Speakers:

James Grice, chair, data centre and digital infrastructure practice – Akerman LLP (chairperson)
Doug Recker, founder/president – Duos Edge AI and Duos Technologies Group
Scott Willis, CEO – DARTPOINTS
Josh Rudin, member – Mintz
Fred Bayles, vice president, corporate development – Cologix

The data centre industry has never moved faster – or faced more pressure. At Metro Connect 2026, a panel of experts gathered to discuss where the next wave of deployment is heading, what is holding it back, and why the enterprise edge market may be the smartest bet in an era of runaway hyperscale ambition.

The conversation, chaired by James Grice of Akerman LLP, brought together Fred Bayles of Cologix, Scott Willis of DARTPOINTS, and Doug Recker of Duos Edge AI.

What emerged was a nuanced picture of an industry being pulled simultaneously in two directions: towards gigawatt-scale campuses serving the hyperscalers, and towards nimble, smaller-footprint deployments serving the enterprises and communities that the big players simply will not touch.

The enterprise edge opportunity

While headlines have been dominated by trillion-dollar hyperscale commitments, several panellists argued that the real opportunity, and the faster return, lies in the one-to-ten megawatt range. Recker, whose business has built a model around deploying in underserved markets, was direct on this point.

“The quickest way to put capital to work is to go to markets where you can get one to five megawatts readily available and deploy infrastructure in a year, not two or three years,” he said. “I think the enterprise data centre is going to be a winner. If I were an investor, I would be in that space – 30 megawatts or below, you’re in great shape.”

The logic is straightforward: smaller deployments sidestep many of the regulatory, utility and community hurdles that plague larger sites, whilst still generating meaningful returns quickly. Recker described how deploying a pod in a market like Amarillo, Texas, where companies had previously been co-locating in Dallas, created a ripple effect of connectivity and commercial activity that attracted further investment into the region.

What struck the panel was how rapidly demand is evolving even in these smaller markets. “The local hospital wants to do AI,” Recker noted. “We’d fill our pods out at 200 kW, and now hospitals are asking for high-density cabinets. We never thought we’d be doing that in a market like that.”

For Willis and DARTPOINTS, the challenge is designing infrastructure flexible enough to serve a wide range of enterprise customers simultaneously. “Customer A’s demands are going to be very different from customer B’s, which will be different from customer C’s,” he explained. “It’s really about how you create that ecosystem that can meet multiple demand requirements in the enterprise market.”

Bayles added that the customer mix is also shifting in unexpected ways. Cologix has seen an uptick in high-frequency trading clients and pharmaceutical companies requesting larger deployments in markets where those sectors had previously been absent. “Being agile for the enterprise is really one of our main focuses,” he said. “We’re building sites with 25 megawatts initially, but with the ability to scale up to over 100 megawatts.”

Power, utilities and the constraint that defines everything

No conversation about data centre deployment in 2026 can avoid the subject of power – and this one was no exception.

Grice, who works extensively on the legal and policy side of large-scale deployments, painted a stark picture of a utility sector struggling to keep pace with demand.

“Utility power is very challenging at scale,” he said. “The idea of priority queue for large-scale sites – you used to have it, you don’t now. If you’re prepared to sign the energy services agreement and put up the collateral, then you can get power, but the collateral numbers on the large side are actually shocking.”

He described clients in Texas deliberately sizing projects at 74.9 megawatts to stay below the threshold that triggers a more complex interconnection process – a telling indicator of how regulatory friction is already shaping development decisions on the ground. The broader concern, he argued, is speculative demand. Utilities are wary of committing capacity to projects that may never materialise, and that wariness is translating into tougher contract terms, higher collateral requirements and slower approvals.

Bayles put the scale of the challenge into perspective. “The run rate goal is 40 gigawatts a year,” he said. “Looking at the next five years, it’s roughly $1.8 trillion. And 70% of that right now is in the US.” He noted that hyperscalers are already thinking in gigawatts, with 300 megawatts representing merely the opening phase of a larger campus. “I just don’t think we’re even close to where the size of this opportunity will be.”

Beyond power, Grice flagged the full checklist that any serious developer must address before breaking ground: land control, water availability, fibre connectivity and, increasingly, community relations.

The last of these has become a material constraint in its own right. “It used to be that everybody wanted a data centre,” he said.

“That’s not the way it is now.” He noted that six US states are currently attempting to pass moratoriums on data centre development, driven largely by community pushback over power consumption and water use – concerns that may or may not be well-founded, but are being heard by local decision-makers regardless.

The retrofit question and what comes next

With new build timelines stretching to two or three years and lead times on equipment remaining unpredictable, attention is turning to whether existing legacy data centres can be upgraded to meet modern density requirements. The panel’s view was cautious. Most corporate data centres were designed around 7–8 kW per rack; retrofitting them for 100–200 kW workloads is, in many cases, structurally and economically unviable.

“You just can’t go into an existing data centre and simply add density,” said Willis. Instead, DARTPOINTS is pursuing a campus model – building new, higher-density facilities adjacent to legacy sites and migrating customers across as their requirements grow.

Bayles noted that Cologix has reviewed around 400 sites over the past 14 months and consistently finds that the cost of modernisation must be factored into any acquisition price from the outset. “You’re talking about 50 sites, times 100 megawatts, times $5,000–$10,000 per kilowatt to get those up to modern standards,” he said. “That’s a big number.”

The overarching message from the panel was one of structured optimism. Demand is real, capital is available and the opportunity set – from edge deployments in mid-sized American cities to eventual expansion across Asia and Europe – is enormous.

But, the path to capturing that opportunity runs directly through a set of constraints – power, supply chain, community sentiment and regulatory complexity – that will separate disciplined operators from those caught overextended. As the panel summarised, the winners will be those who can deploy quickly, in the right markets, at the right scale.