Dan Niles, founder of Niles Investment Management, said the artificial intelligence sector is entering a repricing phase, citing constraints on United States data-center capacity, declining token prices, and rising funding costs as factors forcing a reevaluation of valuations across the industry.

Speaking on Bloomberg's "Insight with Haslinda Amin," Niles said these pressures are reshaping the competitive landscape in AI, a sector in which leading firms have taken on substantial long-term computing commitments — obligations that tie companies to capital-intensive infrastructure for years and become more expensive to carry as funding costs rise. His comments align with recent reports of delayed or canceled data-center capacity projects caused by power and grid limitations — constraints that could affect major AI companies such as Anthropic.

Anthropic, which holds significant multiyear commitments to secure computing capacity, is reportedly facing increased scrutiny as investors reassess its valuation prospects. According to the report, recent market activity, reflected in falling token prices — the per-unit rates charged for AI model usage — and rising operational challenges, is consistent with a decreased likelihood of Anthropic reaching its $600 billion valuation target by the end of the year.

Background: an AI buildout meets infrastructure limits

Anthropic is a San Francisco-based AI company founded in 2021 by former OpenAI researchers Dario Amodei and Daniela Amodei. It is best known for its Claude family of AI models and counts Amazon and Google among its largest backers, with Amazon committing investment of up to $4 billion and Google committing up to $2 billion. In September 2025, the company raised $13 billion in a funding round that valued it at $183 billion, making the $600 billion year-end target a steep step up — more than triple the September valuation.

Industry-wide, the race to secure AI computing capacity has driven record capital spending on data centers, but power availability and grid interconnection have emerged as key bottlenecks in the United States. For AI companies, data-center access shapes how much computing power is available to train and run their models. Utilities and grid operators in several regions have struggled to connect new facilities quickly enough to meet demand from cloud and AI providers, leading some projects to be postponed or scrapped. At the same time, higher funding costs have made capital-intensive infrastructure more expensive to finance, adding to the valuation pressure Niles described.

What to watch

Developments in data-center capacity and power access could influence Anthropic's growth potential, as could announcements from key partners such as Amazon and Google regarding their commitments to the company. Changes in Anthropic's strategic initiatives or funding rounds may also affect its valuation outlook as the December 31 deadline approaches.