Amid AI-driven capacity race and supply chain challenges, data centre spend rushing to $1.6 trillion

Global datacentre capital expenditure is predicted to reach $1.6 trillion by 2030, driven by AI workloads and large-scale infrastructure projects, although supply chain constraints and market fragilities pose risks to this rapid expansion.

The analyst firm Omdia forecasts that global datacentre capital expenditure will expand at a 17 percent compound annual growth rate through 2030, reaching about $1.6 trillion as operators race to build capacity for artificial intelligence workloads and refresh general-purpose infrastructure. According to the original report, rising demand for larger models and greater inference compute is pushing designs towards higher performance, higher power-density servers and racks, and correspondingly more spending on power and thermal systems. [^[1]^](https://www.theregister.com/2025/12/09/omdia_datacenter_capex/)

Omdia’s projection is framed around four scenarios: a baseline that balances confirmed orders against supply constraints (which it judges most likely); an accelerated-growth outcome if constraints ease; a “bubble” case in which investor disappointment in AI productivity slows investment after 2026; and a high-growth “Nvidia” scenario that assumes few constraints and much faster spending. The firm also warns that supply-chain bottlenecks , notably GPU and memory backlogs , are already inflating component prices and, by extension, server costs. [^[1]^](https://www.theregister.com/2025/12/09/omdia_datacenter_capex/)

The scale of the required investment is contested across analysts. McKinsey’s assessment of the compute transition suggests an even larger capital requirement, estimating global datacentre investment needs of about $6.7 trillion by 2030 to meet AI-driven compute demand, of which roughly $5.2 trillion would be for AI-specific centres. Industry forecasters such as Dell’Oro Group offer still different timelines and magnitudes, projecting a 21 percent CAGR to roughly $1.2 trillion by 2029 and noting that GPUs and custom accelerators now represent around one-third of datacentre capex. These divergent estimates underscore uncertainty about pace, geography and what counts as AI-specific versus traditional IT investment. [^[2]^](https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/the-cost-of-compute-a-7-trillion-dollar-race-to-scale-data-centers)[^[3]^](https://www.datacenterdynamics.com/en/news/as-ai-capex-leads-us-economy-global-data-center-capex-projected-to-grow-at-21-percent-cagr-to-12tn-by-2029/)

Market dynamics are already redirecting buyer behaviour. Omdia and market reporting show hyperscalers prioritising high-spec systems for AI over routine server refreshes, creating a temporary suppression of standard server replacement cycles followed by an intensified refresh across segments. Industry data cited by The Register and Omdia indicate server prices could rise materially , channel sources suggested increases of about 15 percent for some configurations as memory and other commodity costs climb. [^[1]^](https://www.theregister.com/2025/12/09/omdia_datacenter_capex/)

The rush to scale has drawn large strategic investments and partnerships. Financial players and sovereign-backed entities are moving to create regional AI hubs: for example, Brookfield and Qatar’s national AI company have formed a $20 billion joint venture to build integrated compute capacity in the Middle East. Major cloud vendors are also committing multibillion-dollar programmes to expand AI infrastructure globally, with announcements this week of sizeable investments aimed at growing cloud footprints and AI capabilities in markets such as India and Canada. These deals illustrate both the appetite for capacity and the geopolitical dimension of AI infrastructure build-outs. [^[4]^](https://www.reuters.com/technology/brookfield-qatars-qai-form-20-billion-jv-ai-infrastructure-2025-12-09/)[^[5]^](https://www.reuters.com/business/retail-consumer/amazon-invest-over-35-billion-india-by-2030-expand-operations-boost-ai-2025-12-10/)[^[6]^](https://www.reuters.com/business/microsoft-invest-more-than-5-billion-canada-over-next-two-years-2025-12-09/)

Yet rapid expansion carries risk. Commentary from market analysts highlights the vulnerability of smaller, emerging ‘neo-cloud’ tenants: if these players struggle financially, they could strain colocation operators and slow demand, potentially choking off a portion of the projected capex growth. Omdia’s scenarios acknowledge that not all developers will succeed if growth accelerates without commensurate productivity gains. The picture that emerges is one of robust investment potential tempered by countervailing market and balance-sheet fragilities. [^[1]^](https://www.theregister.com/2025/12/09/omdia_datacenter_capex/)[^[7]^](https://www.reuters.com/commentary/breakingviews/shaky-data-centre-tenants-could-choke-off-ai-boom-2025-12-10/)

Given these pressures, Omdia advises datacentre and IT teams to re-evaluate design assumptions: future facilities will likely differ markedly from their predecessors in silicon, server and rack architecture, thermal management, and power distribution and backup. Industry participants and governments face a narrow window to align financing, supply chains and regulatory frameworks if the most ambitious capacity build-outs are to be realised without major dislocations. [^[1]^](https://www.theregister.com/2025/12/09/omdia_datacenter_capex/)[^[2]^](https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/the-cost-of-compute-a-7-trillion-dollar-race-to-scale-data-centers)

📌 Reference Map:

##Reference Map:

• [^[1]^](https://www.theregister.com/2025/12/09/omdia_datacenter_capex/) (The Register / Omdia Cloud and Datacentre Market Snapshot) – Paragraph 1, Paragraph 2, Paragraph 4, Paragraph 6, Paragraph 7
• [^[2]^](https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/the-cost-of-compute-a-7-trillion-dollar-race-to-scale-data-centers) (McKinsey) – Paragraph 3, Paragraph 7
• [^[3]^](https://www.datacenterdynamics.com/en/news/as-ai-capex-leads-us-economy-global-data-center-capex-projected-to-grow-at-21-percent-cagr-to-12tn-by-2029/) (Dell’Oro Group / DataCenterDynamics) – Paragraph 3
• [^[4]^](https://www.reuters.com/technology/brookfield-qatars-qai-form-20-billion-jv-ai-infrastructure-2025-12-09/) (Reuters – Brookfield/Qai JV) – Paragraph 5
• [^[5]^](https://www.reuters.com/business/retail-consumer/amazon-invest-over-35-billion-india-by-2030-expand-operations-boost-ai-2025-12-10/) (Reuters – Amazon India investment) – Paragraph 5
• [^[6]^](https://www.reuters.com/business/microsoft-invest-more-than-5-billion-canada-over-next-two-years-2025-12-09/) (Reuters – Microsoft Canada/AI investments) – Paragraph 5
• [^[7]^](https://www.reuters.com/commentary/breakingviews/shaky-data-centre-tenants-could-choke-off-ai-boom-2025-12-10/) (Reuters Breakingviews) – Paragraph 6

Source: Fuse Wire Services