OpenAI and Anthropic vs. app developers: tech's Cronos syndrome (www.economist.com)

Big AI labs such as OpenAI and Anthropic are increasingly competing with the very startups that build apps on their models — a phenomenon dubbed “Cronos syndrome,” where the platform devours its offspring. Rather than just exposing raw APIs, these providers are packaging higher-level features (customizable assistants, tool integrations, retrieval-augmented workflows, built-in fine-tuning) and surfacing them as first‑party products. That makes it easy for end users to get the same functionality without the middle‑man apps, while platform owners control pricing, rate limits, data flow, and distribution channels. This shift matters for the AI/ML community because it changes incentives and technical risk profiles. App developers face vendor lock‑in, sudden model updates that break integrations, and margin pressure when providers internalize value. On the technical side, the rise of turnkey capabilities — embeddings + RAG, managed tool APIs, cheap fine‑tuning/adapter layers and faster inference — lowers the barrier for platforms to replicate third‑party features. The result is a concentration risk for innovation, stronger arguments for open models and standards, and growing regulatory scrutiny. Developers should diversify (multi‑model stacks, on‑device or self‑hosted inference), demand clearer SLAs/data contracts, and design modular apps that can swap model backends to survive this platform‑versus‑app dynamic.