Both US labs slashed model pricing this week as Chinese open-weight rivals erode the cost advantage that underpinned trillion-dollar valuations.
OpenAI and Anthropic released cheaper models in response to gains by Chinese AI labs, according to Ars Technica reporting. The moves mark a shift from the pure capability race that defined 2024-2025 toward a price war on inference costs.
The timing matters: both companies are burning cash on compute while trying to justify valuations built on enterprise lock-in and API margins. Cheaper Chinese open-weight models give buyers leverage they didn't have a year ago.
If frontier labs are competing on price instead of just capability, enterprise buyers gain negotiating power and gross margins on AI infrastructure compress. Boards should model API costs as a moving target, not a fixed line item, and revisit multi-model vendor strategies now.
The daily signal, curated. Get it in your inbox.
Subscribe on LinkedIn →