The AI industry's pricing power is under siege from an unlikely force: free and cheap open-source models. As capable open models proliferate, enterprises increasingly favor them over premium proprietary systems, forcing the likes of OpenAI and Anthropic to cut prices. The open-source surge is reshaping the economics of AI — and challenging the assumption that the most expensive models will always win.The open-source riseThe gap has narrowed. Open-source and open-weight models have grown dramatically more capable, closing much of the distance to the frontier while remaining far cheaper — or free — to deploy. For many real-world tasks, a good open model is now good enough, undermining the case for paying premium prices.The enterprise calculusBusinesses are doing the math. Facing large and growing AI bills, enterprises are turning to cheaper open models for workloads that don't require the absolute cutting edge, controlling costs and avoiding vendor lock-in. That shift in demand directly pressures the revenue of the proprietary labs.The price warThe giants are responding. Reports indicate OpenAI and Anthropic have been forced to cut prices as enterprises favor cheaper alternatives, a sign that competition is squeezing margins. The era of charging premium rates with little competition is giving way to a more contested, price-sensitive market.The strategic dilemmaIt is a bind for the leaders. The proprietary labs must justify premium pricing with clearly superior capability, even as open models keep improving and the cost of training frontier systems soars. Balancing the enormous expense of staying ahead against pricing pressure from below is a genuine strategic challenge.Why it mattersIt reshapes AI's economics. The open-source surge could democratize access, lower costs across the industry, and pressure the business models of companies that have raised billions on the promise of premium AI. It also raises questions about how the labs sustain the spending the frontier demands if pricing power erodes.The bottom lineCheap, capable open-source models are squeezing the AI giants, driving price cuts and reshaping the industry's economics as enterprises embrace lower-cost alternatives. The proprietary labs must now prove their premium is worth paying, even as open models keep closing the gap. The open-source surge is a powerful counterweight — and a sign that AI's competitive landscape is shifting beneath the leaders' feet.