Major AI firms such as OpenAI and Anthropic are struggling to maintain profitability as the rapid commoditization of AI models and token usage pricing pressures margins across the industry.
- AI token costs outpace subscription fees, pressuring margins.
- Open weight AI models from China and elsewhere close performance gaps.
- AI firms shift focus to layered services and regulatory lobbying.
What happened
Leading US AI companies such as OpenAI and Anthropic remain unprofitable as the costs associated with training and inference rise sharply. Reports indicate that subscribers paying $200 monthly may consume tokens worth much more, forcing these companies to adopt metered pricing models to sustain revenue. Despite their innovation and dominant market positions, these firms are enduring significant financial strains owing to the economics of AI token consumption.
Simultaneously, commoditization trends are evident with open weight AI models from China and other markets rapidly approaching parity with advanced US models. This shift is compounded by widespread access to AI tokens via proxy services, bypassing geographic restrictions and controls. As a result, maintaining exclusivity and high margins is increasingly untenable for frontier AI labs.
Why it matters
The commoditization of AI models signals a transformative phase for the AI sector, where core model offerings may soon lose their premium pricing leverage. This evolution challenges the long-held business model of pricing based on cutting-edge AI capabilities and token scarcity. Firms will need to pivot towards monetizing complementary services, platforms, and enterprise-specific solutions to preserve profitability.
Moreover, the growing competition from equally capable open and Chinese-developed models compresses the window for US companies to harness their technological lead. The widespread availability of low-cost tokens incentivizes price-sensitive users to seek alternative providers, undermining the dominance of traditional leaders and heightening the urgency for differentiation beyond raw model performance.
What to watch next
Industry observers should closely monitor how leading AI companies adapt their pricing strategies and product portfolios, particularly their emphasis on developer tools, integrated solutions, and regulatory lobbying to sustain competitive advantage. The success of these strategies will provide insights into the viability of layered service models as core AI model commoditization accelerates.
Additionally, tracking the progress of open weight AI models globally, especially those from China, will be critical. These models are expected to match or surpass current US benchmarks by late 2026, signaling a pivotal shift in global AI innovation dynamics. How enterprises balance cost sensitivity with demand for advanced AI features will shape market structures and margins going forward.