Corporate America is increasingly turning to lower-cost “open” AI models as IT expenses rise, with mentions of “open weight” or “open source” models in earnings calls and investor conferences surging sixfold in August and September compared to the same period last year. While the shift is most pronounced among technology companies, cross-industry adoption is evident, with PNC Financial Services, CH Robinson, and Siemens discussing usage. The open-weight AI market is dominated by Chinese firms, though Mistral (France), Nvidia, Reflection AI, and Thinking Machines Lab (US) also contribute.

Public market investors are closely monitoring AI spending allocations between frontier and open-weight models. Open-weight models accounted for 56% of all tokens processed through Vercel’s AI Gateway in August, up from 7% in December. Vinay Kuruvila, CTO of Tinder (Match Group), reported AI costs rising from $1mn to $10mn annually, with potential to reduce reliance on frontier models like OpenAI’s Astra and Claude if open-weight alternatives improve.

Both Anthropic and OpenAI face pressure as cheaper alternatives threaten revenue, with Anthropic preparing for a $2tn IPO and OpenAI seeking a $1.2tn private funding round. Usage data confirms real adoption beyond rhetoric, with Chinese open-weight labs like DeepSeek and Zhipu generating revenue through token sales.