During the current earnings season, several companies have reported that increasing AI and inference costs are negatively affecting their profit margins. Shopify noted that while economies of scale helped, they were still burdened by higher costs associated with large language models (LLMs).
This situation raises concerns about the projected valuations of OpenAI and Anthropic, both anticipated to exceed $800 billion. These valuations rely on the assumption that these companies will maintain their market share and pricing power, which is now in question as cheaper AI solutions become more prevalent.
Chinese labs are offering comparable AI services at significantly lower prices, with models like DeepSeek's costing as little as $544, compared to Anthropic's Claude at $4,811. This price disparity is prompting enterprises to reconsider their AI spending, with a survey indicating that 45% of companies expect to spend over $100,000 monthly on AI by 2025, up from 20% the previous year.
Google is also responding to this trend by promoting its more affordable AI models, suggesting that large customers could save over $1 billion annually by switching to cheaper options. The shift towards cost-effective AI solutions is evident, with Chinese models now accounting for over 60% of usage on platforms like OpenRouter.
As enterprises adopt strategies to minimize costs, such as using cheaper open-source models for routine tasks, the competitive edge of American AI firms may be eroding. While companies like Cohere are finding success in regulated markets, the broader enterprise sector is increasingly drawn to lower-cost alternatives.
The implications for OpenAI and Anthropic are significant; as they prepare for their IPOs, they must demonstrate robust revenue growth and market dominance in a landscape where their pricing power is under threat