Fluidstack, the AI data center startup that builds specialized compute infrastructure for frontier AI labs, is in discussions to raise a $1 billion funding round at an approximately $18 billion valuation, according to Bloomberg as reported by TechFundingNews and TechCrunch. The round is expected to be led by Jane Street, the quantitative trading firm that has expanded aggressively into venture investing.

If the deal closes, it would more than double Fluidstack’s valuation from December 2025, when the company was reportedly looking to raise approximately $700 million at $7.5 billion, a round that was never formally confirmed. Earlier discussions also reportedly involved Situational Awareness, the fund founded by former OpenAI researcher Leopold Aschenbrenner, and other high-profile tech investors, per TechCrunch.

The Anthropic Anchor

Fluidstack’s valuation trajectory is anchored by a $50 billion compute agreement with Anthropic, signed last year, to develop custom AI data centers in the U.S. That deal, among the largest single compute commitments in the AI industry, gave Fluidstack the revenue visibility that attracts growth-stage investors at these multiples.

The company’s client base extends beyond Anthropic to include Meta, Poolside, and Mistral, according to TechFundingNews. Originally founded in Europe, Fluidstack has since relocated its headquarters to New York and shifted its focus toward the U.S. market.

The Neocloud Premium

Fluidstack represents the “neocloud” category: purpose-built AI data center providers that emerged because hyperscalers’ generalist infrastructure is suboptimal for the specialized GPU cluster configurations required by frontier AI training and inference. Unlike AWS, Azure, or Google Cloud, neoclouds like Fluidstack accelerate AI projects by removing compute bottlenecks through dedicated, AI-optimized infrastructure deployments.

The $18 billion valuation, if confirmed, would make Fluidstack one of the most highly valued private AI infrastructure companies globally. The round is still in discussions and has not closed. Investors are pricing the AI compute supply layer at a premium that reflects a specific bet about the scale of the agentic economy: the same frontier models that power Claude Code, Claude Cowork, and enterprise agent deployments need dedicated data center capacity, and the companies building that capacity are commanding growth-stage valuations that rival the AI application companies they serve.