The emergence of powerful and cost-effective open-weight AI models, especially those developed by Chinese companies, is posing a significant challenge to the business models of leading US AI firms such as OpenAI and Anthropic. These American companies derive their high valuations from their ability to maintain a technological edge that justifies premium pricing and promises substantial future growth. However, if comparable models become widely accessible at little to no cost, investors may question the sustainability of these valuations.
Chinese firms are rapidly advancing in the open-weight AI space, offering models that are not only increasingly capable but also dramatically cheaper, sometimes by 60% to 90% compared to US rivals. For example, DeepSeek's V4 Flash model costs $0.14 per million input tokens, a stark contrast to OpenAI’s GPT-5.5 at $5.00, or Claude Opus 4.8, which can charge around $5 for input and $25 for output per million tokens. This pricing disparity has led to a significant shift, with Juniper Research noting that work on platforms like OpenRouter through leading providers like OpenAI, Google, or Anthropic has fallen from 70% last year to just 30%.
This shift is already impacting pricing strategies; OpenAI recently reduced the price of its smaller GPT-5.6 models, with its Luna model seeing an 80% price cut. While leading closed models still hold a slight performance edge (e.g., Stanford's 2026 AI Index indicates a 3.3% gap as of March 2026), for most enterprise tasks, this gap is not substantial enough to justify significantly higher costs. Enterprises are increasingly considering open models that are 90% to 95% as capable for a fraction of the price, leading to economic rationalization for switching providers.
The increasing adoption of cheaper open models, particularly for frequent tasks like programming and agentic workloads, raises concerns about the funding of the massive AI infrastructure buildout in the West. If inference revenue, which underpins these investments, weakens due to price competition, the complex financing structures used by hyperscalers (like off-balance-sheet financing and circular financing) could become problematic. Chinese companies, often supported by external financial businesses or integrated into cloud provider offerings that sell compute rather than the model itself, face fewer immediate funding pressures in this regard.
Despite the competitive threat, major US tech companies like Nvidia, Microsoft, Meta, IBM, and Palantir are advocating against premature restrictions on open-weight models, arguing that widely available American AI can accelerate adoption, strengthen the domestic ecosystem, and prevent China from gaining a global technological advantage.
