Craig Eadie, Director at Straightline Consulting, outlines why without earlier, deeper commissioning expertise, tomorrow’s high-density AI data centres risk becoming under-powered bottlenecks instead of the engines of a new digital economy.
The rise of generative artificial intelligence is reshaping the global economy at breakneck speed. From autonomous software development to synthetic video and hyper-personalised content, GenAI is rewriting the rules for productivity and innovation. It’s already being credited with the potential to raise GDP across industrialised nations by up to 15% over the next decade. But the real story is happening behind the scenes – in the buildings powering this transformation.
Data centres are the refineries of the AI era. But those refineries only work if they’re properly commissioned. And that’s where the biggest bottleneck in the AI economy may be forming.
The boom in AI-driven infrastructure is changing everything for commissioning professionals. Where commissioning was once seen as the final hurdle in the construction timeline – a tick-box phase between build and handover – it’s now one of the most critical phases in the entire data centre lifecycle. Why? Because GenAI is driving scale and complexity at levels the industry has never seen.
The pressure on data centre commissioning teams is rising across the board. GenAI workloads demand far more energy, far more cooling, and far more resilience than legacy cloud or colocation environments. By 2025, nearly half of all global data centre power could be consumed by AI alone. These aren’t just new builds – they’re next-generation facilities designed to operate on the edge of technical possibility. High-density power environments, novel cooling systems like liquid-to-chip, extreme redundancy requirements, and ultra-stringent uptime SLAs have all become baseline expectations.
Commissioning is where all of this comes together. It’s where the design is validated, where the systems are tested, and where the risks are exposed – or not. And right now, the risks are growing.
Access to power is already becoming one of the most serious constraints. AI needs more electricity than traditional data workloads, and many key markets are already struggling to keep up. Projects are being delayed or mothballed because the energy simply isn’t available. In the UK, local authorities are pushing back on new data centre developments over concerns about grid load and water use. The Government’s AI Growth Zones are a step in the right direction, but infrastructure and approvals still lag far behind demand. The European Union’s plan for AI factories and gigafactories sounds ambitious – but without faster planning and clearer access to power, those plans may never materialise at scale.
For commissioning teams, this means working with incomplete infrastructure, running test procedures under partial power scenarios, and developing new approaches to ensure compliance and resilience in the face of uncertainty. It’s commissioning on a tightrope – balancing technical standards against real-world constraints in an environment where failure is not an option.
The supply chain isn’t helping, either. Long lead times on core equipment – especially for cooling systems and switchgear – are now common, and cost volatility is creating headaches for developers and operators alike. This squeezes commissioning timelines and forces teams to compress months of validation into shorter and shorter windows. The risk here is obvious: if commissioning becomes rushed, quality drops. And in a high-stakes AI data centre, quality is everything.
Then there’s the talent question – the elephant in the room. Commissioning has always been a specialised field, requiring deep electrical or mechanical knowledge, systems-level thinking, and hands-on experience. But the sector simply isn’t producing new talent at the rate needed to meet the current level of demand – let alone what’s coming next. Younger engineers are going into software, finance, and AI itself. Commissioning isn’t on their radar – and that’s a problem. Right now, commissioning engineers are aging out of the workforce faster than they’re being replaced. It’s not just a slow leak. It’s a talent crisis.
And the job itself is getting harder. Commissioning GenAI-ready facilities involves entirely new systems and protocols. Liquid cooling, for example, fundamentally changes the way commissioning is performed. Many of the systems can’t be fully tested until all final hardware is in place, which often means pushing commissioning into the operational phase – creating friction between delivery and operations. Digital twin technology offers new ways to test and monitor, but it also demands a new skillset. Today’s commissioning engineer needs to be part traditionalist, part technologist – comfortable on the floor and in the cloud.
At Straightline Consulting, this shift is already underway. With more than 825 MW of commissioned capacity across the UK, Europe, and beyond, the firm is seeing demand not just grow – but change shape. Clients aren’t asking for help getting over the finish line anymore. They’re bringing commissioning teams in earlier, asking for strategic insight, risk modelling, and support shaping the design itself. The commissioning process is becoming a key driver of project success, and a litmus test for whether a data centre is truly fit for AI workloads.
But the industry can’t do this alone. Unless governments, developers, and investors treat commissioning as a critical link in the AI infrastructure chain – not just a late-stage handover requirement – we risk building data centres that are underpowered, under-tested, and ultimately unfit for purpose. The AI boom isn’t slowing down. But unless commissioning keeps pace, the infrastructure behind it might.
Commissioning used to be invisible. That’s no longer the case. In an AI-powered world, commissioning is the final gatekeeper – the force that ensures tomorrow’s digital economy runs on infrastructure that actually works.
If AI is the new oil, then commissioning is the pressure valve keeping the whole system from blowing wide open.
