Itay Inbar on NYSE Floor Talk: Where Growth-Stage Tech Investing Is Headed
Greenfield partner Itay Inbar sat down with NYSE Floor Talk in San Francisco for a wide-ranging conversation on the state of growth-stage tech investing. The discussion covered AI's shift from scaling compute to pushing the frontier, the changing nature of defensibility, and why this moment feels different from any that came before.
Product depth is the new moat.
One of the sharper ideas from the conversation is worth sitting with. Itay's observation is that the velocity of building has increased by orders of magnitude. With AI in the stack, shipping product is faster, R&D is more efficient, and feature parity happens in weeks instead of quarters.
The counterintuitive implication: deep, defensible product moats matter more now than ever. When anyone can ship a v1, what separates a breakout company from a feature is how much hard, compounding work is buried underneath the surface.
Itay pointed to three places that depth is showing up: research that takes years to replicate, hardware and software working together in ways that can't be cloned from a GitHub repo, and services coupled with software in configurations competitors can't easily match.
There's a fourth worth naming explicitly - distribution. When everyone can build the product, the companies that win are the ones that get it into the hands of the right customers faster, with stickier GTM motions, and with the kind of category ownership that makes them the default answer to a question the market is starting to ask.
The bar for what counts as a moat has gone up, not down.
The second wave of AI is being built alongside the frontier labs.
The first wave of AI investing was about capital and compute — funding the data centers, the chips, the raw infrastructure needed to train these models at all. That wave is still playing out, but Itay framed what's happening now as distinct: a new ecosystem of companies working directly with OpenAI, Anthropic, and other frontier labs to push the boundary of what AI systems can actually do.
This matters because it reframes where the interesting growth-stage opportunities are. It's not just "apply AI to industry X." It's a layer of companies sitting close enough to the frontier to shape what the frontier becomes. Infrastructure for agent workflows. Evaluation and safety tooling. Specialized data pipelines. Orchestration for multi-model systems. The companies building here aren't downstream of the labs — they're partners in pushing what's possible.
Unit economics are being rewritten, and that's not a bad thing.
Gross margins are compressing across AI-native companies because most of them depend on foundation model APIs that come with real costs. On paper, this looks like a worse business than the pure-software SaaS that defined the last cycle. But the math is more interesting than it first appears.
Itay's point: R&D spend is going down as AI makes engineering more efficient, and the markets these companies are addressing are materially larger than what came before. When software was just software, you were selling into a defined budget line. When software is paired with services, or replaces a human workflow entirely, the addressable dollars expand dramatically. A company with 60% gross margins in a $100B market can be far more valuable than one with 85% margins in a $10B market.