Google’s global startup lead Darren Mowry has issued a warning to AI startups, cautioning that two popular business models—LLM wrappers and AI aggregators—are at high risk of failure as the generative AI market matures. Speaking on the Equity podcast, Mowry said these startups now have their “check engine light” on, signaling structural vulnerabilities. LLM wrappers are companies that build products on top of existing large language models like GPT, Claude, or Gemini, often adding only a user interface or minor customization. Examples include startups that use AI to help students study or automate customer service. Mowry argues that simply wrapping a powerful model with a thin layer of branding or UX is no longer enough. “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,” he said. To survive, startups must create deep, defensible advantages—either through horizontal innovation or strong specialization in a specific vertical market. He cited Cursor, a GPT-powered coding assistant, and Harvey AI, a legal-focused AI tool, as examples of LLM wrappers that have built real moats through domain expertise and integrated workflows. These companies didn’t just repackage existing models; they embedded AI into specialized developer and professional workflows. AI aggregators, a subset of wrappers, attempt to solve the complexity of using multiple models by offering a unified interface or API that routes queries across different LLMs. Platforms like Perplexity and OpenRouter fall into this category, providing access to multiple models through a single layer with added features like monitoring and evaluation. However, Mowry advises startups to stay out of this space. “Users want some intellectual property built in” to ensure they get the best model for their needs, not just access based on compute availability or routing logic. Mowry draws a parallel to the early days of cloud computing, when a wave of startups emerged to resell AWS infrastructure, offering billing tools and support. But as Amazon expanded its enterprise offerings and customers gained confidence managing cloud services directly, most resellers disappeared. The survivors were those who added real value—like security, migration, or DevOps services. Today, as model providers like Google, OpenAI, and Anthropic roll out enterprise-grade tools, AI aggregators face similar pressure. In contrast, Mowry remains optimistic about developer platforms and vibe coding tools—areas that saw explosive growth in 2025. Startups like Replit, Lovable, and Cursor have attracted major investment and strong customer adoption, many of them already Google Cloud customers. He also sees strong potential in direct-to-consumer AI, where tools like Google’s Veo video generator empower creatives—such as film and TV students—to bring ideas to life. Beyond AI, Mowry believes biotech and climate tech are experiencing a surge in momentum, driven by both venture funding and unprecedented access to data that enables new forms of innovation. He sees these fields as ripe for breakthroughs that were previously out of reach.