13 February 2026
A majority of senior IT and data centre leaders expect AI workloads to outstrip current capacity by 2027, as demand grows faster than new infrastructure and power connections.
That is a headline finding from Salute's State of the Industry 2026 report, based on a survey of 200 senior IT and data centre professionals in the UK and the US. The study points to mounting pressure on operators as they plan AI and high-performance computing deployments whilst managing electricity constraints and a tightening labour market.
In total, 83% of respondents believe AI workloads will exceed their current capacity within two years. Readiness is already an issue, with 74% saying they are not fully prepared to support AI workloads today.
AI and high-performance computing are increasingly shaping build and expansion plans. Nearly half of respondents (48%) named them as the biggest driver of new data centre projects. The report also points to a shift in how decision-makers assess growth, moving away from footprint-led expansion to a performance-led strategy. Operators are prioritising higher-density deployments and the ability to run specialist workloads reliably.
Utilisation expectations are rising in parallel. The survey found that 80% expect to run AI infrastructure at significant or full utilisation by 2027, and 84% said speed-to-market directly influences investment decisions.
Electricity availability emerged as the biggest barrier to AI expansion, ahead of skills, funding and supply chain issues. In several major markets, grid access is becoming a practical limiter as data centre demand rises and connection queues lengthen. The report cites forecasts that electricity demand from data centres will more than double by 2030. This is pushing power strategy up board-level agendas and reshaping how operators approach site selection, expansion sequencing and risk management.
Erich Sanchack, Chief Executive Officer at Salute, said scaling AI extends beyond technology in the rack. "Scaling AI isn't just about technology in the rack, it includes resilience of supply," said Sanchack. "The grid has become the new bottleneck. Operators that can turn constrained megawatts into dependable performance will own the future."
Operators are putting more emphasis on improving existing estates. Half said they are upgrading efficiency and capacity rather than relying solely on new builds. This includes changes to power and cooling design, retrofits to raise usable capacity, and operational measures to run more compute within the same electrical envelope.
Power procurement and resilience measures are also becoming more central to planning. Renewable power purchase agreements, on-site generation and battery storage are now mainstream in power strategies. These approaches can reduce exposure to price volatility and add flexibility during peak demand, though most large-scale deployments still require grid connections.
Alongside energy constraints, staffing is a growing challenge. The report highlights shortages of specialist operations talent, limited training pathways and competition from hyperscalers as barriers to AI readiness. These pressures affect operational roles that keep facilities stable, efficient and compliant, as well as teams responsible for commissioning, maintenance and optimisation. Higher-density AI hardware also raises the bar for operational discipline, given its sensitivity to power quality, cooling performance and downtime.
Jon Healy, Managing Director EMEA at Salute, said specialist operations talent is critical to realising the full potential of advanced facilities. "You can build the most advanced facility in the world, but without specialist operations talent, it will never realise its full potential," said Healy.
Despite concerns about cyber security and regulation, respondents remained positive about AI investment. The survey found that 75% believe the rewards outweigh the risks, and most expect a return on investment within five years. The findings reflect a market where commercial demand for AI services and enterprise adoption continue to grow, even as regulators move to set clearer rules on data use, model governance and security controls. Operators and IT leaders also face rising requirements for resilience and compliance, particularly when AI infrastructure is deployed in multi-tenant environments.
The report argues that future capacity will depend on design choices, power availability, operational practices and workforce planning, with many organisations expecting infrastructure to run at high utilisation by 2027.
