Agentic AI might be outrunning the one thing it needs most — an AI trust layer.

That challenge is reshaping the ambitions of companies long anchored in data protection. Veeam Software Group GmbH, which spent two decades building the market’s most widely adopted backup and recovery platform, is now staking out an entirely new category — data and AI trust infrastructure — at the center of the enterprise AI stack, according to Veeam Chief Executive Officer Anand Eswaran (pictured). Nvidia Inc. has the compute, Databricks Inc. and Snowflake Inc. own the data layer, Anthropic PBC and OpenAI LLC supply the intelligence — and Veeam wants the layer none of them are building.

“You have your GPUs — Nvidias of the world. You have your data layer, the companies like Databricks and Snowflake. You have the AI models which operate on it, Anthropic, OpenAI. You have your agent orchestration layers,” Eswaran said. “Sam Altman said it best. He said, ‘The problem in the bottleneck is never going to be compute or intelligence, because that is going to become a utility. It is going to be, ‘Can you trust the data feeding it?’ That is at the heart of our thesis, which is that right above the data layer and below the model layer, you need to have a data and AI trust layer.”

Eswaran spoke with theCUBE’s Dave Vellante and Krista Case at VeeamON 2026, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed Veeam’s strategic pivot to the AI trust layer, the Veeam DataAI Command Platform, agentic security and the company’s IPO outlook. (* Disclosure below.)

Veeam claims the AI trust layer for the agentic era

The strategic logic behind Veeam’s new direction traces directly to its $1.725 billion acquisition of Securiti Inc., completed in December 2025, which brought data security posture management, privacy, governance and AI trust capabilities into the fold. Veeam’s thesis surrounding the data and AI trust layer emerged from repeated conversations with customers who found that without well-governed data even the most sophisticated agentic deployments failed to deliver the intended outcomes.

“Today, there are multiple different companies who create individual products, some for data security, some for AI security, some for governance, some for privacy, some for compliance and some for resilience,” he said. “The only way this AI trust layer works is if you are able to unify all of these domains on one platform and one data fabric — one platform bringing together data security, governance, compliance, privacy and resilience together — and that’s what we announced today.”

That announcement introduced the Veeam DataAI Command Platform, described as the industry’s first unified data and AI trust infrastructure for the agentic era. At its core sits the DataAI Command Graph, built on technology from the Securiti AI acquisition, which can visualize data at a granular level — going well beyond the database or storage-bucket view that most tools offer — across more than 300 connectors spanning cloud, SaaS and on-premises environments, according to Eswaran.

“This can actually visualize data at a discrete, granular element level — not at a database [level], not at an S3 bucket level, at a granular level,” Eswaran said. “Nobody else can do that. This is the first one which has breadth. It has 300-plus connectors, which connects to every data system, AI system, identity system and agent system out there. This is the first knowledge graph in the industry which unifies multiple domains.”

The platform also addresses identity, an area inseparable from data security in the agentic era, Eswaran noted. Understanding the entitlements and policies attached to every data element — and tracking which human and non-human identities are accessing it — is what enables the precision resilience posture that Veeam is positioning as its competitive edge. That precision extends to recovery: rather than rolling back 24 hours of operations when an agent goes wrong, the DataAI Command Platform can surgically undo only the specific actions that caused harm, Eswaran explained.

“How can I trust the data feeding my AI and should something go wrong, how can I recover with precision only those five seconds which went wrong, only that one agent action which went wrong, not restore 24 hours of work?” he said. “Precision resilience — that’s the beauty of the DataAI Command Platform.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of VeeamON 2026:

(* Disclosure: TheCUBE is a paid media partner for the VeeamON event. Neither Veeam, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE