George Kurian, Chief Executive Officer of NetApp, said enterprises rushing to deploy AI should leave their data where it is. He told journalists in Abu Dhabi that the average company runs its business on just 15% of the data it holds.
Speaking at a media roundtable at AI Everything Abu Dhabi 2026 at ADNEC Centre, Kurian said that once AI runs through every part of a business, moving data back and forth between environments stops making sense. “We really think you shouldn’t move your data. You should leave your data where it is, because as AI gets pervasive through all of your business, this idea of moving data is stupid,” he said.
Kurian, who has led NetApp since 2015, said the Emirates is adopting AI faster than any other part of the world. He came to the event days after NetApp’s INSIGHT 2026 conference in Las Vegas, and is also scheduled for a fireside chat on the Summit Stage titled “Can AI Be Trusted? The New Architecture of Sovereign, Secure and Safe AI.”
Most enterprise data has never reached an AI model
Kurian said the average enterprise gives about 12% of its employees access to its data. Most of that is structured records held in ERP and CRM systems and surfaced through business intelligence tools. The other 85% of a typical organisation’s data is unstructured. He did not give a source for the figures.
NetApp’s answer is a data intelligence layer that finds and catalogues data across siloed systems. AI models work from that layer, and the underlying files stay where they are. NetApp can show customers which data is duplicated or stale and hasn’t been opened in a long time, but Kurian said the clean-up is their job.
“The responsibility on curating the data is really the customer’s,” he said. Compute has outrun the data feeding it
Asked whether attention is moving from GPUs to data infrastructure, Kurian called it a natural progression. Organisations had to train models before they could use them, he said, and the data problem now matters in almost every large organisation, both for security and for value.
He said many deployments hit “a memory wall or a data wall, where they cannot feed the compute fast enough.” NetApp says the systems it announced at INSIGHT remove that bottleneck at the scale of 100,000 GPUs, though no performance baselines were shared to back the claim.
NetApp also launched its AI Data Engine. Kurian said it applies classification, access and privacy controls so that agents acting for employees cannot misuse the data they can reach.
Sovereignty sits with the data a company is liable for
Kurian argued that the sovereignty debate focuses too much on which model an organisation chooses, and that legal liability should be what drives it, “because ultimately you’re legally liable for your data.” When every company can use the same models, he added, “data, and how you use it, is the intellectual property and the competitive moat.”
He described sovereignty as a question of control. That covers business continuity and also visibility over how data is being used. NetApp has run in disconnected data centres where control stays inside the country and with its citizens, and Kurian said its software can flag when data crosses a geographic boundary.
Companies will answer for what their AI agents do
Asked who is responsible for decisions as AI becomes more autonomous, Kurian put the responsibility on the organisation. Enterprises have always been liable for mistakes their staff make with data. If agents now replace those staff and act for them, “the enterprise is accountable,” he said.
He said security has to move as fast as the threat. “If the malicious actor is a machine, the protection also has to be at machine speed,” he said. NetApp has built deep learning models into its products to block malicious activity at the point where data is created. Kurian said the technology blocked a recent AI-powered ransomware campaign but gave no details. NetApp is also working with model and harness providers to extend the identity-based access controls on customers’ data to the AI agents that act for employees.
NetApp is counting on customers wanting cloud choice
Kurian said NetApp was the first storage provider built natively into Amazon’s cloud and is still the only one. He listed Microsoft, Google, IBM, Oracle and SAP as its other partners. Customers want to choose where their data is used, he said, and the more platforms they can connect it to, the more their data infrastructure is worth.
