No data, no defense: How AI startups survive the tech giants

This article summarizes an episode of VC10X with Prashant Choubey’s video series featuring David Hornik, founding partner at Lobby Capital.

David Hornik, founding partner at Lobby Capital/ Photo credit: Lobby Capital

For David Hornik, a founding partner at Lobby Capital, the safest bet in AI investing is software that makes human work more valuable.

He argues that venture capital should only flow to startups that meet three conditions. The startups must hold advantages incumbents cannot copy.

They must also offer entry prices that make venture returns mathematically possible, and empower human workers in roles where physical presence is non-negotiable.

AI startups need something rivals cannot copy

AI startups are not entering an empty field. The largest tech companies understand the threat, own the distribution channels, and have the capital to respond quickly.

Because a polished demo offers weak defense, Hornik looks for advantages that incumbents cannot easily rebuild. To identify true defensibility, he applies these practical tests:

• Pass on industry AI apps if Claude, Gemini, or OpenAI can easily copy the main workflow.
• Require proprietary data that the company owns and improves continuously as customers use the product.
• Look for deep integrations into customer systems that rivals cannot quickly map or replicate.
• Back difficult-to-replicate technical work in memory, data handling, infrastructure, or model quality.
• Assume tech giants like Google, Meta, Microsoft, Amazon, and Apple will enter the market if it becomes large enough.

Hornik explains, “You need to have… a proprietary position in the market. You either need proprietary data integrations and understanding that are distinct, or you need unique tech.”

This standard quickly narrows the deal list. Basic, single-industry AI apps face price pressure and short sales windows. Conversely, the strongest companies build structural advantages that compound with customer use.

Investors face a tougher environment today than during earlier software waves. During the internet boom, incumbents often reacted late. In the AI era, large platforms are already spending billions to add features to existing products and use their reach to box out startups.

Capital must flow toward companies where workflow knowledge or technical design creates a widening barrier. A demo cannot compete when a rival already owns the model, the sales channel, and the cloud infrastructure.

Regret over missed winners cannot replace price discipline

If defensibility is the first filter, price discipline is the second. Even a highly defensible company becomes a poor investment if the entry valuation is too high.

Hornik separates the sting of missing outlier companies from the discipline of securing meaningful ownership in the startups he does back. Missing early investments in LinkedIn and Uber taught him to respect network effects and shifting consumer behaviors, but it did not teach him to chase hot deals at any price.

Helping people may be the overlooked AI market

Care work turns human presence into software demand