Kathy Gibson reports from VeeamON – After 20 years in business, 2026 may be the best forward-looking year yet for Veeam.
This is the word from Tim Pfaelzer, senior vice-president: EMEA at Veeam, speaking at the VeeamOn Tour 2026 in Riversands today.
Veeam is coming from a strong position, he adds: it is now a $2,1-billion company, with more than 550 customers in 150 countries, including 80% of the Fortune 500.
Having come through the era of backup and recovery, then cyber resilience, Pfaelzer says Veeam is now addressing the era of agentic drift.
In the new era, customers still need to have backup, they still need cyberresilience and th ability to recover – but they also need to shift to assuming drift.
This time it’s not one shift, but three, Pfaelzer says. Firstly, data gravity has flipped from structured to unstructured. “Yesterday’s apps are now systems of record.”
In the agentic era, agents reach into every system – whether customers intend them to or now. “Dorman data is now the fuel”. In fact, 230 zettabytes of date was created this year, 90% of it unstructured. And this is set to at least triple by 2029.
This gravity of data now fuels the AI that weaponizes it, Pfaelzer says.
The new failure mode isn’t a breach like companies understand it: 88% of customers are already using or piloting AI agents -but only 7% are truly AI-ready. At the same time, AI has acted on this data.
Secondly, the infrastructure to deploy AI is growing and growing large. AI infrastructure spend is at $410-billion today, set to grow to $3-trillion by 2028.
But the data foundations aren’t growing at this rate. Today, there are 82 agents for every human.
Meanwhile, risk in the agentic era obeys different laws of physics: 250 000 non-human identities per enterprise; 7 000 passwords are blocked per second by Microsoft, and 97% of agents carry excess privileges so the blast radius is larger than it needs to be.
“This isn’t theoretical,” Pfaelzer says. Breaches has wiped data – and backups – in the space of seconds.
For IT, their job has just changed. “If you have a dwell time of nine seconds, everything you are trained to do will be too late.”
Zero Trust was built for humans, Pfaelzer says – and may not hold up in the agentic AI era.
There is a layer missing today – and that is trust.
The industry is converging on this same realisation, he adds. Everything that goes into the AI produces something – so if the data is wrong, the output will be wrong.
Veeam and SecurityAI are converging to create a new category, where data and trust converge.
“This will generate that missing layer – that data and trust layer that is missing now.”
DataAI includes access and identity, blended with AI.
Veeam’s DataAI Command Graph on the Data Command platform adds a layer of resilience, complemented with a layer that includes security, governance, compliance and privacy.
Agents operating on top of this platform include services including dashboard, threat detection, graph, backup, restore and more.
“This is not a vision: it is a product available today,” Pfaelzer says.
Why the DataAI Command Graph
The typical enterprise today has hundreds of petabytes of data, held in 10s of billions or 100s of billions of files in 100s of millions tables.
And when we talk about artificial intelligence (AI), this data is key to using it effectively.
But the sprawl and sheer volume of data is a massive challenge. “We have more data than ever before, but don’t necessarily have the right data for IA use cases,” says Tahir Latif, chief trust officer: EMEA at Veeam.
And this is why a vast majority of AI projects fail, he believes.
All data is not created equal; and it Is held in a variety of different environments. Meanwhile, the agents now being used communicate with LLMs and with various system to extract data.
Traditional thinking was that the LLM was the biggest risk, Latif says. What’s changed is that the agent is now the single biggest risk. The agent can now operate at machine speed, it can act autonomously, and it can connect with multiple different systems.
“The goal can no longer be to protect the model or the prompt – because we have given agents permissions,” Latif explains. “So companies need to be thinking about what agents are able to touch; and how they can be resilient if something goes wrong.
“When you have agents engaging with other agents, it becomes even more complex – it gets exponential and the blast radius increases. You just need one agent that is over-permissioned connecting to another one that has access to tools and you create risks.
“And, unfortunately, these risks are largely invisible.”
So having security controls in place isn’t enough any more, Latif says. “The agents have potential access to all our system – and that’s where the risk lies.”
The DataAI Command Graph is a living network consisting of a company’s data, identity, AI, activity and risk.
Veeam adds powerful dashboarding that connects across the ecosystem – on-premise, Saas, cloud or multi-cloud – and to literally hundreds of different systems in an ecosystem that is growing all the time.
The tool includes data discovery and classification that offers contextual DataAI intelligence. It also links the data to the user, adding a layer of intelligence that includes what permissions a human or agent has.
The DataAI platform can assist companies in building out their data inventory, Latif points out. “You can’t protect what you don’t know. So you need to have an inventory of the agents acting on your behalf.”
Privacy is a key concern, so the platform is able to help customers with privacy compliance by automating a lot of operations.
These features are all contained in the same visibility platform. “You probably have all these tools already, but in a disparate, unconnected manner. The DataAI Platform gives customers all of these features in one place.”