Google VP Warns That Two Types of AI Startups May Not Survive

When the generative AI boom began, it seemed like a new startup launched every minute. Founders raced to capitalize on the excitement, building products that promised to revolutionize everything from coding to content creation. But as the dust settles and the market matures, a stark reality is emerging: not every AI business model is built to last.

Darren Mowry, who leads Google’s global startup organization across Cloud, DeepMind, and Alphabet, has issued a sobering warning. Two once-hot categories of AI startups—LLM wrappers and AI aggregators—are now flashing what he calls the “check engine light.”

“If you’re really just counting on the back-end model to do all the work and you’re almost white-labeling that model, the industry doesn’t have a lot of patience for that anymore.” — Darren Mowry, Google VP

The Wrapper Problem

LLM wrappers are startups that essentially wrap existing large language models—like Claude, GPT, or Gemini—with a product or user experience layer to solve a specific problem. Think of a study aid for students or a simple content generator for marketers.

These businesses thrived in the early days of the AI boom. When OpenAI launched its ChatGPT store in early 2024, the barrier to entry seemed impossibly low. Slap a user interface on top of a powerful model, and you had a product. But Mowry argues that those days are over.

Wrapping “very thin intellectual property” around existing models signals a lack of differentiation, he says. The market has moved on. Users and investors alike are now looking for genuine innovation, not just a prettier window into someone else’s technology.

The Aggregator Squeeze

AI aggregators face a similar predicament. These startups combine multiple LLMs into a single interface or API layer, routing queries across different models and offering users access to several options at once. They typically provide orchestration layers that include monitoring, governance, or evaluation tooling.

Companies like Perplexity and OpenRouter fall into this category. While they’ve gained traction by offering convenience, Mowry sees trouble ahead. As model providers like Google, OpenAI, and Anthropic expand their own enterprise features, aggregators risk being squeezed out of the value chain.

The comparison to cloud computing is telling. In the late 2000s, startups that simply resold AWS infrastructure found themselves obsolete once Amazon built its own enterprise tools. Only those offering real services—security, DevOps consulting, genuine expertise—survived. Mowry believes AI aggregators are heading for the same fate.

“You’ve got to have deep, wide moats that are either horizontally differentiated or something really specific to a vertical market to progress and grow.” — Darren Mowry

What Will Survive

Not all wrappers are doomed. Mowry points to Cursor, a GPT-powered coding assistant, and Harvey AI, a legal AI assistant, as examples of sustainable models. These companies have built deep, defensible moats in specific vertical markets. They’re not just wrapping a model—they’re creating genuine value through domain expertise, specialized workflows, and proprietary data.

Developer platforms like Replit and Lovable also offer a roadmap for success. These tools saw record-breaking traction in 2025 by going beyond simple model access to create integrated development environments where AI is just one component of a larger, stickier product.

Mowry is also bullish on direct-to-consumer AI tools that enable creation. He cites Google’s video generator Veo as an example of how students and creators can bring stories to life—tools that democratize creative expression rather than just automating tasks.

Beyond the AI Hype

Looking past the immediate AI landscape, Mowry sees strong momentum in biotech and climate tech—industries where massive datasets can be leveraged to create real, measurable value. These sectors require deep technical expertise and significant capital, creating natural barriers to entry that protect innovative companies.

The message is clear: the AI gold rush is entering a new phase. The easy wins—the wrappers and aggregators that rode the wave of excitement—are facing a reckoning. The startups that will thrive are those that took the harder path: building proprietary technology, developing domain expertise, and creating genuine moats around their businesses.

For founders still betting on thin wrappers, the check engine light is on. The question is whether they can adapt before the market leaves them behind.

This article was reported by the ArtificialDaily editorial team. For more information, visit TechCrunch.