Tania Velarde has watched the data center change shape in her 25 years in the industry. The biggest change is happening right now.
“AI is no longer just another workload. It is becoming a foundational pillar of modern data center strategy, driving the rise of AI-native and NeoCloud infrastructure.” said Velarde, who leads partnerships and alliances at T-Systems North America ,an IT service provider. And the data centers themselves are multiplying: “The Giga factories are exploding around the globe.”
Her colleague Alex Martinez, who leads sales for T-Systems in the U.S. and Canada, sees where this is going next. “Our managing operations are moving now to support AI,” he said. “Our future as part of AI is based on an industrial AI cloud.”
Martinez and Velarde’s perspectives are in line with market predictions. According to Mordor Intelligence, the AI factory infrastructure market will be worth almost three times as much ($782B) in five years as it is now ($266.8B). This is in part due to the shift from standard data center layouts toward AI-native facilities, as well sovereign compute rules and data residency requirements that make hybrid deployments more relevant.
But factories and industrial clouds are only as good as the data that feeds them, and that’s where the challenge for this vision of the future emerges: while AI is moving fast, enterprise data is not.
According to proprietary data from Hitachi Vantara, organizations with successful AI project outcomes are twice as likely to cite data quality as a key component than organizations that struggle to find AI success. Fragmented data, ungoverned pipelines, and infrastructure built for outdated workloads are often why AI projects stall before production.
Whether an organization is looking at the AI factory or the AI industrial cloud, the journey toward success is clear: a data foundation that’s built for today, and flexible enough to handle tomorrow.
What a Factory-scale Foundation Requires
In mid-October, NVIDIA published guidelines on how large AI facilities should behave during utility grid disturbances. According to the guidelines, facilities should stay connected, respond automatically, and return to near-full load in under two seconds. Such guidance is directional for the market and the emerging AI factories category, and Velarde agrees.
“Complex businesses rely on infrastructure to support digital and critical operations,” Velarde said, which means the foundation has to deliver on three fronts: “You need reliability. You need availability. And you need resilience of your data.”
Martinez adds that storage alone no longer cuts it. “Data and moving data is not about just storing it anymore,” For T-Systems, which runs mission-critical applications for its customers, the stakes are plain: “The core of the business is going to be basically based on a solid foundation of data.”
The same logic applies to the industrial cloud layer. Martinez says if an organization has the right technology and the right infrastructure underneath its IT and managed services, replication is simple.
Work Backwards to Feed the Factory
A traditional factory starts with what it needs to produce, then works back to the raw materials. AI works the same way.
"I think first [organizations] need to identify what is the value that they need to get out of AI, Martinez said. From there, identify the data AI will consume, then the governance that data requires, then the level of reliability and security it needs.
A common objection to governance is that it slows innovation. But Martinez sees it the other way around, noting that without control of data, initiatives will actually slow down. And skipping the remaining steps will amplify the weaknesses of the information.
Once the foundation is in place, the payoff is freedom. When trust is designed in from the start, organizations don’t need to worry about adding additional layers of security and reliability, and engineers can work without rebuilding the foundation each time.
Building a Foundation for What’s Next
What makes the data foundation work — and what creates a path toward an AI industrial cloud — is not novelty, but dependability.
The AI factory and the AI industrial cloud represent two facets of the same future: AI capacity that can operate at massive scale, ensure dependability, and deliver managed operations. Without a reliable data foundation, this future is impossible.
Whether it’s a GPU-dedicated factory or an industrial cloud running managed operations, the infrastructure underneath sets the bar for what AI can deliver.
Learn more about VSP One, the data foundation that keeps pace with your AI ambition.
Jeff Lundberg
Jeff Lundberg is Principal Product Marketing Manager for the Hitachi iQ solution portfolio at Hitachi Vantara.