The internet seems almost supernatural. Tap a piece of glass in Vancouver and a film begins playing. Ask an artificial-intelligence system a question and, seconds later, it produces an answer assembled from patterns learned across vast quantities of information. A bank transfers money, a hospital retrieves a scan and a mine models the geology beneath a mountain all through systems most of us never see.
But the “cloud” is not floating in the sky. It is built on the ground.
Every search, video stream, payment, cloud backup and AI request eventually travels through fibre-optic cable to a physical building filled with computers. These buildings are data centres: the factories, warehouses and power stations of the digital economy. Inside them, some of the simplest materials on Earth—water, sand and rocks—are reorganized by Human Intelligence™ into one of the most sophisticated forms of infrastructure ever created.
That transformation is the real marvel of the industry. Water moving through a turbine becomes electricity. Silica sand becomes a semiconductor. Copper and aluminum become electrical networks. Lithium and other minerals become backup batteries. Glass becomes fibre-optic cable. Human creativity brings these materials together and turns them into computing power, valuable services, high-wage employment, intellectual property and national capability.
What happens inside a data centre?
A data centre is a purpose-built facility containing computers that store, process and transmit digital information. Some are operated for a single bank, government or telecommunications company. Colocation facilities function like highly secure hotels for computers: customers rent rack space, electricity, cooling and network access for machines they own. Cloud providers such as Microsoft, Amazon and Google allow customers to rent computing and storage without owning the underlying hardware.
AI data centres—sometimes called “AI factories”—are the newest and most demanding version. Instead of relying mainly on conventional central processing units, they contain dense clusters of graphics processing units and specialized accelerators. Thousands of these chips can work together to train an AI model, run scientific simulations or generate answers for millions of users.
Imagine someone in Vancouver asking an AI assistant a question. The request leaves the user’s device, passes through an internet provider and travels along fibre-optic networks. Routers direct it to the appropriate data centre. There, processors perform billions of mathematical operations, storage systems retrieve relevant information and networking equipment coordinates the machines. The response then travels back to the user, often in a fraction of a second.
Making that exchange feel effortless requires an extraordinary amount of supporting infrastructure. Electricity arrives through substations and transformers before passing through switchgear, uninterruptible power supplies and distribution equipment. Batteries keep machines operating through brief interruptions; generators can provide emergency power during longer outages. Security systems protect the site, while redundant fibre routes ensure that information can keep moving if one connection fails.
Cooling is equally important. Nearly all the electricity consumed by a processor ultimately becomes heat. Fans, chillers, heat exchangers, cooling towers and increasingly direct-to-chip liquid systems carry that heat away. Operators measure efficiency using indicators such as Power Usage Effectiveness, which compares total facility electricity with the portion used by computing equipment, and Water Usage Effectiveness, which measures water consumption relative to computing output.
The scale is growing rapidly. The International Energy Agency estimates that data centres consumed about 485 terawatt-hours of electricity worldwide in 2025 and projects consumption of roughly 950 terawatt-hours by 2030. AI-focused facilities are expected to grow much faster than conventional data centres. That is an immense new industrial demand—but it is also evidence that computation has become a foundational economic input, much like transportation, finance or electricity itself.
Water, sand and rocks become intelligence
The magic of a data centre begins with materials.
Water is an obvious example in British Columbia. Rain and snow accumulate in watersheds; gravity carries that water through rivers and reservoirs; turbines run by BC Hydro convert its movement into electricity. That electricity travels through transmission lines to a data centre, where it powers processors performing calculations that would have been unimaginable a generation ago. A river flowing through a turbine can eventually help design a medicine, animate a film, optimize an electrical grid or translate a conversation.
Water can also be used in cooling, although requirements vary dramatically. Evaporative systems may consume significant volumes, while dry cooling and closed-loop liquid systems can greatly reduce onsite use, sometimes in exchange for higher capital costs or electricity consumption. Cooling loops can require corrosion inhibitors, scale-control chemicals, biocides, glycol or specialized dielectric fluids. The relevant public-policy question is therefore not whether a data centre “uses water,” but how much it uses, what type it uses, how it is treated and whether the design is appropriate for the local climate and water supply.
Then there is sand. Silicon, the basic material in most computer chips, is derived from silica. Through an almost astonishing sequence of purification, crystal growth, precision slicing, chemical deposition and microscopic etching, an abundant mineral is turned into a semiconductor containing billions of transistors. Those transistors switch on and off at tremendous speed, translating electrical current into logic.
Around the chip sits a miniature mineral economy: copper and aluminum carry electricity; gold, silver, tin, tantalum, tungsten and palladium appear in electronic components; gallium, germanium and rare-earth elements support specialized semiconductors and optical systems. Steel forms racks and structures. Cement and aggregate become the building. Lithium, nickel, manganese, cobalt or lead may be used in backup batteries. Glass carries pulses of light through fibre-optic networks.
The internet is therefore not post-industrial. It is minerals and energy refined to an extraordinary degree. A data centre is what happens when mining, chemistry, electricity, manufacturing and software converge.
How the business creates value
The business model is easier to understand when the industry is viewed as a chain.
A landowner or developer supplies the site and building. A utility sells electricity and connects the project to the grid. Telecommunications companies provide fibre. Hardware manufacturers sell servers, processors, storage and networking equipment. The data-centre operator supplies secure space, power, cooling and reliability. A cloud provider packages that infrastructure into computing, storage and software services.
At the end of the chain are the customers: governments, banks, hospitals, universities, streaming platforms, mines, engineering firms, software companies, AI developers and small businesses. They may pay for rack space, reserved power, data storage, network capacity, cybersecurity, processing time or “GPU hours.” Increasingly, they buy computation as a metered service, just as an earlier generation bought long-distance telephone minutes.
An AI factory converts electricity and semiconductor capacity into model training, predictions, simulations, images, software and tokens of generated language. If those outputs help a company discover a mineral deposit, automate a business process, create a new product or sell a service around the world, physical inputs in British Columbia have been converted into income and intellectual property.
The facilities themselves are capital-intensive. Construction requires engineers, electricians, concrete and steel suppliers, cooling specialists, fibre installers and equipment technicians. Permanent operations require network engineers, electrical and mechanical technicians, cybersecurity specialists, facilities managers and security staff. Municipalities receive property taxes, permit revenue and new assessment value.
It is also important not to oversell direct employment. A data centre may employ fewer permanent workers than a mine or factory of comparable cost. Its greatest economic promise is the ecosystem that can grow around it: AI companies, research laboratories, visual-effects studios, life-sciences firms, clean-technology developers and existing resource businesses able to purchase advanced computing close to home.
From local infrastructure to national sovereignty
Data centres can also create value that does not appear neatly in a jobs estimate.
Locating computing closer to users can reduce latency, improve network resilience and provide backup capacity during emergencies. New substations and fibre routes can strengthen local infrastructure. The enormous quantity of usable heat produced by servers can even become an asset. Instead of releasing it into the air, a facility can transfer it into district-energy systems serving homes, offices, pools, greenhouses or industrial users.
Domestic computing capacity is also becoming a matter of sovereignty. Canadian governments, hospitals, businesses and researchers generate sensitive information. Keeping that information in Canada can place it under Canadian jurisdiction and reduce dependence on foreign infrastructure. Physical location alone is not sufficient—ownership, encryption, contracts and legal access still matter—but a country cannot exercise meaningful digital sovereignty without infrastructure on its own soil. The AI and data factories storing our sensitive information need to be hosted in our home and native land.
The same is true of artificial intelligence. Countries without adequate computing capacity may find themselves renting access from foreign companies, waiting behind foreign customers or sending valuable information and research abroad. Sovereign computing can support cybersecurity, emergency management, defence, scientific research and the development of Canadian AI systems. In that sense, a data centre can be both commercial infrastructure and a strategic national asset.
Why British Columbia and why now?
B.C. has several natural advantages: a hydro-dominated electrical system, a relatively cool climate, strong universities, an established technology and digital-media workforce, and connections to Canadian, American and Pacific rim markets. Vancouver offers fibre connectivity, customers and specialized talent. Interior and northern communities can offer cooler conditions, industrial land and proximity to electrical infrastructure.
Interest now exceeds the power initially made available. BC Hydro’s 2026 Call for Demand received 15 applications representing close to 800 megawatts. For the first two-year allocation period, the Province has provided for up to 100 MW for conventional data centres and 300 MW for AI facilities, with a maximum of 145 MW at one site. Rather than simply connecting projects in the order they apply, BC Hydro is assessing factors including price, readiness, economic and community benefits, environmental performance and data sovereignty.
The competition matters because electricity has other uses. Homes, electric vehicles, mines, manufacturing plants and building heating all require additional supply. BC Hydro expects provincial electricity consumption to grow by approximately 50 per cent by 2050. Data centres can be a remarkable use of clean power, but proponents should still show what British Columbians receive for each scarce megawatt: investment, taxes, durable employment, grid flexibility, Indigenous participation, local purchasing, heat recovery, Canadian computing access and locally retained intellectual property.
TELUS’s proposed B.C. AI cluster illustrates both the opportunity and the questions. The company has announced an expanded Kamloops facility, a Mount Pleasant facility intended to open and scale through 2028, and a proposed facility at 150 West Georgia Street in Vancouver for 2029. TELUS says the three sites could exceed 150 MW by 2032. The Vancouver facilities are designed to connect with the city’s Neighbourhood Energy Utility and Creative Energy’s downtown system, potentially turning waste heat from computation into heat for surrounding buildings. Timelines, approvals, electricity allocations and commercial arrangements still matter, but the concept is compelling: hydroelectricity enters a building as power; computation and economic value leave through fibre; useful heat flows into the neighbourhood.
Choosing what B.C. does with its electrons
A serious data-centre strategy must ask practical questions. Who pays for grid upgrades? How much potable water will a facility consume? Can it reduce demand during system peaks? How many jobs and how much tax revenue will remain in B.C.? Will Indigenous communities participate as owners, suppliers or revenue partners? Will equipment be reused or recycled? Will British Columbian businesses and researchers have access to the computing capacity?
These questions do not diminish the wonder of the industry. They are how B.C. ensures that the wonder produces broadly shared benefits.
A data centre takes things that appear ordinary—flowing water, sand, metal-bearing rocks, a parcel of land and ideas in the human mind—and arranges them into machines capable of storing civilization’s knowledge and performing trillions of calculations. It converts natural resources into digital infrastructure, and digital infrastructure into services, discoveries, businesses, public revenue and national power.
There is something genuinely magical in that. But it is not magic in the supernatural sense. It is the more impressive kind: the result of geology, physics, engineering, investment and human creativity working together. If British Columbia chooses its projects carefully, the province can do more than supply electricity to the global digital economy. It can use its own rivers, resources and talent to build wealth, capability and sovereignty here at home.
Siavash Tahan can be reached at [email protected].
Resource Works News
