What Colocation Providers Should Expect From The Event

How can we modernize, intensify, and grow existing facilities to become AI-capable facilities without categorizing every data center as a new hyperscale AI campus?

Our colocation research revealed a significant finding: the market is much more diverse than just "AI data centers."

A colocation platform may simultaneously support cloud computing, enterprise workloads, interconnection, networks, hybrid infrastructure, and AI customers. An existing facility might have dedicated years to serving conventional cloud or enterprise demand but is now gradually introducing higher-density AI capabilities. This approach is distinctly different from starting a new gigawatt AI campus from the ground up.

However, many of these facilities find themselves involved in the same public and regulatory discussions surrounding "data centers."

This is why our agenda has been meticulously tested against the realities of retail, regional, multi-tenant, and hyperscale-focused colocation models, instead of blindly applying the hyperscaler playbook to every scenario.

Not every operator can tackle the challenges like a hyperscaler.

Some hyperscalers can address infrastructure constraints by investing significant capital into next-generation dedicated substations, transmission systems, reclaimed-water technologies, or custom cooling infrastructure. Yet for many colocation providers—especially regional and mid-market platforms—this isn't the operational reality.

Our research clearly indicated this:

"Many operators lack hyperscale budgets or engineering teams."

Instead, numerous colocation providers are evolving their existing facilities incrementally, carefully balancing customer demand with retrofit feasibility, financial discipline, operational risk, and the need to maintain live customer environments during upgrades.

The event will not merely highlight capital-intensive hyperscaler solutions or suggest that all others should replicate them. It will also explore practical options available to colocation providers: targeted modernization, contained AI zones, selective cooling upgrades, increased power capacity, infrastructure replacements, partial redevelopments, and phased expansions.

The crucial question isn't:

"What is the most advanced technical solution?"

But rather:

"What is the most proportionate solution for this asset, this customer mix, this market, and this stage of growth?"

AI may arrive as an upgrade rather than through new campuses.

This insight is especially relevant for the installed colocation base. AI capabilities won’t be limited to new greenfield campuses. They could emerge as an additional 5MW or 20MW, heightened rack densities, liquid-cooling retrofits, more extensive electrical distribution, a new cooling plant, additional substations, or even another building. From the outside, the facility might appear unchanged, yet internally it could be significantly more intensive.

This evolution raises vital operational and commercial questions:

At what stage does modernization transform into redevelopment?

When does an increase in AI capacity necessitate revisiting the original assumptions regarding power, water, generators, cooling, noise, or community impact?

Our research emphasized understanding when an internal technology shift becomes externally significant. A building may not physically expand, but utility loads may increase substantially, the water system's cooling profile may change, and regulatory scrutiny might intensify regarding new generation or infrastructure needs.

Multi-tenant facilities face different accountability challenges compared to single-user sites.

Colocation introduces complexities that aren't present in single-user facilities. A retail colocation facility might serve hundreds of customers, encompassing numerous cloud on-ramps, networks, and digital services. Conversely, a hyperscale-oriented colo facility could cater to just a few major customers, driving much greater concentrated demands. 

Both categories are called data centers, but their customer structures, demand profiles, local roles, infrastructure needs, and accountability models differ significantly.

These distinctions matter when local authorities evaluate planning conditions, infrastructure costs, or community impacts. They are equally crucial when the operator forecasts future demand. A colocation provider might build capacity without knowing the eventual server utilization, rack density, deployment speed, cooling specifications, or whether a customer’s future workload will involve AI training, inference, or traditional cloud computing.

Consequently, operators must ensure flexibility without creating the impression that the facility's overall impact is uncertain.

The event will also address the increasingly indistinct lines between colocation providers and their customers.

Who determines future density? Who specifies cooling requirements? Who owns which components of a liquid-cooling system? Who is accountable when customer deployments deviate significantly from initial forecasts? And when a major tenant increases power or cooling demands, who engages with the community regarding that change?

For colocation providers, these are not merely technical distinctions; they increasingly influence forecasting, contracts, transparency, infrastructure investments, and community accountability.

Customer reporting must align with the actual performance of the site.

This is another key issue concerning colocation that will be integrated into the agenda. A multi-tenant operator might furnish individual customers with data regarding their allocated electricity, carbon, or water usage. Nonetheless, customer-level figures do not inherently clarify the facility's total impact on the host community.

Thus, the event will pragmatically examine what is: directly metered, calculated, allocated, shared, and excluded—which will include how customer reporting correlates back to the site-level power, water, and infrastructure status.

This will necessitate confronting difficult but constructive questions regarding shared cooling, vacancies, shifting utilization, and how impacts are distributed when multiple customers utilize common infrastructure.

Common principles mandate proportional applications.

Ultimately, this may convey the most crucial message to colocation providers. A highly interconnected 10MW urban facility should not automatically be treated the same as a several-hundred-megawatt hyperscale campus. However, the industry should not argue that all data centers are entirely unique, making universal expectations unattainable.

Our findings suggest a more effective middle ground:

Common principles. Proportionate application.

The event aims to assist colocation providers in understanding what principles genuinely apply across the industry and what aspects should vary based on scale, customer mix, operating model, location, and actual local impact.

The objective isn't to transform colocation operators into hyperscalers but rather to empower them to make better decisions about where to modernize, where to expand, what to invest in, how to disclose, how to share responsibility with customers, and when the operational proposition has evolved enough that community dialogues must also change.

In short: How do you ensure an existing asset remains commercially relevant, becomes AI-capable where necessary, and is primed for further growth without compromising the flexibility, credibility, and local permissions that initially made the site valuable?