Cloudera has expanded its data services portfolio with a release that takes generative AI directly into the enterprise data center. The update allows organizations to run AI securely behind their firewall with GPU acceleration, addressing one of the biggest blockers to adoption: protecting sensitive data and intellectual property.
Tackling Security and Cost Barriers
For many enterprises, concerns around data security and regulatory compliance have slowed AI projects. Research shows most organizations still lack foundational safeguards to protect critical models and data pipelines. Cloudera’s approach places AI where the data already lives, cutting out the need to move sensitive information into public clouds. That not only reduces security risks but also brings down infrastructure costs by consolidating workloads and automating complex tasks.
Leo Brunnick, Chief Product Officer at Cloudera, told ChannelE2E, the goal is to give regulated industries more control. “They can run powerful AI models securely within their own data center to address strict compliance and data privacy needs that public cloud options cannot always meet. This approach delivers major improvements in performance and scalability to allow businesses to get value from their AI projects much faster, while making better use of their existing hardware.”
He added that what makes the platform stand out is its consistency across environments: “We are effectively bringing the benefits of the cloud anywhere, whether in a private data center, the public cloud, or a hybrid environment.”
Extending Cloud-Native AI On Premises
With this release, Cloudera brings two key services into the data center. AI Inference Service, accelerated by NVIDIA, streamlines deployment and management of large-scale models directly on premises. AI Studios extends the low-code environment for building and running generative AI applications and agents without relying solely on public cloud infrastructure.
Brunnick stressed that the platform doesn’t force enterprises to choose between cloud and on-premises. “Our platform is truly hybrid, so enterprises don’t have to move from our cloud AI services to the on-premise version. A team could build a model in the public cloud to take advantage of its flexible compute, and then deploy that same model on premises to run securely against sensitive production data. The key is that the user experience is consistent across both environments, which gives teams the flexibility to choose the right location for each part of the AI lifecycle without disruption.”
Deployment Speed and Efficiency Gains
Independent research highlights measurable benefits for organizations adopting Cloudera Data Services on premises. Reported improvements include faster workload deployment, higher productivity for data teams, and significant cost savings through better hardware utilization. For enterprises under pressure to modernize, these gains translate to quicker AI adoption and reduced operational overhead.
A big part of that efficiency comes from NVIDIA integration. Brunnick pointed out that “adding NVIDIA NIM microservices is a game-changer for getting the best performance from large language models quickly and easily. Because NVIDIA NIM is already embedded in the Cloudera AI Inference service, we uniquely streamline the deployment and management of large-scale AI models. This reduces deployment time from over a week with a DIY approach to as little as five minutes, while delivering up to 36x faster performance on NVIDIA GPUs. These efficiencies directly reduce operational costs, with customers seeing estimated savings of about 54% compared to running all workloads in a non-optimized environment.”
What This Means for Enterprises
Cloudera’s move to bring private AI into the data center takes away the old choice between speed and security. Enterprises can now scale AI projects while keeping sensitive data under their own control and meeting compliance demands. For IT leaders, it marks a shift from testing AI at the edges to running it in secure, production-ready environments.
