Energy Vault, Inc. has commenced construction on a powered AI infrastructure campus in Snyder, Texas, marking the first physical deployment phase of its broader strategy to serve artificial intelligence and high-performance computing customers through modular data center infrastructure.

Project Scope and Initial Deployment

The Snyder AI campus represents the first phase of Energy Vault's development strategy for the site. The initial build is designed to deliver 8 megawatts of powered shell capacity, with a second phase planned to expand that footprint to 25 megawatts.

The company has set a target of commercial operation in the first quarter of 2027 for the initial deployment. Energy Vault, listed on the New York Stock Exchange under the ticker NRGV and headquartered in Westlake Village, California, describes itself as a global energy infrastructure company supporting grid reliability and next-generation AI and high-performance computing infrastructure.

The company has a longer-term ambition for the Snyder site, with planned capacity expansion up to 500 megawatts and annual revenue per megawatt in line with prior guidance.

Crusoe Spark Modular Data Centers at the Core

The technical foundation of the Snyder AI campus rests on the Crusoe Spark modular data center platform, developed by Crusoe Cloud.

Energy Vault is purchasing ten Crusoe Spark modular data center units from Crusoe, which Crusoe will then lease back from Energy Vault as powered shell infrastructure.

Crusoe Spark will deploy Crusoe Cloud computing capacity for training, fine-tuning, and inference for customers.

Construction activities currently underway at the site include fiber installation, site preparation, power infrastructure, and enabling works required to support the deployment of the Crusoe Spark units.

The platform is designed for rapid deployment, with installation and commissioning timelines measured in weeks rather than the years typically associated with traditional data center infrastructure.

The campus has been designed to combine powered land, electrical infrastructure, N+1 redundancy, battery energy storage systems, and the modular data center infrastructure provided by Crusoe Spark.

N+1 redundancy means the system is built with one additional component beyond what is required for normal operation, so that if one element fails, operations continue without disruption.

Commercial Agreement and Business Structure

The groundbreaking follows a previously announced commercial agreement between Energy Vault and Crusoe.

The structure of the deal illustrates what Crusoe describes as the business opportunities enabled by vertical integration, combining Crusoe Spark physical capacity with Crusoe Cloud offtake commitments.

An offtake arrangement in this context refers to a contract in which a buyer commits in advance to purchase a company's future output, providing predictable revenue and reducing the risk that infrastructure capacity will go unused.

Marco Terruzzin, Chief Commercial and Development Officer of Energy Vault, framed the announcement explicitly around pace of execution.

"Earlier this year, we announced our commercial agreement with Crusoe.

Today, we are building it," Terruzzin said.

"Our initial 8MW deployment is targeted for commercial operation in Q1 2027. Energy Vault is delivering AI infrastructure by combining land, power, redundancy, controls, and modular infrastructure to help AI compute customers move from concept to capacity faster."

Terruzzin also highlighted the company's relationship with the local government, noting that community engagement and responsible site development were intended to create lasting local benefits for the City of Snyder.

Crusoe's Strategic Perspective

Cully Cavness, President, Co-founder and Chief Strategy Officer of Crusoe, described the Crusoe Spark modular data center program as both a capacity accelerant and a strategic differentiator for his company and its partners.

Cavness pointed to the Snyder AI site as a demonstration of a repeatable and scalable model for the rapid deployment of modular computing capacity at powered land sites.

"We are excited to be working with Energy Vault at their flagship Snyder AI site, which will demonstrate a repeatable and scalable model for the rapid deployment of modular computing capacity at powered land sites," Cavness said.

Addressing AI Infrastructure Bottlenecks

Energy Vault has positioned the project as a direct response to deployment bottlenecks that AI and cloud customers are encountering in the current infrastructure market.

The company identified land availability, power access, reliability, redundancy, speed-to-power, and execution as the critical constraints its model is designed to address.

The company states that by combining its development capabilities with powered land, electrical infrastructure, on-site power, battery energy storage, and modular data center infrastructure, it is offering a differentiated path for AI compute customers seeking faster routes from planning to operational capacity.

The Snyder AI campus is described by Energy Vault as its flagship site for this approach, and the company has indicated it views the project as a demonstration of a model it intends to replicate and scale further.