NVIDIA Fleet Expansion as Voltage Park Becomes Reference Platform NVIDIA Cloud Partner

Voltage Park is in the process of bringing 2,000 new NVIDIA HGX™  B200 GPUs online at its data center in western Washington. Built for data scientists and developers alike, the HGX B200 systems provide customers with a massive 1.4 terabytes (TB) of GPU memory, 64 terabytes per second (TB/s) of aggregated memory bandwidth, and 14.4 TB/s of NVLink Switch bandwidth  to handle more complex AI workloads. 

This expansion comes on the heels of Voltage Park becoming one of the few Reference Platform NVIDIA Cloud Partners in the world. This recognition validates the quality, strength and scale of the infrastructure backbone of our AI Factory, and affirms our mission to make a flexible sandbox where enterprise teams can build complete, customized AI systems with a combination of models, tools, frameworks, NVIDIA accelerated computing  and orchestration layers they choose. 

As a leading neocloud, this designation further affirms our unique position as the only vertically integrated AI platform purpose-built for both enterprise and research teams with no engineering lift, vendor lock-in, or performance trade-offs. 

A new standard for production AI 

As a Reference Platform NVIDIA Cloud Partner, every layer of hardware our customers use - whether on-demand, reserved, or in the AI Factory - is validated by NVIDIA for performance, reliability, and scale. This includes the new NVIDIA HGX B200 systems and future NVIDIA GPU procurements. This allows us to consistently deliver modular, enterprise-ready solutions that move with your team’s pace and priorities. 

What this unlocks for our customers 

This isn’t just about hardware, it’s about providing the most up-to-date capabilities for founders, researchers and enterprises: 

● Access to NVIDIA’s latest architectures, including Blackwell and Hopper, with performance validated at the cluster level

● Modular, production-grade environments tuned for generative AI, vision models, and low-latency inference 

● Faster deployment cycles via NVIDIA AI Enterprise, NVIDIA NIM microservices, and pre-optimized models all integrated into our platform 

● Co-innovation pathways across NVIDIA’s developer and partner ecosystem

● A global footprint, built for regionally distributed teams and edge deployment 

This hardware advantage is only the beginning. Customers can now combine our world-class compute with the models, orchestration and software tools of their choice to build a customized AI Factory for building, training, and deploying AI systems at scale. 

AI Factory: Built for enterprises, not just engineers 

Many AI platforms are software-first and rent compute as an afterthought. We took the opposite approach by merging our owned hardware with software modularity to create a unified sandbox where developers can transform raw data - structured or unstructured - into production AI. No infrastructure gaps, no hyperscaler constraints. 

This is how we’ve surpassed competitors in fleet size - while staying radically simple to adopt. Our model is designed for enterprise teams, not just ML engineers. That means: 

● No surprise bottlenecks 

● Predictable, transparent pricing 

● Direct bare-metal access 

● Enterprise-level support with no lift from your team 

We give you control without the complexity. 

Accelerating adoption, not just experimentation 

Since day one, we’ve built for one thing: real-world deployment at scale. 

Becoming a Reference Platform NVIDIA Cloud Partner validates the AI Factory’s readiness to support enterprise-grade production workloads. This isn't about experimenting with AI. It's about moving to production faster, at lower cost, and with fewer compromises. 

Want to build AI on NVIDIA HGX B200 and take your applications to the next level? Explore more about our NVIDIA-powered infrastructure.