Increasing use of AI is putting pressures on servers, cooling systems and the like that they weren’t designed for, which means infrastructure is an important conversation – and resellers can add value here, as Martin Ryder, channel sales director Northern Europe at Vertiv, explains.
Artificial Intelligence (AI) is no longer emerging technology. It’s active in everything from financial modelling to image recognition and customer demand is growing. But while most of the attention has focused on software, the impact on physical infrastructure is becoming impossible to ignore.
The servers, cooling systems and power setups that support traditional enterprise IT weren’t designed with AI in mind. The shift towards high-density, high-heat, power-hungry compute environments is already underway – and it’s not limited to hyperscalers.
For channel partners, including resellers, systems integrators and MSPs, the implications are technical and commercial. As customers start to ask how AI fits into their existing infrastructure, they’ll want practical guidance. The channel is in a strong position to provide the infrastructure needed – but need to move quickly if it is to keep up with demand and take advantage of the opportunities this presents.
AI is changing infrastructure assumptions
One fundamental change is that AI workloads behave differently from most enterprise applications. They run faster, draw more power, and fluctuate far more aggressively in comparison to conventional workloads that typically operate at a steady baseline. AI training and inference jobs can shift from low to high utilisation in seconds.
This creates difficulties for existing power and cooling systems, especially in environments where capacity planning was based on older compute patterns. Racks designed for 10 kW–15kW loads are now being pushed towards 50kW or more.
The infrastructure required to support this shift looks very different to traditional systems. Liquid cooling is starting to be implemented to work along side traditional airflow models. Power distribution units and uninterruptible power supplies need to handle surges and unpredictable spikes. Meanwhile, software-defined power management is emerging as a way to smooth loads and improve resilience.
For most customers, adapting to these conditions will require more than a hardware upgrade. It may mean revisiting the underlying architecture.
Customers feel pressure
AI is now an essential part of infrastructure conversations. Some organisations are exploring early-stage deployment of AI tools, while others are further along the road and starting to hit real limitations – whether in their power systems, thermal management or physical capacity.
In both cases, they’re looking for advice. Many haven’t had to think about their on-premise estate in years, having focused investment on cloud migration or user-facing upgrades. Now they’re dealing with unfamiliar questions: can the site support direct-to-chip cooling? What happens if our AI training workloads trip a breaker? Do we need a separate high-density zone?
These questions don’t have simple answers – and that’s where channel partners can add value. The ability to assess existing infrastructure, make informed upgrade recommendations and support implementation will become a key differentiator in the months ahead.
Scaling smart
Not every customer has the budget, or appetite, for a full infrastructure rebuild. For many, the priority will be phased upgrades that extend the life of existing assets while making room for AI.
That might mean deploying modular power units, upgrading cooling systems to support hybrid environments, or isolating high-density AI workloads within separate parts of the facility. It could also mean investing in better monitoring, to understand exactly where capacity is being stretched.
Channel partners can guide customers through these choices and help them to avoid expensive missteps. Retrofitting isn’t always the cheapest route long-term, but can provide breathing space while a longer-term strategy is put in place.
New demand, new revenue models
Ultimately, there’s no single way to monetise the AI infrastructure shift. Some partners will benefit from hardware refreshes, as customers invest in new power, cooling and density systems. Others will build longer-term value through consultancy, design services and ongoing support.
Infrastructure-as-a-service models are gaining traction among customers looking for flexibility. There’s also increasing interest in prefabricated modular deployments that allow for AI-specific zones without needing to rework the entire estate.
Channel partners that can combine product expertise with broader strategy – balancing cost, risk and long-term performance – will be best placed to lead these conversations.
Questions to ask
Partners should be having infrastructure-focused discussions with customers exploring AI. Here are a few starter questions:
? What is your current rack density, and how much capacity is left?
? Have you assessed your cooling systems for high-density workloads?
? Are your power systems equipped to handle rapid load changes?
? Do you have visibility into where your energy and thermal hotspots are?
? Are there AI use cases on the horizon that could shift infrastructure needs quickly?
These questions aren’t just technical, they’re strategic. Helping customers work through them can position the channel as a critical partner in their AI journey.
Planning beyond the hype
AI’s infrastructure implications aren’t a theoretical problem for tomorrow – they’re a practical challenge today. Power and cooling bottlenecks are already holding back deployments in some environments and as adoption accelerates, these issues could become more widespread unless they are addressed.
Channel partners that understand the pace and nature of this shift won’t just be responding to change, they’ll be shaping it – supporting smarter deployment models, more resilient infrastructure, and a new baseline for how enterprise IT operates.
The infrastructure conversation is back on the table. And for the channel, it’s an opportunity to become strategic long-term partners.
