Novemind
News

Anthropic Closes $30B At $900B Valuation: What It Means For AI Buyers

25 May 2026

Illustration of two AI vendors on a balance scale tipping toward Anthropic

Anthropic is on track to close a funding round of more than $30 billion at a valuation above $900 billion, with Sequoia Capital, Dragoneer, Altimeter, and Greenoaks each putting in roughly $2 billion. If finalized at that level, Anthropic would overtake OpenAI ($852 billion in March) as the world's most valuable AI startup.

[Source: Bloomberg]

Why This Matters

The frontier AI market is no longer OpenAI plus also-rans. Anthropic moving ahead of OpenAI on valuation confirms what enterprise buyers have been signalling for two years. There are now two credible frontier vendors with real revenue, distinct safety postures, and platform deals across AWS, Google Cloud, and Azure. The "default to OpenAI" reflex from 2023 has quietly ended.

Capacity, not just capability, is being funded. A $30 billion raise is overwhelmingly about compute, data centers, and the talent to run them. This is the round that pays for the next generation of Claude models and the inference capacity to serve them at enterprise scale. Practically: fewer rate limits, more reserved capacity options, and continued downward pressure on per-token pricing in 2026 and 2027.

Concentration risk is shifting, not vanishing. Three or four hyperscalers and frontier labs now sit at the center of almost every business AI deployment. Diversifying across providers is more feasible than it was a year ago, but the underlying compute stack is increasingly held by a small number of players.

Our Take

For most SMEs, this is good news in the medium term and a reminder in the short term. Good news because real competition between Anthropic, OpenAI, Google, and the open-source ecosystem keeps prices honest and feature velocity high. A reminder because the cost of betting your AI stack on a single vendor is rising, not falling.

The practical move is the same one we have been recommending all year. Build an abstraction layer in front of your model calls, run periodic head-to-head evaluations on the workloads that matter, and make sure the team can answer "what happens if our primary AI vendor doubles prices or changes terms" without panic. We laid out a full framework in our guide on choosing Claude, GPT, or open-source LLMs for European businesses, and the same principles apply directly here.

If your AI strategy is still single-vendor by accident, now is the right week to revisit it. Talk to us about AI agent development and we will help you map a vendor-resilient architecture for the next 18 months.