Google Cloud VP Darren Mowry has issued a stark warning to AI startups: businesses built as LLM wrappers or AI aggregators face an uncertain future. Speaking on the Equity podcast, Mowry said startups that simply layer a user interface on top of existing models like GPT or Gemini are seeing their "check engine light" come on as investors lose patience with thin intellectual property.

Mowry, who previously worked at AWS and Microsoft, draws parallels to the early cloud computing era when startups reselling Amazon's infrastructure were squeezed out once AWS built its own enterprise tools. The same pattern is emerging in AI, where model providers are expanding directly into enterprise features. Startups like Cursor and Harvey AI have survived by building deep moats through specialized vertical applications, while generic wrappers struggle to differentiate.

For AI aggregators that route queries across multiple models, Mowry's message is equally blunt: "Stay out of the aggregator business." These platforms face margin pressure as users demand more than just orchestration layers - they want built-in intellectual property that ensures intelligent routing based on specific needs rather than backend constraints. Meanwhile, Mowry sees opportunities in vibe coding platforms, direct-to-consumer AI tools, and sectors like biotech and climate tech where massive datasets enable real value creation.