Hammerspace Posts Nearly 14x YTD Bookings Over Full-Year 2025 as AI Shifts to Inference and Physical AI

Hammerspace, an high-performance data platform for AI, announced exceptional business momentum as the market moves into a new phase – shifting from building models to operationalizing AI across enterprise and government environments.To scale inference and prepare for physical AI, organizations need platforms that can make existing infrastructure AI-ready now. That urgency is driving strong adoption of Hammerspace’s data platform across global enterprises running training and inference across data centers and clouds, government and sovereign AI initiatives, neocloud environments and large-scale cloud deployments.

“AI has moved beyond model building,” said David Flynn, founder and CEO, Hammerspace. “The next phase is operational: inference at scale, faster time to first token, lower cost per token, and readiness for physical AI. Hammerspace turns the infrastructure customers already own into the platform they need to win.”

Hammerspace’s momentum reflects a simple market reality: organizations cannot wait for infrastructure shortages to ease, for new capacity to arrive, or for copy-based architectures to catch up. They need to move now with the storage, GPU servers, data centers and cloud resources they already have.

From hyperscalers like Meta to Vanderbilt ACCRE, and from the US government to sovereign cloud, neocloud and high-frequency trading environments, organizations are choosing Hammerspace because they need both: seamless access to distributed data across hybrid environments and the high-performance data delivery GPUs require at scale. That combination is becoming essential across training, inference, physical AI and ultra-low-latency data-intensive workloads – and is driving strong global growth for Hammerspace.

Hammerspace Growth Drivers

• Bookings in 2026 are already up nearly 14x vs. full-year 2025. Customers are now moving to make existing infrastructure AI-ready rather than waiting for ideal future buildouts
• Massive Time-To-First-Token (TTFT) Advantage: Seconds to First Token, Not Days, Weeks or Months. Hammerspace delivers a major time-to-first-token advantage by making distributed data usable in place on the infrastructure customers already own, where it already lives, eliminating the delay of copying, migrating and staging data in a new AI infrastructure silo before inference can begin
• Global enterprise growth across training and inference. Hammerspace is gaining traction with global enterprises unifying data across data centers and the cloud to accelerate AI without new silos or disruptive migrations
• Government and sovereign AI adoption. Demand continues to grow across government and sovereign AI initiatives, where organizations must operationalize AI quickly while maintaining control of distributed data
• Neocloud adoption is accelerating. Government and public neocloud providers are choosing Hammerspace for GPU-scale performance and direct access to enterprise data without an additional copy or another silo
• OCI growth at a global scale. Demand is growing for Hammerspace on Oracle Cloud Infrastructure to support large-scale enterprise AI and globally distributed workloads across cloud, hybrid and sovereign environments
• Strategic partnership momentum with Hitachi Vantara. The Hitachi Vantara partnership is expanding Hammerspace’s reach through integrated solutions for hybrid cloud and AI infrastructure
• Expanded go-to-market with Supermicro. The Supermicro partnership is accelerating adoption through pre-qualified infrastructure that simplifies deployment and shortens time-to-value
• Strong renewals and customer satisfaction. Gross retention above 95% and Net Promoter Score (NPS) reaching 71, signifying strong customer loyalty and satisfaction, and Hammerspace’s growing strategic importance as customers move from AI pilots to operational AI

Built for AI Anywhere, From Training to Physical AI
AI now demands a platform that can do more than support isolated stages of training, inference and physical AI, or small, curated data sets designed for proof-of-concept work. It requires a platform that can use existing infrastructure immediately, make enterprise data AI-ready, and eliminate the delay, cost and complexity of copy-based architectures.

Hammerspace’s data platform is built to accelerate training with high-speed access to large, distributed datasets, efficient data movement to GPUs and the flexibility to use storage, GPU servers, data centers, cloud and neocloud resources already in place. At the same time, it delivers the performance, efficiency and data accessibility organizations need to operationalize inference and prepare for physical AI at scale.

Hammerspace is being selected at a rapid pace because its technology uniquely addresses the needs of this moment, helping organizations move faster across training, inference and physical AI with their existing infrastructure and their preferred cloud providers.