Google vice president predicts the demise of two types of AI startups

One of Google's executives publicly stated what the venture capital industry has been whispering about for over a year: a significant portion of generative AI startups are built on sand. The company's vice president warned that two specific classes of AI businesses — LLM wrappers and AI aggregators — are facing mounting pressure that could prove fatal for them.

To understand the essence of this warning, it's worth clarifying the terms. LLM wrappers are startups that take someone else's language model (typically from OpenAI, Anthropic, or Google itself) and build a product on top of it with their own interface and minimal added logic. A classic example is a service that connects to the GPT-4 API and offers, say, generation of marketing copy or legal documents. AI aggregators operate similarly, but attempt to combine access to multiple models simultaneously, positioning themselves as a "one-stop shop" for users. Both types of companies experienced explosive growth in 2023–2024, when simple access to generative AI seemed like a sufficient competitive advantage.

The problem that the Google representative points to is structural in nature. The margins of wrappers are being squeezed from both directions simultaneously. From below, model providers are pressing in: OpenAI, Anthropic, and Google systematically lower API prices while simultaneously expanding their own functionality. Every update to the base model — whether improved context handling, built-in internet search, or native multimodal support — eats away at the value the wrapper tried to create independently. From above, users are pressing in, increasingly understanding that the difference between a startup's polished interface and direct access to the model is often minimal.

For aggregators, the situation is even more complex. Their key promise — giving users a choice of the best model for a specific task — loses value as leading models converge in capabilities. When the difference between Claude, GPT, and Gemini is measured in nuance rather than a chasm, the very idea of aggregation loses its meaning. Moreover, major platforms are actively integrating AI directly into their ecosystems: Google into Workspace and Search, Microsoft into Office and Windows, Apple into its devices. The intermediary between user and model becomes redundant.

The context in which this warning was delivered is also important. The venture market for AI startups is undergoing a sobering phase. According to analysts, in 2025 the volume of funding for early-stage AI companies began to contract, and investors increasingly demand evidence of sustainable unit economics rather than just impressive demos. Startups whose business model amounts to markup on someone else's API find themselves in a vulnerable position: they are easy to replicate, their margins are transparent to investors, and their dependence on the model provider creates existential risk with any change in conditions.

Of course, the Google vice president's words are not without self-interested undertones. A company that itself is a provider of base models has a vested interest in clients working with it directly rather than through intermediaries. Nevertheless, the objective logic of the market confirms his thesis. The startups that will survive are those that create their own models for narrow tasks, accumulate unique data, or build deep vertical integration in specific industries — medicine, law, manufacturing. A simple layer on top of someone else's intelligence is not a technology company; it is a temporary arbitrage opportunity.

The generative AI industry is entering a phase that can be compared to the evolution of mobile apps: after the initial boom, when any flashlight app attracted millions of downloads, the market ruthlessly culled those who could not offer real and sustainable value. For AI startups, a similar moment of truth is arriving. Those who built their business on access to others' technology rather than on their own expertise risk discovering that the ground beneath them has vanished faster than they could build walls.