LLMs

Enterprise AI Loyalty Is a Myth: Companies Are Switching Labs With Every Model Release

The most uncomfortable truth in enterprise AI right now is that nobody is actually winning. New data shows businesses are flip-flopping between OpenAI and Anthropic as each lab ships a new model, exposing a fundamental fragility in the business case both companies are selling to investors. Stickiness, the quality that makes enterprise software so valuable, appears to be nearly absent from the AI stack so far. Both labs are responding not by building deeper moats through capability, but through trust: OpenAI this week launched Zero Data Retention for API customers and announced new privacy protections designed to out-flank Anthropic on the enterprise compliance checklist, while Anthropic is rolling out watermarking for Claude-generated content. Meanwhile, OpenAI published a headline-grabbing case study claiming Asana used Codex to complete five years of engineering backlog in two weeks for roughly $12,000, a powerful ROI story that signals the lab understands the battle is now being fought on CFO spreadsheets, not benchmark leaderboards. NVIDIA publicly endorsed ChatGPT Work as a workflow tool across its global teams, giving OpenAI a marquee brand endorsement that few competitors can match. The competition is fierce, the products are improving fast, and the customers are watching, wallets open but fingers hovering over the exit button.