<\/p>\n
<\/p>\n
<\/p>\n
\n 67 percent<\/strong> of all AI workloads are now run outside the cloud. 88 percent<\/strong> of companies run at least one AI workload on-premises. It\u2019s not about leaving the cloud entirely\u2014it\u2019s about hybrid AI<\/strong>, where control, security, and intellectual property are best protected on-premises.\n <\/p>\n<\/div>\n Target Audience:<\/strong> CTO, Infrastructure Manager, AI Lead | Reading Time:<\/strong> 6 minutes<\/p>\n TL;DR:<\/strong> 67 percent of all AI workloads are now running outside the cloud. 88 percent of companies run at least one AI workload on-prem. It\u2019s not about leaving the cloud entirely\u2014it\u2019s about hybrid AI, where control, security, and IP are best protected on-premises. GPU clusters based on the NVIDIA DGX platform, such as NVIDIA DGX systems, are becoming strategic investments, not experiments. <\/p>\n Three years ago, the answer was obvious: AI runs in the cloud. AWS, Azure, and GCP offer on-demand GPU instances, and startups built their entire businesses around API calls to OpenAI and Anthropic. <\/p>\n But in 2026, the tide has turned. Dell Technologies World in May presented figures that silenced the audience: 67 percent of all AI workloads are now running outside the public cloud<\/strong>. Four out of five companies\u201488 percent\u2014run at least one AI workload on-premises, in a colocation facility, or in an edge environment. <\/p>\n This is not a minor shift. It is a fundamental reevaluation of how AI infrastructure should be built. <\/p>\n There are five driving forces behind this shift:<\/p>\n GPU hours in the cloud are expensive. An NVIDIA H100 instance on AWS costs about $4\u20135 per hour. For a training job that requires 1,000 GPU hours per day, the annual cost can easily reach $1.5\u20132 million\u2014just for compute. <\/p>\n With an on-premises NVIDIA DGX H100 system, the break-even point is often reached in 18\u201324 months for constant workloads. Over 5\u20137 years, the difference amounts to several million dollars. <\/p>\n Nordic companies in the finance, healthcare, and public sectors have strict requirements regarding where data may be stored and processed. The GDPR, the Patient Data Act, and defense-related confidentiality regulations mean that sensitive data cannot leave the country\u2014or even the organization\u2019s own data centers. <\/p>\n The cloud can solve this with region-specific instances, but you never have the same level of control as when you own the infrastructure yourself.<\/p>\n When training an AI model on proprietary data, the model itself is a competitive advantage. Allowing it to be trained in a shared cloud environment\u2014where the cloud provider could theoretically observe or replicate patterns\u2014is a business risk that more and more companies are refusing to take. <\/p>\n The cloud is shared. Other users’ workloads affect yours. Latency varies. Bandwidth is not guaranteed. <\/p>\n “On-prem” means dedicated GPUs, dedicated network bandwidth, and full control over the entire stack. For real-time inference\u2014especially agent-based AI that interacts with internal systems\u2014this is crucial. <\/p>\n Cloud environments lock you into a vendor\u2019s ecosystem. On-premises solutions based on open standards allow you to switch vendors, migrate workloads, and maintain control over your technology roadmap. <\/p>\n
\n
\nWhat’s happening with AI and the cloud right now?<\/h2>\n
\nWhy are companies moving AI from the cloud to on-premises?<\/h2>\n
1. Costs Are Skyrocketing in the Cloud<\/h3>\n
2. Data Residency and Regulatory Requirements<\/h3>\n
3. Intellectual Property Protection and Competitive Advantage<\/h3>\n
4. Predictable performance<\/h3>\n
5. Scalability Without Lock-in<\/h3>\n
\nDell, IDC, and Gartner: What Do the Numbers Say About 2026?<\/h2>\n