AI App Crisis, OpenAI Does Math, Big Nvidia Deal

The latest RevenueCat analysis reveals a stark retention gap between AI‑powered and traditional subscription apps. While AI apps monetize quickly, their 12‑month retention hovers around 21 %, compared with over 30 % for non‑AI counterparts. High churn is driven by over‑hyped promises, early‑stage bugs, and a flood of novelty tools that fail to deliver lasting value. Categories such as photo and video see the fastest AI adoption, whereas gaming and travel lag far behind, highlighting uneven market penetration and the need for focused product‑market fit.

OpenAI’s recent rollout of dynamic visual explanations transforms ChatGPT into an interactive learning platform. Users can now manipulate variables in real‑time diagrams covering over 70 math and science concepts—from Pythagorean theorem to kinetic energy—turning static answers into hands‑on experiments. With more than 140 million weekly users leveraging ChatGPT for math and science, this feature promises deeper comprehension and higher engagement, positioning AI as a catalyst for smarter education rather than a shortcut.

In parallel, Thinking Machine Labs announced a multi‑year partnership with NVIDIA, committing at least one gigawatt of Vera Rubin AI systems to power its Tinker API and future models. The deal, slated to begin in 2027, underscores the escalating competition for compute resources as the industry projects $3‑4 trillion in AI infrastructure spend. By securing massive GPU capacity and a strategic investment from NVIDIA, Thinking Machine Labs aims to deliver more reliable, replicable AI outputs, signaling a broader shift toward heavyweight compute alliances that will shape the next generation of AI services.