Nscale’s AI Ambition: Why Data Infrastructure Matters More Than GPUs

According to DCD, Nscale has selected Vast Data’s AI Operating System to support its massive GPU expansion plans, including a previously announced deal to provide Microsoft with 116,000 Nvidia GB300 GPUs across clusters in Portugal and Norway. The neocloud company plans to scale to 300,000 Nvidia Grace Blackwell GPUs globally and recently secured significant funding with $1.1 billion in September 2025 and $433 million in Series C SAFE funding in October. Nscale’s chief revenue officer Tom Burke emphasized that the partnership aims to create “an integrated, globally distributed AI cloud” by simplifying data access across continents while maintaining the lowest compute costs through vertically integrated infrastructure. This strategic move signals a fundamental shift in how AI infrastructure is being architected at scale.

The Quiet Revolution in AI Infrastructure

While much of the AI industry focuses obsessively on GPU counts and compute power, this partnership reveals a more sophisticated understanding of what actually enables successful AI deployment at scale. The bottleneck in modern AI isn’t just computation—it’s data movement, storage efficiency, and orchestration across distributed environments. Vast Data’s approach of using off-the-shelf hardware with specialized software represents a maturation of the AI infrastructure market, where optimized data flow becomes as critical as raw processing power. This evolution mirrors the broader transition in cloud computing from simple resource provisioning to integrated platform services.

The Vertical Integration Playbook

Nscale’s strategy of building “from the ground up the underlying data center, to the GPU, to the software orchestration layer” represents a significant departure from traditional cloud providers who often rely on third-party infrastructure. This vertically integrated approach could provide substantial cost and performance advantages, but it also carries significant execution risk. Building custom data centers requires massive capital expenditure and deep expertise in power management, cooling, and physical infrastructure—areas where many pure-play software companies have struggled. The success of this strategy will depend on Nscale’s ability to manage complexity across multiple domains while maintaining the reliability expected from enterprise AI infrastructure providers.

Redefining the Cloud Competitive Landscape

This partnership positions Nscale and Vast Data as challengers to the hyperscale dominance of Microsoft, Google, and Amazon in the AI infrastructure space. By focusing specifically on AI workloads rather than general-purpose cloud computing, they’re carving out a specialized niche that could appeal to enterprises with massive AI training requirements. The timing is strategic—as companies move from experimental AI projects to production deployments, they’re discovering that generic cloud infrastructure often falls short for demanding AI workloads. The combination of Nscale’s GPU scale and Vast Data’s data architecture could create a compelling alternative for organizations building the next generation of AI applications.

The Technical Implications of Data-Centric AI

The integration of Vast Data’s platform with Nvidia BlueField-3 DPU technology represents a sophisticated approach to solving one of AI’s most persistent challenges: data movement bottlenecks. By disaggregating storage from compute and optimizing data flow at the hardware level, this architecture could significantly reduce the time AI models spend waiting for data rather than processing it. This becomes increasingly important as model sizes grow and training datasets expand into the petabyte range. The ability to maintain consistent performance across distributed global clusters could become a key differentiator for enterprises running continuous training pipelines across multiple geographic regions.

Market Outlook and Strategic Implications

The substantial funding rounds—$1.1 billion followed by $433 million—indicate strong investor confidence in Nscale’s approach, but also highlight the capital-intensive nature of building AI infrastructure at this scale. The success of this partnership will likely influence how other specialized AI infrastructure providers approach the market. If Nscale can demonstrate superior performance and cost efficiency compared to hyperscale providers, we may see more vertically integrated specialists emerging to serve specific AI workload categories. However, the path forward is fraught with challenges, including managing global operations, maintaining competitive pricing, and staying ahead of rapidly evolving AI hardware and software stacks.