A persistent misconception is slowing AI adoption across UK organisations: that embracing AI demands a wholesale overhaul of existing IT infrastructure, or a significant upfront capital commitment. 

According to Tim Loake, Vice President of UK Infrastructure Solutions Group at Dell Technologies, this belief is not merely incorrect – it is actively harmful.

"I've seen organisations invest heavily in AI infrastructure without really understanding the use cases, let alone the models or modalities required to deliver outcomes against those use cases," he says. "The result is expensive infrastructure sitting idle for months while teams work through upfront design and data preparation before anything meaningful reaches production."

Tim's advice is to start small, build a proof of concept on minimum viable infrastructure and grow from there.

A single server building block, he argues, may be sufficient to begin. The discipline of running real workloads, rather than debating hypothetical ones, accelerates learning and surfaces problems far earlier.

He is equally firm on architecture choices. "The only thing worse than building a bespoke 'snowflake' system is a big, expensive snowflake," he says, advocating instead for standards-based, scalable platforms that avoid lock-in and preserve flexibility. 

For Tim, the central conversation should always return to time-to-value and the cultural changes required to embed new tools: "If you don't get people using the tools quickly, you don't realise any of the benefits you built into the business case."

AI has become a standing agenda item for infrastructure leaders

The pace of change in how organisations are treating AI has been striking. Twelve months ago, the technology was frequently framed as something to address in future planning cycles. That framing has shifted decisively.

"Today, it's part of almost every infrastructure and data conversation I have," Tim says. "It has moved from a future topic to a standing agenda item."

Most organisations remain in the experimentation or piloting phase, but the number moving into full production is growing. As that transition happens, capacity planning is emerging as a consistent pain point. Demand almost always ramps more slowly than anticipated before accelerating well beyond original projections – a pattern with direct consequences for networking, infrastructure sizing and cost management.

The cloud-to-on-premises journey is also becoming a familiar trajectory. Pilots frequently begin in public cloud environments, where speed and accessibility make early experimentation straightforward. As usage scales and value proves itself, however, the cost profile changes. "Costs often spiral up faster than expected," Tim notes. "At that point, many organisations look to shift more of their AI estate on-premises to regain control over performance, cost and data location."

Use case clarity and business focus drive smarter AI deployments

Three structural factors help explain why organisations continue to misjudge their AI infrastructure requirements. The first is a lack of use case clarity. Without precise objectives, engineering teams default to over-building.

"A red flag for me is when an organisation is trying to tackle too many projects or use cases at once," Tim says. His rule of thumb is to begin with fewer than five, prioritising those that are both feasible and high in business value. Early wins generate confidence and provide the financial justification for subsequent investment.

The second factor is starting with technology rather than the business problem. The convergence of data, compute, storage and networking in AI workloads creates genuine complexity – and that complexity can push procurement teams towards acquiring broad capability on a speculative basis. The more disciplined approach is to deploy precisely what current workloads require, while retaining the architectural agility to scale as AI maturity develops.

The third factor is legacy infrastructure. When existing environments are not designed for AI workloads, the instinct is often to add hardware rather than revisit architecture. "The biggest issue I see is data silos," Tim says. 

"The work required to build a usable knowledge graph across multiple data sources is almost always massively underestimated." Bolting AI capability onto unsuitable foundations creates bottlenecks; the more effective path is adopting modern, validated architectures that are purpose-built for AI demands.

Leading organisations prioritise adoption and validated architectures

Among the organisations that are seeing consistent returns from their AI programmes, a number of common characteristics are emerging. They focus on high-impact use cases where value can be demonstrated quickly, then scale with evidence behind them rather than assumptions.

Adoption, however, is where Tim sees the clearest differentiator. "The best technology has no value if no one adopts it," he says. "They invest in training, they measure usage and outcomes, and they inspect what they expect – continuously."

Dell has applied this logic internally. "At Dell, for example, we used exactly that approach with our own sales teams to drive adoption of new tools designed to make them more efficient," Tim explains. "By focusing on training, measurement and inspection, we've dramatically increased our face time with customers."

Flexible consumption models are also gaining traction among more mature adopters, particularly where utilisation is difficult to predict. The ability to scale in line with actual demand provides a useful hedge against both hardware price volatility and the uneven ramp patterns that characterise early AI deployments.

Validated, pre-tested architectures are the third hallmark of leading organisations. By removing integration uncertainty and reducing deployment time, these frameworks allow technical teams to focus on building applications rather than managing infrastructure complexity. 

"The key benefits I emphasise are speed and risk reduction: you get to value faster and with fewer surprises by leaning on the experience of your vendor partners," Tim says.

Regulation, energy efficiency and production readiness define the road ahead

The commercial pressure to demonstrate returns from AI investment is intensifying, and it is reshaping conversations at the C-suite level. For Tim, the concerns he hears most consistently from senior leaders are the risk of competitive disadvantage and the challenge of managing sovereign data within regulatory frameworks. This is particularly relevant across EMEA, where energy costs and compliance requirements vary considerably by country.

This context places new demands on infrastructure design. "Modern AI infrastructure must be balanced and integrated," Tim says. "Powerful GPUs in servers need fast storage and high-bandwidth fabric to prevent data starvation and maximise utilisation." Energy consumption, data management and compute costs are each subject to growing scrutiny.

Tighter integration between compute and storage will also become a competitive differentiator. Organisations that treat both as components of a single, integrated system rather than separate infrastructure decisions will benefit from lower latency and faster time-to-insight.

Above all, Tim sees the sector moving decisively from experimentation into structured, repeatable production environments. The framing of the central question is itself changing. "The conversation is shifting from 'how big can we build?' to 'how intelligently can we build?'" he says, "and that shift will define the most successful organisations of the next decade."