Anthropic Plans to Raise Up to $50 Billion This Summer, Pre-Funding Valuation at $900 Billion
Anthropic is preparing for a new round of massive financing, with a pre-funding valuation expected to reach $900 billion, surpassing OpenAI's record of $852 billion set in March this year.
The company's annualized revenue is about to exceed $45 billion, a fivefold increase from $9 billion at the end of last year, primarily driven by the Claude Code developer tools and Cowork collaboration products, which have captured market share from OpenAI in the enterprise sector.
The funding will be used to fill the computing power gap, as the next-generation Mythos model is currently limited to a few partners due to resource constraints. Anthropic has signed long-term agreements with SpaceXAI, Google, Broadcom, and AWS.
Source: Public Information
ABAB AI Insight
Anthropic previously received substantial cloud credit support from Amazon and Google in multiple funding rounds in 2025. This $900 billion valuation financing continues its strategy of "locking in computing power with capital + enterprise deployment." Founder Dario Amodei has repeatedly emphasized the need for hundreds of billions in computing resources to catch up with the forefront. This round of financing will further solidify its advantages in enterprise AI security and compliance.
In terms of capital strategy, Anthropic is rapidly amplifying enterprise subscription revenue through its Claude Code and Cowork products. The massive financing will primarily flow into long-term computing power procurement contracts and data center construction, motivated by the need to secure supply before the full release of Mythos and to avoid falling behind OpenAI's GPT-5.5 series, while paving the way for an IPO by the end of the year to allow some early investors to exit.
Similar to OpenAI's rapid iteration after its high-valuation financing in March this year, and companies like xAI binding hardware supply chains through financing, top AI labs are currently transitioning from competition in model training to a dual-driven control of computing power and enterprise revenue.
Essentially, this represents a concentration of capital: achieving pricing power through ultra-large-scale financing accelerates the shift from general AI to enterprise security verticals. The mechanism is that the fivefold growth in annualized revenue provides valuation support, allowing capital to concentrate from dispersed startup projects to a few AI platforms with closed-loop implementations, forming an infrastructure barrier of "winner takes all."
ABAB News · Cognitive Law
When revenue grows fivefold, a $900 billion valuation is no longer a bubble but a ticket to the computing arms race. What is truly valuable is not the model, but the subscriptions that sell the model to enterprises and the locked-in computing power. When AI company financing exceeds national budgets, technological competition becomes a race of capital and supply chains.