From generalist to deep: Why venture is being forced to specialize

For more than a decade, I invested as a generalist, backing exceptional founders across enterprise software and fintech. That model worked extraordinarily well for a long time. But over the past few years, it has become increasingly clear that the ground beneath venture capital is shifting.

Open-source models, low-code tools, and foundational AI have dramatically lowered the cost of building digital products. Creation is faster, cheaper, and more accessible than ever. As a result, differentiation has moved downstream, away from software itself and toward the physical systems that make the digital world run.

Compute, energy, manufacturing, and hardware-software integration are becoming the new frontiers of value.

We are living through a rare convergence of forces: large-scale reshoring and reindustrialization across the U.S., an accelerating technological arms race between the U.S. and China, and the AI supercycle, the most profound technological transformation since the dawn of the internet.

But unlike prior software waves, this one is constrained not by code, but by physics.

The next generation of trillion-dollar companies will be built by those who can solve hard, non-abstractable problems: scaling compute, generating abundant energy, automating physical labor, and pushing the limits of materials, manufacturing, and robotics. These are not marginal improvements, they are the foundations that determine whether the AI revolution can scale at all, and which nations ultimately lead it.

In short, advantage is shifting to those who can break through the walls of compute, energy, and manpower.

This shift is forcing a structural change in venture itself.

In deep tech, surface-level understanding is no longer sufficient. Conviction requires technical fluency, long feedback loops, and the ability to assess risk across science, manufacturing, regulation, and geopolitics, not just product-market fit.

Increasingly, successful investors in frontier technology are thesis-led and expert-driven. They build narrow focus areas, deep sourcing channels, and long-term relationships with researchers, operators, and institutions. The era where pattern recognition alone could carry an investor is fading.

Government also plays a different role in this world. In frontier technology, states are no longer passive regulators. They are customers, funders, and strategic partners. Non-dilutive capital, from grants, contracts, and public-private programs, has become a critical accelerant, often determining which technologies reach scale first.

Frontier technology is not evenly distributed.

A small number of ecosystems combine elite research institutions, defense and industrial know-how, and entrepreneurial density. The U.S. remains the largest such ecosystem. Israel, despite its size, has quietly become the only meaningful other, with exceptional depth across semiconductors, optics, RF, robotics, and defense-driven engineering.

The U.S.–Israel corridor, in particular, has emerged as a powerful engine for deep-tech innovation: scientific discovery on one end, scale and market access on the other. As capital increasingly flows toward physical infrastructure and strategic technologies, that corridor is becoming more important, not less. That is also the basis for the recently signed Pax-Silica, a deliberate collaboration across the deep tech ecosystem.

Richard Feynman once said, “What I cannot create, I do not understand.”

That idea increasingly applies to venture capital itself.

The generalist era wasn’t a mistake, it was a product of a software-dominant world. But as software becomes abundant, advantage shifts to what cannot be abstracted away: atoms, energy, and physical limits.

The next generation of outsized outcomes will belong to those who can operate at the intersection of science, scale, and capitalץ

Depth is no longer a preference. It is the price of admission.