There’s a difference between backing a fast follower and backing a frontier founder.

We’ve seen incredible generalist VCs do both — but when it comes to deep tech— AI infrastructure, robotics, advanced systems — the approach needs to change. Pattern-matching doesn’t work when the pattern hasn’t been built yet.

At VentureCapital, we spend most of our time in high-complexity, low-visibility territory. This post breaks down a few things we think generalist investors often miss when evaluating deep tech founders — and how we do it differently.

1. Surface Signals ≠ Real Readiness

A slick demo and an impressive GitHub repo aren’t enough.

Many generalists overweight what they can see — the UI polish, the deck flow, the buzzwords — because the tech itself is harder to evaluate. But in deep tech, the real progress often lives in:

• A novel systems architecture or compiler-level insight

• Edge-case testing environments

• The founder’s understanding of downstream failure modes

We go deeper. We bring in technical experts. And we ask questions that don’t show up in standard DD checklists.

The TAM Trap

We often hear: “Is this market big enough?”

In deep tech, that question is backwards.

Most breakout deep tech companies create new markets or reshape old ones. There was no TAM for “cloud GPUs” or “AI-native observability” ten years ago. There is now.

Instead of top-down TAM math, we look for:

• Pain that’s visceral and unsolved

• Clear starting wedges with room to expand

• Macro or technical tailwinds that shift market dynamics over time

The best deep tech startups don’t ride trends — they become them.

Founders Sound Different

Deep tech founders often sound… less polished. Less rehearsed. Sometimes even a little scattered. That’s not a red flag — it’s the cost of thinking at the edge.

They speak in systems. They over-explain because they care. They reference preprint papers and system limitations — not competitors and blitzscaling.

We don’t penalize this. We lean into it. And when the ambition and clarity emerge underneath, it’s obvious we’re looking at something special.

Traction Looks Different Too

For an AI infra startup or a robotics platform, “revenue” at seed may not be the signal.

What we look for instead:

• Usage by technical teams solving real problems

• Repeatability in deployment or integration

• Signs of obsession from early users — even if the product breaks sometimes

In deep tech, traction often starts with love, not money.

Our POV at VentureCapital

We believe evaluating deep tech requires a different lens:

• One that’s willing to read technical docs, not just Twitter threads

• One that values first-principles clarity over slide polish

• One that plays long games, not fast flips

We’re not here for hype cycles. We’re here to fund the systems that reshape what’s possible.

Final Word

If you’re building something that sounds complicated, uncertain, or too early — talk to us.

We don’t expect a perfect pitch. We expect the ambition to matter.

And we’ll meet you in the complexity — and help you build from it.