The two companies announced a strategic partnership on Monday to deliver what they describe as a unified AI factory: a production environment where data moves continuously from ingestion through to model training and inference without the bottlenecks that leave GPU clusters starved.
GPU starvation occurs when compute hardware runs ahead of the data pipeline. Accelerators optimised for sustained parallel throughput sit idle while storage systems catch up, or while data preparation stages complete. In environments running large-scale training or continuous inference, those idle periods translate directly into wasted capital. The partnership targets that gap by combining Cloudera's containerised data services with VAST Data's AI Operating System.
Cloudera contributes its next-generation lakehouse architecture, including data engineering, streaming, analytics, machine learning, and AI services, all deployable consistently across on-premises data centres, private cloud, and public cloud. VAST Data provides the storage and compute fabric underneath, built on its Disaggregated Shared Everything architecture, which combines high-performance storage, a vector database with NVIDIA cuVS acceleration, and a global namespace that can scale to exabytes. The reference design follows the NVIDIA AI Data Platform specification, making the combined stack compatible with NVIDIA AI Enterprise and NIM microservices for model serving.
"Enterprises are investing billions in GPUs, yet many struggle to achieve full utilisation due to data bottlenecks," said Abhas Ricky, chief business officer and general manager of applied AI at Cloudera. "Our partnership with VAST eliminates GPU starvation and enables customers to build true AI factories where data flows seamlessly from ingestion to insight."
The combined customer base manages 60 exabytes of data, which the companies said creates a significant immediate addressable market for the joint solution. They intend to expand reference architectures and industry-specific validated deployment patterns through 2026.
Jeff Denworth, co-founder at VAST Data, described the central problem: "Most enterprises already have the data they need for AI. The challenge is unlocking the value in data to create a continuous pipeline of AI inference, fine-tuning, and data analysis."
The joint solution is available now through both companies' enterprise sales teams. Customers running Apache Spark workloads will also be able to leverage NVIDIA cuDF GPU acceleration on top of VAST's high-throughput data services, allowing Cloudera Data Engineering jobs to use GPU processing without pipeline changes.
For regulated industries, the combined stack is positioned as a full private AI infrastructure, covering data sovereignty requirements while delivering the throughput necessary for production-scale AI workloads.
To stay across the latest in cloud, AI and enterprise tech analysis from Compare the Cloud, subscribe to our weekly newsletter at https://www.comparethecloud.net/newsletter
