Key Highlights
- Qumulo introduces Cloud AI Accelerator to enable real-time access to enterprise data across cloud and on-premises systems.
- Platform aims to reduce GPU downtime caused by delays in moving and preparing data.
- It is available across AWS, Microsoft Azure, Google Cloud, and Oracle Cloud Infrastructure.
Qumulo has introduced Cloud AI Accelerator, a new platform designed to help enterprises run AI workloads more efficiently by improving how data is accessed across cloud, edge, and on-premises environments.
The company said the platform combines its Cloud Data Fabric, Cloud Native Qumulo (CNQ), and NeuralCache technologies to give businesses real-time access to distributed datasets without the need for repeated copying or long staging processes.
For many organizations investing in artificial intelligence, moving and preparing data before workloads can begin remains a major hurdle.
Qumulo believes this process is one of the reasons why expensive GPU infrastructure often remains underused.
According to industry analysis cited by the company, enterprise GPU utilization averages just around 5%, leaving significant computing power sitting idle.
Qumulo Looks to Solve Enterprise AI Data Bottlenecks
According to Qumulo, the new platform addresses what it calls the challenge of “data gravity”, the difficulty of running AI workloads when data is locked in different locations.
Instead of forcing organizations to bring workloads to where the data is stored, Cloud AI Accelerator allows enterprises to run workloads wherever GPU resources are available.
“Every enterprise we talk to is focused on GPU availability, but availability is only half the problem. The deeper issue is utilization, and the culprit is data gravity,” said Douglas Gourlay, CEO of Qumulo, in a statement.
The platform is designed to connect both cloud-native and on-premises Qumulo environments with AI services such as Microsoft AI Foundry, Amazon Bedrock, and Google Vertex AI, without requiring separate data copies.
Qumulo said this approach can help organizations reduce delays tied to data staging, minimize idle GPU time, and avoid maintaining multiple storage silos across different environments.
Broader Push Into AI Infrastructure
The launch builds on Qumulo’s broader AI infrastructure strategy. The company had previewed Cloud AI Accelerator in November 2025 as part of its efforts to better connect enterprise data with cloud-based AI computing resources.
For hybrid deployments, the platform also incorporates Cisco networking, security, and Unified Computing System (UCS) infrastructure, enabling organizations to adapt workloads based on changing GPU availability across regions and cloud providers.
Cloud AI Accelerator is available immediately across AWS, Microsoft Azure, Google Cloud, and Oracle Cloud Infrastructure, with support for hybrid deployments using Cisco UCS environments.