Oxide Computer Raises $445M Series D as AI Demand Pushes Enterprises Back to Owned Infrastructure
Oxide Computer Company raised $445 million in a Series D on October 9, 2026, led by Eclipse. Existing investors US Innovative Technology Fund (USIT), Riot Ventures, and Jane Street participated. The Emeryville, California company builds full-stack cloud infrastructure that enterprises own and operate in their own facilities — an architectural choice that has moved from niche to mainstream as AI workload costs have made public cloud unit economics hard to justify at scale.
The round is the largest in Oxide’s history and comes after the company reached profitability, an unusual milestone for infrastructure hardware startups.
The On-Premises Resurgence
Oxide competes against the dominant model of renting compute from hyperscalers. Its pitch is that large AI training and inference workloads have changed the calculus: at sufficient scale, owned infrastructure is cheaper, more predictable, and keeps data and model weights behind enterprise firewalls.
The market argument has sharpened as GPU costs have risen. Enterprise AI teams running persistent inference workloads on reserved or on-demand cloud instances increasingly hit a point where a capital expenditure in owned rack hardware pays back in 18 to 24 months.
Oxide’s hardware is designed to look and feel like public cloud from a software and operations standpoint — APIs, orchestration, and tooling consistent with what developers expect from AWS or GCP, but running on hardware the customer controls.
What the Investors Are Betting On
Eclipse, which led the round, has a track record in industrials and deep tech infrastructure. Jane Street’s participation carries a different signal: the firm’s involvement typically reflects conviction in the unit economics model at scale, not just the technology thesis. USIT involvement points to a sovereign and defense compute angle — on-premises infrastructure that does not route through non-U.S. data centers is increasingly required for government and defense AI contracts.
The $445M gives Oxide the balance sheet to accelerate manufacturing capacity, expand sales into the enterprise accounts that are actively modeling cloud repatriation, and build out the software layer that makes owned hardware competitive with managed cloud on operational complexity.
Context
AI infrastructure investment in 2026 has split into two visible tracks: continued massive capex at hyperscalers building out GPU clusters for model training, and a parallel build-out of on-premises and sovereign compute for inference and data-sensitive workloads. Oxide’s raise is the largest single data point yet for the second track.