harder conversations around business fundamentals.

The right question is not whether model companies will kill software. They won’t. It’s which software companies will become stronger as models improve, and which will become increasingly commoditised by them. That distinction matters because the bar to raise has doubled while investor attention has narrowed dramatically. Raising money outside AI is a tough ask in 2026, but that doesn’t mean every company needs to become an AI company. It means founders need to ask harder questions.

If OpenAI or Anthropic launched this feature tomorrow, what would still belong to us? Are we building a product, or are we building an advantage? Do we control proprietary data, hardware, scientific know-how, distribution, or regulation? Does our business become stronger as models improve, or weaker? Would customers still choose us if intelligence itself became abundant and cheap? These are the questions investors are asking too.

The Best Offense Is a Good Defense

Whatever you do as an early-stage founder right now, defense is key.

Investors are gravitating towards sectors like robotics, defence, photonics, biotech, and novel compute. This isn’t because they are fashionable, but because they possess characteristics that are difficult for foundation model providers to replicate.

Robotics and defence require integration with the physical world, complex systems, and long customer relationships. Photonics and novel compute rely on years of scientific research and manufacturing expertise. Biotech combines proprietary data, regulatory barriers, and deep domain knowledge. Intelligence alone is not enough. 

Investors are looking for companies with assets that cannot simply be replaced by agents. Proprietary data, specialised hardware, scientific expertise and difficult-to-reproduce infrastructure are becoming durable moats. 

No matter what, we are in for a turbulent remainder of 2026. Capital is concentrating at every level, the revenue bar to raise has roughly doubled in five years, and the companies getting funded look nothing like they did three years ago.

This looks bad for early-stage founders, but it doesn’t need to be. Every technological wave creates new giants, but it also creates new opportunities for smaller players that understand where defensibility lies. For founders, the message is simple but not easy: assume intelligence becomes abundant and build the things that remain scarce.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.