DigitalBridge CEO on the AI revolution’s impact on investments

Marc Ganzi: The past year has validated our thesis—generative AI isn’t just a trend; it’s a fundamental shift in how digital infrastructure needs to be built.

What has changed is the pace of this journey. Demand has accelerated, intensifying pressure on power availability, network density, and compute capacity.

At DigitalBridge, we’ve leaned into this by scaling our data centre investments. We now own and operate seven global platforms representing more than 180 data centres, with strategic power agreements for up to 16GW of power.

We are also expanding our fibre and edge networks to handle this growing AI workload.

Ganzi: These breakthroughs reinforce the accelerating demand for AI-native infrastructure. Innovations like DeepSeek and projects like Stargate are helping push compute requirements to unprecedented levels, requiring purpose-built facilities with high-density power, liquid cooling, and ultra-low-latency connectivity.

At DigitalBridge, we remain technology-agnostic—we build infrastructure that supports the most advanced AI workloads, regardless of the specific model or architecture. Adaptability is key; our platforms are designed to scale with evolving AI demands, from training massive models to enabling real-time inference at the edge.

Ganzi: Edge computing is evolving from concept to necessity. As AI-driven applications demand real-time processing, we are seeing a clear need for smaller, high-density, low-latency data centres close to consumers and enterprises.

This is changing investment strategies—hyperscale alone isn’t enough. We need a layered approach, combining core data centres with distributed edge infrastructure. The industry is innovating with new designs, better cooling technologies, and network architectures that prioritise proximity and efficiency.

Another key consideration is power and capacity availability. Forward-looking strategies involve pre-emptive planning—securing land, energy, and network capacity well in advance to avoid bottlenecks.

Ganzi: Three major trends stand out:

First, continued AI development, which is driving much of the accelerating growth in the digital infrastructure space.

Second, energy innovation—whether through nuclear, hydrogen, or grid-interactive data centres, we’ll see new solutions to power AI at scale.

Third, the convergence of connectivity and compute—edge networks will increasingly integrate with AI to drive new use cases in automation, industrial IoT, and immersive experiences.

Ganzi: First of all, we see a step function change in demand for fibre-based networks to support both the rise of AI network architecture and the rising data needs of consumers and enterprise. The rise of AI-driven workloads is fundamentally reshaping digital infrastructure, and connectivity is a critical piece of the puzzle.

The sheer volume of data being generated, moved, and processed will dramatically increase overall network utilisation. More importantly, the extraordinary power needs of AI, will result in data centres and compute to be significantly more distributed geographically than ever before.

Our largest customers are looking to expand into areas where there is high power density and access to significant bandwidth with low latency. Right now, the demand is for very large pipes for the next generation of large AI-native data centres. We think the next trend will move to the edge where we see a requirement for more expansive bandwidth, closer to the edge, that is optimised for inference.

Practically speaking, this means we are seeing a few trends emerge:

New fibre routes to support the new AI compute facilities – We’re seeing a step-function increase in demand for high-capacity fibre builds between a rapidly expanding set of AI compute hubs. We are seeing hyperscale clients acquiring capacity in the billions of TCV and we think we are still in the early innings.

A new wave of subsea cables – With workloads growing globally, we’re seeing renewed interest in this foundational part of our digital world. As traditional data centre hubs are effectively sold out of available capacity, new AI hubs are rapidly developing around the world that will need to be interconnected. This will drive significant demand for undersea connectivity.

Fibre densification at the edge – The upgrade of networks at the edge, whether it be residential or commercial, fibre to the tower or small cell, will continue.

Ganzi: Right now, capital is flowing at an accelerated pace into data centres—rightfully so, given AI’s demands.

The biggest gap between what we have and what we will need is in energy. AI and high-performance computing workloads are consuming power at an unprecedented rate, and grid constraints are real. We need more investment in renewable energy and battery storage and more widespread use of power purchase agreements to sustain this growth. If there’s one area that needs urgent focus, it’s aligning digital infrastructure with scalable, sustainable power solutions.

As part of this alignment, investors and operators should carefully consider whether to bring power to the data centre or position the data centre closer to power sources. This strategic decision will be critical in ensuring long-term sustainability and efficiency in an era of increasing energy demands.

As I mentioned earlier, FTTH investment is steady and demand for gigabit connectivity is strong, but there have been a lot of either poorly developed or executed business plans out there and a lot of investors are learning the hard way that success in fibre requires industry knowledge, strong business models and access to best-in-class management. I think there have been a lot of investments that don’t have those aspects, and we are already seeing several players exit the market due to poor performance.

Ganzi: Jensen’s vision is spot on. AI is transitioning from software to the physical world, embedding itself into robotics, autonomous systems, and smart devices.

The proliferation of IoT and M2M communications will further shape infrastructure needs, requiring seamless connectivity and real-time processing. This evolution means infrastructure must adapt to support constant AI inference at the edge while still feeding massive training models in core facilities.

The winners will be those who can build for both—balancing high-performance compute at scale with ultra-low-latency delivery at the edge.

Ganzi: AI infrastructure isn’t a partisan issue—it’s a national competitiveness issue. Whether it’s tax incentives for data centre development, streamlined permitting for new facilities, or investments in broadband and energy, continued bipartisan support is critical.

The US must ensure that its infrastructure—whether cloud, fibre, or power supply—is robust enough to support AI-driven economic transformation. Federal backing can accelerate deployment timelines, address zoning and energy constraints, and create a more predictable investment environment.

Ganzi: It’s a big deal. AI data centres require vast amounts of land, power, and connectivity—federal sites can help solve those constraints. This move would open up new geographies for development and may shorten the permitting process—while creating strategic hubs for AI compute. From an investment standpoint, it makes infrastructure more scalable and predictable.

Q: What do you see as the most significant challenges facing the digital infrastructure sector in 2025?

Ganzi: The biggest challenge is power availability. AI and HPC workloads are consuming more energy than traditional cloud applications, and securing reliable, low-carbon power is becoming a gating factor for growth.

Beyond that, we’re seeing geopolitical risk in supply chains, regulatory scrutiny around data centre energy consumption, and the need for workforce development to support this industry at scale. These aren’t unsolvable problems, but they require coordinated effort between investors, operators, and policymakers.

Ganzi: The US remains a leader, but competition is heating up. Markets like the Middle East, Singapore, and parts of Europe are aggressively scaling AI-ready infrastructure.

To stay ahead, the US needs to accelerate power grid modernisation, streamline permitting for new data centres, and continue attracting private capital into digital infrastructure.

The good news is that we have the right mix of talent, technology, and investment appetite—it’s just about execution.

CEO of Digital Bridge Marc Ganzi, 'we have a once in a lifetime opportunity to build AI infrastructure'

Metro Connect USA 2025: Networking redefined for digital infrastructure leaders