Data centres may not keep up with AI capacity requirements – analysis

Demand for information centre capacity may outstrip supply as a effect of the immense requirements of artificial quality (AI), according to a caller report.

GlobalData’s Data Centers report explains that the information centre marketplace is becoming much challenging successful ample portion owed to a surge of AI-related developments and rising request for computing power, which leads to accrued electrical powerfulness requirements.

“Since the merchandise of ChatGPT, the tech manufacture has recognised that existing information centre infrastructure is approaching its limits,” it says. “Server racks successful information centres volition soon beryllium incapable to supply the electrical powerfulness required to bid progressively larger and much analyzable ample connection models.”

Through their powerful computing abilities, vast capacities and fast, reliable operability, information centres allowing businesses to deliver online services, store invaluable information and run power-hungry applications. Indeed, GlobalData’s Cloud Computing report forecasts that the global unreality computing market underpinning overmuch of this and largely facilitated by information centres will beryllium worthy $1.5trn in 2028 having grown astatine a compound yearly complaint of 17% since 2024.

“The unreality computing manufacture is being shaped by multi-dimensional forces, specified arsenic booming AI demand, edge-based low-latency requirements, regularisation and sustainability concerns,” the study says. “Cloud providers are progressively investing successful customized chips and AI tools to conscionable evolving demands.”

However, shifts successful concern trends like growing request for integer and online services, greater automation crossed industries and accrued usage of integrated networks of systems and platforms are increasing request for information centres.

Of the interaction of AI’s emergence, the Data Centers report states: “AI created caller demands connected information centre infrastructure, arsenic AI workloads necessitate important computing power. In parallel, the maturation of borderline computing (where information is processed person to the user) has driven the improvement of smaller, localised borderline information centres.”

To accommodate the increasing demand, more and more data centres are being constructed. GlobalData research shows that the fig of information centre projects by operation commencement date remained relatively stable between 2019 and 2024 earlier surging successful 2025 and 2026 owed to companies scaling up information centre projects to enactment increasing AI workloads. This is problematic successful itself, though, owed to the power required to tally the increasing fig of projects.