The data centre industry is at a turning point. What was once driven mainly by scale and cost efficiency is now being reshaped by artificial intelligence, power availability and network connectivity.

Demand for computing is growing faster than ever, but success is no longer just about building larger facilities. Operators must also consider where power is available, how data moves and how networks of data centres, fibre and edge infrastructure can work together.

Marc Ganzi, CEO of DigitalBridge sees this as a structural shift, not a temporary trend.

Instead, it is a force that is transforming every layer of the digital infrastructure ecosystem, as a result the industry is being forced to rethink old assumptions. Planning for speed, resilience and sustainability is now just as important as planning for scale.

Read more:  DigitalBridge CEO Marc Ganzi to deliver keynote at 25th Metro Connect

The AI super-cycle is far from over

According to Ganzi, the most important force shaping data centres today is not cyclical, but structural. As a result, demand driven by artificial intelligence continues to accelerate rather than plateau.

“The AI infrastructure super-cycle will continue, which will be reflected in the sustained build-out of capacity for training and increasingly for inference at the edge,” he says.

This distinction between training and inference is becoming central to data centre strategy. Large-scale training clusters continue to require massive, power-dense campuses, often built for a single hyperscale customer. At the same time, inference workloads are increasingly moving closer to end users, enterprises and devices, changing both the geography and the design of compute infrastructure.

This shift is occurring as long-haul connectivity regains importance.

Ganzi explains: “Connectivity will re-emerge as a strategic asset, as hyperscalers and AI platforms seek greater diversity and sovereignty in global data movement.”

Alongside this, while higher interest rates have changed financing conditions, they have not reduced appetite for digital infrastructure.

“We will see continued capital innovation, including securitisation, private credit and hybrid structures that are unlocking large-scale projects in an environment where traditional cost of capital has increased,” he notes.

However, Ganzi emphasises that capital alone does not resolve today’s biggest constraints, particularly around power availability and permitting.

AI is redefining the data centre

Ganzi notes that AI workloads are reshaping how data centres are built, operated and run.

“In data centres, power-dense AI campuses are driving larger site acquisitions, grid-adjacent strategies and new partnerships with utilities and renewable providers. At the same time, modular builds and advanced cooling are becoming more important to manage density and sustainability targets,” he says.

Liquid cooling, prefabricated modules and more sophisticated power management systems are moving from niche solutions to mainstream requirements, particularly for AI-heavy deployments.

However, data centres do not operate in isolation. As power availability reshapes where data centres can be built, network infrastructure is being forced to follow.

“In fibre and subsea, dark fibre and new routes are becoming critical to unlocking AI growth,” he says. “This includes not only connectivity to existing hubs, but also new corridors and landing points closer to renewable generation.”

AI also remains “the unifying demand driver”, tying all layers of infrastructure together.

“As more inference moves closer to users and devices, we expect a three- to five-times increase in mobile and edge traffic over the next few years, requiring investment across towers, small cells, fibre backhaul and edge data centres.”

This comes as last year, the growing importance of AI inference at the edge, was a key lesson for Ganzi.

“AI inference at the edge was a major lesson. As models became more efficient and devices more capable, it became clear that regional data centres, wireless sites, and fibre aggregation points will play a larger role in compute distribution,” he says.

Alongside this, efficiency-driven technologies are proving to be just as critical as headline features. “We also learned that efficiency-driven technologies such as advanced cooling, smarter power management, and network automation often create more value than headline features because they directly affect operating costs, sustainability, and utilisation.”

On the industry misunderstanding AI

“One misunderstood idea is that AI is only a data centre story,” Ganzi states.

“AI is a network story that stresses every layer of infrastructure, from subsea cables to street-level small cells. Underinvesting in any layer creates bottlenecks that limit overall value,” he notes.

“Another misconception is that AI infrastructure can scale without rethinking power and sustainability. The reality requires disciplined, long-term coordination with utilities, regulators, and communities.”

While this process takes time and effort, Ganzi argues that it is ultimately “where durable returns are created”.

Read more: AI now lies, denies, and plots: OpenAI’s o1 model caught attempting self-replication

Power as the defining variable for 2026

If AI defined the last two years, Ganzi believes power will define the next phase.

“The defining trend for 2026 will be power-informed deployment of digital infrastructure. Site selection, fibre builds, and M&A will increasingly start with one question: where can we secure long-term, low-carbon power at scale, and how do we move data efficiently from there?” he notes.

This shift is already redrawing the map. As a result, new AI regions and data centre clusters will “emerge in markets that were not historically considered tier-one, driven by power availability and policy alignment”.

“Secondary and tertiary markets with strong renewable resources are attracting unprecedented interest,” he explains.

This shift is not just about new locations for data centres, but it is also reshaping the fibre sector that connects them.

“The fibre industry is in both a rationalisation and acceleration phase. On one hand, overbuilt or sub-scale networks are consolidating. On the other, demand for high-capacity routes to AI data centres and edge locations continues to grow,” Ganzi says.

“We are seeing a shift toward open-access and wholesale models, with infrastructure owners partnering more closely with cloud and AI players through long-term capacity commitments.

“This makes fibre a more strategic, contracted asset rather than a commodity,” he concludes.

Ganzi will be speaking and sharing further insights at Metro Connect 2026 on 23 – 25 Feb in Ft Lauderdale with 4,000+ executives across digital infrastructure and finance.

RELATED STORIES

DigitalBridge CEO Marc Ganzi to deliver keynote at 25th Metro Connect

Marc Ganzi: ‘Power, not GPUs, is our biggest challenge’