Mentions of open models in latest US earnings calls surged 6x YoY, with open models hitting 56% of Vercel tokens in August and 40% of AT&T's AI workloads

The cost of running AI models has dropped significantly for some workloads.

Mentions of open-weight AI models in U.S. corporate earnings calls surged sixfold year-over-year, indicating a significant increase in corporate discussion and adoption. In August, open-weight models constituted over 56% of token usage on Vercel's platform and 40% of AT&T's artificial intelligence workloads. Despite this high usage, open-weight models captured only 14% of the spending on AI models. This trend suggests a market preference for more cost-effective, widely available AI solutions, with the expectation that value will shift from model providers to companies controlling distribution, data, and customer relationships.

The substantial growth in open-weight model usage, outpacing their market share of spending, signals a strategic shift in AI adoption. Companies are prioritizing "good-enough" AI capabilities that are both cheaper and more accessible, moving away from the necessity of expensive, frontier proprietary models for many applications. This indicates a potential decoupling of AI performance and cost, where innovative companies can leverage open models for efficiency.

This trend has profound implications for the AI market's future economics. As open models mature, the competitive advantage is expected to transfer from those who develop the core models to those who manage distribution channels, possess unique data sets, and maintain strong customer relationships. Businesses focusing on integrating AI into existing workflows and customer touchpoints are poised to capture greater value.