The AI infrastructure gold rush just claimed another unicorn. Fluidstack, a datacenter startup that's become essential plumbing for AI training, is in talks to raise $1 billion at an $18 billion valuation, according to TechCrunch. That's a staggering 140% jump from the $7.5 billion valuation it commanded just months ago, fueled by a $50 billion contract to build datacenters for Anthropic. The deal underscores how AI companies are racing to lock down scarce computing capacity, turning infrastructure providers into some of the hottest bets in venture capital.
Fluidstack is suddenly one of the most valuable private infrastructure companies in tech, and it's all happening at breakneck speed. The company is currently in advanced talks to raise $1 billion at an $18 billion valuation, sources familiar with the matter told TechCrunch. That's a remarkable 140% increase from the $7.5 billion valuation it secured in its previous round just months earlier.
The catalyst? A monster $50 billion agreement with Anthropic to build out dedicated datacenter capacity for the AI company's growing computing needs. The multi-year deal represents one of the largest infrastructure commitments in AI history and signals how seriously frontier model companies are taking compute scarcity. For Anthropic, which competes directly with OpenAI and Google in the large language model space, securing reliable access to GPU clusters isn't just strategic, it's existential.
Fluidstack's ascent mirrors the broader frenzy around AI infrastructure. While software companies like OpenAI grab headlines, the real bottleneck in AI development has become physical hardware and the datacenters to house it. Training runs for cutting-edge models now require tens of thousands of Nvidia H100 or H200 GPUs running simultaneously for weeks or months. Companies that can deliver that capacity at scale have become kingmakers.
The startup originally made its name by aggregating spare computing capacity from gaming rigs and crypto mining operations, then evolved into building purpose-built AI infrastructure. That pivot proved prescient. As Meta, Microsoft, and other tech giants soak up available cloud capacity from Amazon Web Services and other hyperscalers, newer AI companies have been forced to look elsewhere. Fluidstack stepped into that gap.
The $50 billion Anthropic deal isn't just about raw compute. It includes the construction of new datacenter facilities, procurement of next-generation GPUs, and long-term operational agreements. For Fluidstack, it provides the kind of revenue visibility that justifies sky-high valuations. For Anthropic, it's insurance against getting priced out or rationed by larger cloud providers who are also building competing AI models.
Investor appetite for AI infrastructure deals has remained remarkably strong even as other sectors cool. CoreWeave, another GPU-focused datacenter operator, went public earlier this year at a $19 billion valuation after raising billions in venture and debt financing. Lambda Labs and other specialized providers have also raised substantial rounds. The common thread: guaranteed demand from AI companies desperate for compute.
The timing of Fluidstack's raise is particularly notable. While venture funding overall has contracted from 2021 peaks, infrastructure deals tied to AI continue to command premium valuations. Investors are betting that whoever controls the compute capacity controls access to the AI economy. It's a familiar dynamic, cloud infrastructure providers like AWS proved enormously valuable by becoming the backbone of the internet economy a decade ago.
But there are risks. Datacenter construction is capital-intensive and takes years to complete. GPU shortages remain acute, with lead times stretching months even for well-funded buyers. And if AI development hits an unexpected plateau or efficiency improvements reduce compute requirements, companies sitting on billions in datacenter obligations could face serious overcapacity.
For now, though, the momentum is undeniable. Anthropic's willingness to commit $50 billion suggests the company expects its compute needs to grow exponentially. And Fluidstack's ability to more than double its valuation in a matter of months shows that investors are willing to bet big on that thesis. The round, if it closes at the rumored terms, would rank among the largest venture deals of 2026 and position Fluidstack alongside CoreWeave as one of the most valuable independent AI infrastructure providers.
Neither Fluidstack nor Anthropic responded to requests for comment on the reported funding talks. But the deal terms, if accurate, would represent a validation of the strategy of building specialized infrastructure for AI workloads rather than relying on general-purpose cloud providers. As the AI arms race intensifies, expect more companies to follow Anthropic's playbook: lock down compute capacity now, worry about costs later.
Fluidstack's reported $18 billion valuation is more than just another big number in venture capital. It's a signal that the AI infrastructure layer has become as valuable as the AI models themselves. As Anthropic, OpenAI, and others race to build more capable systems, the companies that control access to compute are emerging as critical gatekeepers. Whether Fluidstack can execute on its $50 billion Anthropic commitment will determine if this valuation proves visionary or overheated. But right now, in a market starved for GPU capacity, being the company that can deliver tens of thousands of chips on demand makes you worth your weight in silicon.
