Why AI Models Are Now Being Built By Former Rivals

This episode of AI Business with Fexingo explores a counterintuitive trend in the AI industry: former competitors are teaming up to build the next generation of models. Lucas and Luna break down the economics behind this shift, using the recent partnership between Cohere and Mistral AI as a focal point. They discuss the staggering capital requirements for frontier model training—hundreds of millions in compute alone—and how even well-funded startups are realizing that going it alone is no longer viable. The conversation touches on why this consolidation mirrors earlier tech cycles, and what it means for enterprise buyers who now face even fewer independent model providers. With specific references to NVIDIA's hardware costs and the soaring price of H100 clusters, the episode grounds the discussion in real numbers. Key takeaways: the cost of training at scale is forcing collaboration, talent is pooling in fewer orgs, and the era of the solo AI lab may be ending. #AI #ArtificialIntelligence #Business #Technology #Cohere #MistralAI #ModelTraining #AIInfrastructure #NVIDIA #H100 #ComputeCosts #Consolidation #Startups #EnterpriseAI #FrontierModels #FexingoBusiness #BusinessPodcast #AIModels Keep every episode free: buymeacoffee.com/fexingo