Organisations are increasingly recognising they already possess the data needed to realise AI’s transformative potential. But, the challenge is no longer data acquisition. The focus is now on activating enterprise data through continuous, governed pipelines that power every stage of the AI lifecycle, from model training and fine-tuning to inference, automation, and advanced analytics.
This is where the AI factory needs to become a foundational layer of the emerging AI infrastructure landscape. AI factories provide the capabilities needed to operationalise and scale AI across the enterprise. In the words of Nvidia CEO Jensen Huang, this marks a fundamental shift in how organisations build and deploy intelligent systems: “AI is now infrastructure, and this infrastructure, just like the internet, just like electricity, needs factories,” he said, at last year’s Computex event.
From AI experimentation to industrialised intelligence
An AI factory is a unified, automated operating model that enables organisations to build, deploy, and scale AI through a fast, repeatable, and above all trusted process. It integrates the core capabilities required to operationalise AI at enterprise scale, including data pipelines, MLOps, infrastructure orchestration, governance, security, and continuous monitoring, into a single, end-to-end platform.
At the heart of a successful AI factory is a trusted data foundation. It provides the operational backbone that ensures data quality, availability, governance, and secure access across complex hybrid and multi-cloud environments. With this foundation in place, organisations can move beyond isolated proofs of concept to industrialise AI, scaling trusted, enterprise-wide capabilities that deliver measurable business outcomes with speed, resilience, and confidence.
Compute powers AI. Data creates value. As organisations move from AI experimentation to enterprise-scale deployment, the limiting factor is no longer GPU capacity. Rather, it’s the ability to continuously provide trusted, resilient data. Recent innovations across the infrastructure ecosystem have cemented the AI factory as the blueprint for enterprise AI. Yet despite the focus on accelerated computing, the true differentiator is the trusted data infrastructure that keeps AI running reliably, securely, and at scale.
From compute capacity to AI capability
Organisations are investing millions in AI infrastructure, but GPUs create value only when they have continuous access to the right data. GPU starvation has become one of the costliest (and still least visible) barriers to AI at scale. The initial instinct is often to add more compute, yet the true bottleneck typically lies in the underlying data infrastructure: fragmented storage, disconnected platforms, inefficient data movement, and inconsistent governance that leave high-value AI resources underutilised.
The AI factory overcomes these challenges by treating data as a continuously flowing production asset, not something confined to disconnected systems. Rather than stitching together isolated AI tools, organisations are creating integrated pipelines that connect data ingestion, engineering, analytics, model development, and inference into a single operational framework. This shift transforms AI from a series of projects into an enterprise capability. In the end, AI factories are built on compute resources that consist of powerful GPU clusters and associated CPUs, but they succeed because of data infrastructure.
Making data work harder for AI
Every unnecessary copy creates additional cost. To overcome this challenge, organisations are increasingly shifting toward architectures that bring AI workloads closer to the data rather than continuously moving petabytes between fragmented systems. By reducing data movement, enterprises can improve performance, simplify governance, accelerate model development, and maintain a single trusted source of truth. This data-centric approach is critical to the AI factory model, ensuring GPU resources remain productive, infrastructure operates efficiently, and AI investments deliver greater business value.
Trust is now as critical as performance
While GPU efficiency remains a central focus of AI discussions, organisations are facing a parallel challenge: protecting sensitive data from growing threats such as ransomware, corruption, and unauthorised access. As enterprises deploy AI against sensitive operational data, intellectual property, healthcare records, financial information, and public sector datasets, security and governance become fundamental design considerations. The question is no longer just, “Can we run AI?” It is, “Can we run AI with the confidence that our data remains secure, governed, and compliant?”
As generative AI becomes a core component of enterprise operations, organisations require confidence that their data remains secure, controlled, and governed throughout its lifecycle. This means having complete visibility into where data resides, who can access it, how it is protected, and how policies are consistently applied across hybrid environments.
Trust in the AI factory extends beyond security alone. It requires a data foundation where the information powering AI models is accurate, protected, and governed by design, regardless of where data is stored or where workloads run. As AI moves from experimentation into mission-critical decision-making, governance must become an integral capability embedded across the entire data pipeline, not a control applied after the fact.
Sovereign infrastructure enables enterprise AI
Many enterprises are rethinking the assumption that public cloud is the default destination for every AI workload. Increasingly, hybrid architectures are becoming the foundation for enterprise AI, enabling organisations to run data and workloads in the environments best suited to their business, regulatory, and performance requirements: whether on-prem, in sovereign cloud environments, or across multiple cloud providers. This approach provides the flexibility to scale AI while maintaining consistent governance, security, and operational control.
Data sovereignty extends far beyond regulatory compliance. It is a dynamic, strategic capability that allows organisations to retain control over one of their most valuable assets while maintaining the flexibility to deploy AI where it delivers the greatest business value. Whether data resides on-prem, in sovereign cloud environments, or across multiple cloud providers, enterprises need a consistent approach to security, governance, and management across the entire AI factory.
A sovereign AI strategy is about creating the trusted foundation that enables innovation at scale. By allowing governance policies to follow the data wherever it resides, organisations can build AI environments that are secure, flexible, and aligned with business, regulatory, and operational requirements.
Cyber Resilience: A core capability of the AI factory
AI systems are only as trustworthy as the data they consume, making data protection and integrity essential to successful AI operations. If training datasets are corrupted, encrypted by ransomware, or unintentionally modified, the impact can extend from degraded model performance to compromised business decisions.
For this reason, cyber resilience must be built into the foundation of AI infrastructure. Immutable storage, ransomware protection, and rapid recovery capabilities ensure that the data powering AI remains protected, recoverable, and trustworthy throughout its lifecycle. In the AI factory, resilience is not an optional safeguard, it is a core capability that enables organisations to operate AI with confidence. Resilient infrastructure provides the foundation for AI factories, enabling organisations to innovate at scale with confidence.
The future AI factory is built around data
The excitement surrounding AI factories is well deserved. They mark a shift from isolated AI experimentation to industrialised, repeatable AI operations that can scale across the enterprise. The organisations achieving the greatest impact are those establishing continuous data pipelines, enabling enterprise data to be ingested, prepared, governed, analysed, and transformed into actionable intelligence with minimal friction. Yet an exclusive focus on GPUs risks missing the larger opportunity: AI success depends not only on accelerated compute, but on the trusted data infrastructure that enables it.
Compute resources will continue to matter, but the next phase of AI advantage will be determined by the quality of the data infrastructure beneath it. Organisations that succeed will be those that build AI factories on trusted foundations: agile architectures that combine high performance with governance, resilience, operational simplicity, and sovereignty.
Paul Speciale is chief marketing officer at Scality.
S 004