The surge in generative AI led to the rapid emergence of startups. Yet, as the hype fades, two business models, LLM wrappers and AI aggregators, are appearing as warning signs. At the helm of Google’s global startup division at Cloud, DeepMind, and Alphabet, Darren Mowry cautions that startups using these models should consider their “check engine light” to be on.
LLM wrappers are startups that layer products or UX over existing large language models like Claude, GPT, or Gemini to address specific issues, such as AI aiding student studies. Mowry emphasizes the fleeting industry interest in those relying solely on backend models. LLM wrappers encasing models like Gemini or GPT-5 imply a lack of differentiation. Startups must establish “deep, wide moats,” horizontally or vertically, for progress, exemplified by tools like Cursor or Harvey AI.
Startups can’t simply overlay a UI on a GPT and expect success, as seen during OpenAI’s ChatGPT store launch in mid-2024; they must deliver sustainable product value now.
AI aggregators aggregate multiple LLMs into singular interfaces or APIs to manage queries. Often, they offer monitoring, governance, or evaluation tools, but Mowry advises startups to avoid this approach. Aggregators lack growth as users demand intellectual property ensuring proper model routing based on need.
Mowry, seasoned in the field from his tenures at AWS and Microsoft and currently at Google Cloud, observes the landscape resembling early cloud computing days when startups arose to resell AWS infrastructure. As AWS introduced its enterprise tools, many of those startups vanished, except those offering real services. Today, AI aggregators encounter similar challenges as model providers advance, potentially diminishing intermediary roles.
Mowry is optimistic about vibe coding and developer platforms. In 2025, startups like Replit, Lovable, and Cursor gained significant investment, supported by Google Cloud, as Mowry highlights. He foresees growth in direct-to-consumer technology, highlighting opportunities for students in film and TV to utilize AI tools like Google’s video generator, Veo, to realize their visions.
Beyond AI, Mowry sees biotech and climate tech as booming sectors, attracting venture investment and providing startups access to substantial data, enabling unprecedented value creation.