Data centres and cloud

Stratum 4 · Networks and systems

Data centres and cloud

Data centres house the servers, storage and networks that run cloud computing, streaming, payments and AI, and cloud firms rent that capacity as services. Cloud infrastructure services earned about US$ 419 billion in 2025 (Synergy Research), while hyperscale operators spent about US$ 142 billion on infrastructure in Q3 2025 alone. The IEA estimates data centres used about 415 TWh of electricity in 2024, about 1.5% of world demand, and projects about 945 TWh by 2030; the US hosts 55% of hyperscale capacity and AWS, Microsoft and Google 58%.²

India. India's operational data-centre capacity reached about 1.75 GW in H1 2026 and is on track for 2 GW by end-2026, with about US$ 173 billion of investment commitments since 2021; Mumbai took 94% of new H1 2026 supply and Mumbai, Chennai and Delhi-NCR hold over 80% of capacity⁴. Cheap renewable power, data-localisation rules and subsea landings in Mumbai and Chennai are the draws.

Where ₹100 of data centres and cloud output goes

Where ₹100 of data centres and cloud output goes Every rupee of sales is paid out to someone along the chain. This splits ₹100 of output by who finally receives it, so you can see how exposed the industry is to commodity, chemical and energy prices. Commodities ₹1.3 Chemicals ₹0.2 Energy ₹2.1…

Price links with supplier industries

Price links with supplier industries How the industries this one buys from shape its prices: the share of its costs they make up, and how their prices have moved together. Tap a row for the charts. Hover or tap ? for what each measure means. Supplier industry Cost share Raw-material mix…

Follow one offering through the chain

Follow one offering through the chain 4 Example story · 6 steps · India An AI cloud in a Navi Mumbai tower Follow Yotta's NM1 campus at Panvel from land, capital and power to Nvidia GPUs rented by the hour. 1 / 6 · Capital, land and power Capital and land at Panvel $173bn India data-centre…

Buys directly from

Buys directly from Computers and electronics Construction and infrastructure Electrical equipment Telecommunications Depends on through its suppliers Building materials Machinery and equipment Metal products Plastic and rubber products Semiconductors Water and sanitation Wood products Sells…

Value chain, from upstream to the customer

Value chain, from upstream to the customer Criticality to the chain: 5 · Chokepoint 4 · Critical 3 · Important 2 · Supporting 1 · Commodity service Stage 1 Capital, land and power Infrastructure investors and data-centre developers 3 · Important Raise equity and debt, buy land with power access…

What the whole chain draws on

What the whole chain draws on Every commodity, chemical and energy tile named by the enterprises above; ×n shows how many of them depend on it. Open any tile for its full detail. Commodities (13) Cu Copper ×5 Al Aluminium ×3 Si Silicon ×2 St Steel ×2 Gas Natural gas U Uranium Li Lithium Pb Lead Au…

Chokepoints and risks

Chokepoints and risks Power is the binding constraint: the IEA projects data-centre demand more than doubling to about 945 TWh by 2030, ahead of many grids' ability to connect, while transformer waits reach up to four years in the US 1 , 5 . AI accelerators depend on one dominant designer (Nvidia…

Sources

Sources Energy and AI: executive summary International Energy Agency, 2025 GenAI helps drive quarterly cloud revenues to $119 billion as growth rate jumped yet again in Q4 Synergy Research Group (TechInsights), 2026 Hyperscaler capex and cloud AI building boom (Synergy Research Group data) CIO…

Relationship cards

• Software and IT services: Cloud platforms and software are two layers of the same stack: software firms rent compute from cloud providers, and cloud providers buy and build vast amounts of software to run it. For India's IT services firms, cloud migration and AI work on the hyperscalers' platforms are a main source of growth, and some, like TCS, are now building data centres themselves. (researched)
• Telecommunications: Data centres are only useful when they are connected, so they cluster where fibre and subsea cables land, and cloud and content firms now generate most international traffic. Telecom operators in turn run their core networks and IT in data centres. In India the same few coastal hubs, Mumbai and Chennai, hold both the cable landings and most data-centre capacity, which concentrates risk. (researched)
• AI and digital platforms: AI models are trained and served in data centres, so AI spending turns directly into data-centre construction and power demand: five large tech firms spent over USD 400 billion of capex in 2025, data-centre electricity use rose 17%, and power and equipment are now the bottleneck. (researched)
• Semiconductors: Data-centre spending on AI is now the main engine of the chip industry: accelerators, server processors and high-bandwidth memory make up most of what goes into new AI halls. The tie runs both ways, since chip designers rent cloud compute for design runs, but the dominant risk is the capex cycle: if hyperscalers slow spending, chip sales fall with it. (researched)
• Natural gas: Gas is the largest single source of power for US data centres and the IEA expects it to supply the biggest share of their added demand to 2030, so the AI build-out has become a new source of US gas demand and a new buyer of scarce gas turbines. (researched)
• Solar photovoltaics: Cloud and AI firms are the largest corporate buyers of renewable power, and solar is a main part of what they contract, but solar output peaks at midday while data centres run all day, so the pair is tied by contracts and by the need for storage or firm power alongside. (researched)
• AC transmission: Large data centre campuses now ask for 100 MW to 1 GW each and must be connected to the high-voltage AC grid, so the pace of transmission and substation build, not servers or capital, sets how fast they can open; in Northern Virginia waits for big connections reached up to seven years in 2024. (researched)
• Copper: A data centre uses roughly 27 tonnes of copper per megawatt in its power, cooling and server systems, and more again in the grid that feeds it, so AI build-out has become a visible new source of copper demand, though still small next to grids, buildings and vehicles. (researched)
• Computers and electronics: Data centres and cloud buys from Computers and electronics; computers and electronics buys from Data centres and cloud; computers and electronics depends on data centres and cloud; data centres and cloud depends on computers and electronics.
• Finance: banking, insurance, markets: Finance: banking, insurance, markets buys from Data centres and cloud; data centres and cloud buys from Finance: banking, insurance, markets; data centres and cloud depends on finance: banking, insurance, markets; finance: banking, insurance, markets depends on data centres and cloud.
• Building materials: Data centres and cloud buys from Building materials; data centres and cloud depends on building materials.
• Construction and infrastructure: Data centres and cloud buys from Construction and infrastructure; data centres and cloud depends on construction and infrastructure.
• E-commerce and digital marketplaces: E-commerce and digital marketplaces buys from Data centres and cloud; E-commerce and digital marketplaces depends on data centres and cloud.
• Electrical equipment: Data centres and cloud buys from Electrical equipment; data centres and cloud depends on electrical equipment.
• Healthcare services: Healthcare services buys from Data centres and cloud; healthcare services depends on data centres and cloud.
• Hospitality and tourism: Hospitality and tourism buys from Data centres and cloud; hospitality and tourism depends on data centres and cloud.
• Machinery and equipment: Data centres and cloud buys from Machinery and equipment; data centres and cloud depends on machinery and equipment.
• Media and entertainment: Media and entertainment buys from Data centres and cloud; media and entertainment depends on data centres and cloud.
• Professional services: Professional services buys from Data centres and cloud; professional services depends on data centres and cloud.
• Public administration, defence and security: Public administration, defence and security buys from Data centres and cloud; public administration, defence and security depends on data centres and cloud.
• Real estate: Data centres and cloud buys from Real estate; data centres and cloud depends on real estate.
• Water and sanitation: Data centres and cloud buys from Water and sanitation; data centres and cloud depends on water and sanitation.
• Aluminium: Data centres and cloud depends on aluminium.
• Cement: Data centres and cloud depends on cement.
• Construction chemicals: Data centres and cloud depends on construction chemicals.
• Crude oil: Data centres and cloud depends on crude oil.
• Data centres: Data centres and cloud depends on data centres.
• Distribution grids, mini-grids and interconnectors: Data centres and cloud depends on distribution grids, mini-grids and interconnectors.
• Electronic specialty gases: Data centres and cloud depends on electronic specialty gases.
• Epoxy resins: Data centres and cloud depends on epoxy resins.
• Flame retardants: Data centres and cloud depends on flame retardants.
• Fluorochemicals and refrigerants: Data centres and cloud depends on fluorochemicals and refrigerants.
• Fluoropolymers (PTFE, PVDF): Data centres and cloud depends on fluoropolymers (PTFE, PVDF).
• Gas turbines and combined cycle: Data centres and cloud depends on gas turbines and combined cycle.
• Germanium: Data centres and cloud depends on germanium.
• Gold: Data centres and cloud depends on gold.
• Heat pumps, air conditioning and refrigeration: Data centres and cloud depends on heat pumps, air conditioning and refrigeration.
• Indium: Data centres and cloud depends on indium.
• Internal combustion engines and gensets: Data centres and cloud depends on internal combustion engines and gensets.
• Lead: Data centres and cloud depends on lead.
• Lithium: Data centres and cloud depends on lithium.
• Lithium iron phosphate batteries: Data centres and cloud depends on lithium iron phosphate batteries.
• Monoethylene glycol: Data centres and cloud depends on monoethylene glycol.
• Nuclear fission reactors: Data centres and cloud depends on nuclear fission reactors.
• Onshore wind: Data centres and cloud depends on onshore wind.
• Silicon: Data centres and cloud depends on silicon.
• Steel: Data centres and cloud depends on steel.
• Transformers and power electronics: Data centres and cloud depends on transformers and power electronics.
• Uranium: Data centres and cloud depends on uranium.
• Water: Data centres and cloud depends on water.
• Water treatment chemicals: Data centres and cloud depends on water treatment chemicals.