OpenAI, Anthropic slash token prices as enterprises curb AI spend

OpenAI and Anthropic have announced token‑price reductions of roughly 50% and 67% respectively, a move forced by enterprises that are tightening AI budgets and demanding lower‑cost alternatives [1].

Price cuts and enterprise pushback

Enterprises such as Uber have already capped AI spending, limiting usage of Claude Code and Cursor to $1,500 per month per tool after burning through its 2026 AI coding budget in just four months [1]. The pressure has pushed OpenAI to roll out “Flex” processing at a half‑price discount, while Anthropic’s latest Opus 4.5 launch slashed Claude Opus pricing by two‑thirds [1]. Companies are responding by routing simple tasks to far cheaper open‑source models from Chinese firms like Alibaba and DeepSeek, reserving the most expensive models for only the most complex work [1]. This cost‑optimisation is reflected in a 1,001% surge in business token consumption from January 2025 to April 2026, outpacing the 497% rise in overall spend [1].

Ripple effects across the AI supply chain

The token‑price war could compress margins for AI chatbot providers, with Anthropic already lowering its gross‑margin outlook to 40% because of inference costs on Google and Amazon servers [1]. Nvidia’s dominance may erode as custom chips—Google TPUs, AWS Trainium, Microsoft Maia, and Meta MTIA—are projected to cut inference costs by 30‑50% and capture a growing share of AI servers, potentially reaching 40% of deployments by 2030 [1]. Meanwhile, neocloud operators such as CoreWeave, which carries $29 billion in liabilities and $9.7 billion due within twelve months, could face tighter cash flow if lower token prices depress revenue streams [1].

What to watch

• IPO pricing – Monitor the valuation adjustments for OpenAI and Anthropic as token‑price cuts reshape revenue forecasts.
• Nvidia market share – Track the adoption rate of custom AI chips versus Nvidia GPUs through 2028.
• Neocloud debt service – Watch CoreWeave’s upcoming $9.7 billion maturity and any refinancing activity.

The steep token‑price cuts underscore a shift from “token‑maxxing” to cost‑conscious AI deployment, raising the question of whether lower prices will revive enterprise demand or accelerate a broader pullback from high‑margin AI services.