Google VP Predicts Shakeout: Two Kinds of AI Startups May Not Last
He calls for these startups to have their “check engine light” on.
Topics
News
- Google VP Predicts Shakeout: Two Kinds of AI Startups May Not Last
- Anthropic Expands Claude Into Code Security With AI Reasoning
- Middle East Asserts Strategic Clout at India AI Impact Summit 2026
- OpenAI Nears $100B Raise at $850B Valuation in Historic Funding Push
- Saudi Arabia AI Firm Humain Invests $3bn in Musk’s xAI
- Perplexity Rolls Back AI Ads to Prioritize User Experience
In 2025, over 90,000 AI startups operated globally, with AI being a focal point for investment by more than 70% of venture capital firms. The AI funding trend continues to maintain momentum today. However, navigating the future will be a challenging task for AI startup founders.
Darren Mowry, Vice President at Google Global Startups, notes two in-demand business models that are soon likely to fall under the cautionary radar: LLM wrappers and AI aggregators.
“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,” said Mowry on a podcast.
He calls for these startups to have their “check engine light” on.
LLM wrappers provide a coating of existing large language models, like Claude, GPT, or Gemini, over a product or UX layer to solve a specific problem. With intellectual property around such LLMs, you’re not differentiating yourself, Mowry says.
“You’ve got to have deep, wide moats that are either horizontally differentiated or something really specific to a vertical market” for a startup to “progress and grow,” he said.
Cursor, a GPT-powered coding assistant, or Harvey AI, a legal AI assistant, are a few examples of the deep-moat LLM wrapper type.
For potential AI aggregator startups, Mowry’s message is clear: “Stay out of the business.” A subset of LLM wrappers, AI aggregators aggregate multiple LLMs into a single interface or API layer to route queries across models and give users access to multiple models.
They often provide a holistic layer that includes monitoring, governance, or evaluation tooling.
AI search startup Perplexity and developer platform OpenRouter provide access to multiple AI models via a single API.
However, all the rage around aggregators has subsided because, he says, users are looking for “some intellectual property built in” to ensure they are directed to the right model at the right time, rather than behind-the-scenes compute or access constraints. They face margin pressure as model providers expand their enterprise features, potentially sidelining middlemen.
Mowry is bullish on vibe coding and developer platforms, expects strong growth in direct-to-consumer tech, and feels biotech and climate tech are having a moment, with the startups leveraging “incredible amounts of data” to create real value “in ways we would never have been able to before.”
