DCE-420 · Lesson 1 of 12 · 31 min
How electricity is supplied in each province
Electricity is the largest operating cost of a data centre and, increasingly, the factor that decides whether a project can be built at all. A 100 MW campus running at high load factor consumes about 800 GWh a year, as much as 70,000 to 80,000 homes, and its electricity bill can exceed $50 million a year. How that power is bought, at what price, with what reliability and carbon content, and how quickly it can be connected, depends on the structure of the electricity sector where the campus is located. This course examines power supply from the data centre owner's point of view: market and regulatory structures, wholesale prices, interconnection, flexibility, renewable procurement, storage, on-site generation and gas supply, contracts, nuclear and emerging sources, and the economics of choosing among them; the section at the end of this lesson sets out how the twelve lessons fit together. It builds on the power systems of ELE-380 and the site selection of DCE-410.
In Canada, electricity is a provincial responsibility, so there are effectively ten different electricity systems, plus the territories. This lesson compares the structures of the main provinces, explains how large customers are billed, and identifies the features that matter most to data centres.
Industry structures
An electricity system has four functions: generation, transmission, distribution and retail (selling to customers). Around the world, these have been organized in two broad ways:
- Vertically integrated utilities own generation, transmission and distribution in their territory and sell to customers at rates set by a regulator based on their costs plus an allowed return. Planning is centralized: the utility decides what to build, subject to regulatory approval.
- Restructured (deregulated) systems separate the functions. Generators compete to sell into a wholesale market run by an independent system operator; transmission and distribution remain regulated monopolies ("wires companies"); and in some places customers can choose a competitive retailer.
Canadian provinces span this range:
| Province | Structure | Key bodies | Notes for data centres |
|---|---|---|---|
| Québec | Vertically integrated Crown corporation | Hydro-Québec; Régie de l'énergie | Abundant low-carbon hydro; large-load allocations controlled by the government; separate data centre rate proposed for loads over 5 MW |
| Ontario | Hybrid: wholesale market with long-term contracts | IESO; Ontario Energy Board; Hydro One; OPG | Global Adjustment charge allocated to large customers by peak contribution (ICI); ministerial approval for large data centre connections |
| Alberta | Competitive energy-only wholesale market and retail choice | AESO; Alberta Utilities Commission; Market Surveillance Administrator | Volatile pool prices; Data Centre Regulation (2026) favours bringing your own generation |
| British Columbia | Vertically integrated Crown corporation | BC Hydro; BC Utilities Commission | Hydro-based; limited power for data centres allocated competitively |
| Manitoba, Saskatchewan, Atlantic | Mostly vertically integrated utilities | Manitoba Hydro, SaskPower, NB Power, Nova Scotia Power, NL Hydro | Smaller systems; large loads need careful planning |
The provinces in more detail
Québec
Hydro-Québec generates almost all of its electricity from hydroelectric stations and sells it at some of the lowest rates in North America, which made Québec attractive for data centres for years. Large-load connections are now subject to government authorization, and Bill 69 (2025) reorganized the governance of energy resources, including how large blocks of power are allocated to projects that serve provincial priorities. Hydro-Québec has proposed a separate rate for new data centres above 5 MW of roughly double the large industrial rate, subject to approval by the Régie de l'énergie.
Ontario
Ontario's Independent Electricity System Operator (IESO) runs a wholesale market and, since 2025, a renewed market design with locational pricing. Much of the province's generation, including nuclear from Ontario Power Generation and Bruce Power, hydro and contracted renewables and gas, is paid under regulated rates or long-term contracts. The difference between those payments and the wholesale market price is recovered through the Global Adjustment (GA), which forms the majority of a large customer's commodity cost. Under the Industrial Conservation Initiative (ICI), Class A customers (generally those with average peak demand above 1 MW) pay GA in proportion to their share of the province's five highest demand hours of the year, which rewards reducing load at those times. The province has proposed treating new data centres as a separate rate class that cannot use ICI, and its 2025 legislation allows ministerial approval of large data centre connections.
Alberta
Alberta has a competitive, energy-only wholesale market run by the Alberta Electric System Operator (AESO): generators are paid the hourly pool price, which can range from zero to the $999.99/MWh cap, and there is no separate capacity payment. Customers buy through retailers or directly from the pool, often with financial hedges. Transmission costs are allocated partly by each customer's contribution to the twelve monthly system peaks. Alberta is restructuring its market and, through its Data Centre Regulation (2026), gives priority to large data centres that bring their own generation, with limited grid capacity for "bridged" projects. Its natural gas supply and lighter regulatory approach have made it a target for very large campuses.
British Columbia and others
BC Hydro is a vertically integrated Crown corporation with mostly hydroelectric supply. Facing rapid load growth, British Columbia limited new connections for energy-intensive industries and now allocates a fixed block of power (400 MW over two years from 2026) to data centres and AI projects through a competitive process, capping individual projects. Manitoba Hydro and Hydro-Québec export large amounts of hydro to the US; SaskPower, NB Power, Nova Scotia Power and Newfoundland and Labrador Hydro serve smaller systems in which a single large data centre can be a significant fraction of demand.
Worked problem
Global Adjustment and peak hours in Ontario
A Class A customer's demand during the province's five highest peak hours averaged 40 MW, while the province's demand in those hours averaged 22,000 MW. The Class A share of Global Adjustment for the year is $5.5 billion. Estimate the customer's GA cost. How much would it save by reducing load to 25 MW during those five hours?
Show solution
- Peak demand factor: \(40/22{,}000 = 0.182\%\).
- GA cost: \(0.00182 \times 5.5 \times 10^9 = \$10.0\) million.
- At 25 MW: \(25/22{,}000 \times 5.5 \times 10^9 = \$6.25\) million, saving about $3.75 million.
Answer: About $10 million a year, reduced by $3.75 million by curtailing 15 MW for five hours. This is why Ontario industrial customers run backup generation or shift load at predicted peaks, and why the treatment of data centres under ICI is contentious. (Figures are illustrative; the actual GA pool and peak values vary each year.)
How large customers are billed
A data centre's electricity bill typically combines several components:
- Energy: the cost of the electricity consumed, per kWh or MWh, at a fixed rate, time-of-use rates, or wholesale market prices passed through.
- Demand: a charge per kW or kVA of the highest demand in the billing period, recovering the cost of capacity in transmission, distribution and sometimes generation. Demand ratchets may bill a minimum percentage of the highest demand in the previous 11 months.
- Transmission and distribution: delivery charges, often based on peak demand, coincident peaks or contracted capacity.
- Capacity or adequacy charges, where separate.
- Riders and taxes: environmental programs, regulatory charges, debt retirement, carbon pricing pass-throughs and sales taxes.
- Power factor charges (ELEC-210).
Large data centres often take service at transmission voltage (69 kV and above), owning their own substations. This avoids distribution charges and losses but requires the customer to fund, build and maintain the substation.
Worked problem
An annual electricity bill
A campus averages 70 MW with a peak of 80 MW. Its tariff charges $45/MWh for energy, $12/kW-month for demand and $6/MWh in riders. Estimate the annual bill and the average cost per MWh.
Show solution
- Energy: \(70 \times 8760 = 613{,}200\) MWh; \(613{,}200 \times 45 = \$27.6\) million.
- Demand: \(80{,}000 \times 12 \times 12 = \$11.5\) million.
- Riders: \(613{,}200 \times 6 = \$3.7\) million.
- Total: \$42.8 million; per MWh \(42.8 \times 10^6/613{,}200 = \$69.8\)/MWh.
Answer: About $43 million a year, averaging $70/MWh. Because demand charges are spread over very high load factor, data centres achieve low average costs per MWh compared with peaky loads; a load with the same peak but half the energy would pay much more per MWh.
Managing price risk
Where customers buy at or near wholesale prices, as in Alberta, the price of energy can change dramatically from hour to hour and year to year. Alberta's average pool price has ranged from under $50/MWh to over $160/MWh in different years, and individual hours can reach the price cap. A large customer chooses how much of this risk to keep:
- Fixed-price contracts: a retailer supplies all energy at a fixed price for one to five years, taking the price and volume risk for a premium.
- Index pricing: the customer pays the hourly wholesale price plus a retailer fee, keeping all the risk and all the potential savings.
- Block and index: the customer buys fixed blocks of energy (for example 50 MW around the clock) at fixed prices and pays index prices for the remainder, hedging most of its constant load.
- Financial hedges: swaps or options with banks or generators fix the price of a quantity of energy without changing physical supply. A virtual PPA (Lesson 5) is a long-term hedge of this kind.
- Demand response: agreeing to reduce load when prices spike or the system is stressed, earning payments and avoiding the highest prices.
Because a data centre's load is nearly constant, it is easy to hedge with flat blocks, unlike loads that vary with weather. The remaining risks are mainly that load ramps more slowly than planned (leaving the customer over-hedged and selling surplus blocks at market prices) and changes in regulated charges, which hedges do not cover.
Worked problem
Block-and-index hedge
An Alberta data centre uses a constant 60 MW. It buys a 50 MW around-the-clock block at $70/MWh and pays the pool price for the rest. In a month of 720 hours, the average pool price is $130/MWh. Compare its energy cost with buying entirely at the pool price.
Show solution
- Block: \(50 \times 720 \times 70 = \$2.52\) million.
- Remainder: \(10 \times 720 \times 130 = \$0.94\) million. Total \$3.46 million.
- All at pool price: \(60 \times 720 \times 130 = \$5.62\) million.
Answer: The hedge saves about $2.2 million in this high-price month. In a low-price month (say $40/MWh), it would cost more than the pool: \(50 \times 720 \times 30 = \$1.08\) million extra. Hedging reduces variability rather than guaranteeing savings.
Large-load tariffs and special rates
Since 2024, many utilities and regulators have created rate classes or contracts specifically for very large new loads. Typical features include minimum bills or take-or-pay provisions based on contracted demand, contract terms of 10 to 20 years, exit fees if the customer leaves early, collateral requirements, and in some cases higher energy rates or a requirement to pay the full incremental cost of new generation. The aim is to ensure that other customers do not pay for generation and transmission built for data centres, and that speculative projects do not tie up capacity. Lesson 3 examines these provisions in the context of interconnection. For the data centre owner, they turn electricity from a variable operating cost into a long-term fixed commitment that must be weighed against uncertain demand for computing capacity.
What matters to a data centre
From the owner's perspective, the electricity sector is evaluated on:
- Availability and timing: whether the system can supply new large loads and when.
- Price level and volatility: regulated rates offer stability; markets offer opportunities and risks; hedging strategies (Lesson 5) manage them.
- Carbon intensity and procurement options: whether clean energy can be bought locally, through PPAs or utility programs.
- Reliability: transmission strength, outage history, and exposure to extreme weather and supply shortfalls.
- Regulatory direction: whether policy makers welcome large loads or are adding conditions, special rates, curtailment obligations or approval requirements, as many jurisdictions have done since 2024.
Because rules change quickly, data centre developers employ energy specialists and regulatory counsel, participate in rate proceedings, and build relationships with utilities and system operators years before they need power.
How this course is organized
The twelve lessons follow the decisions a data centre owner makes about power:
- Lesson 1 (this lesson): how electricity is supplied and priced in each province, and how large customers are billed.
- Lesson 2: wholesale electricity markets, prices, capacity and hedging.
- Lesson 3: interconnection studies, queues and strategies for securing power.
- Lesson 4: data centre load, flexibility and demand response.
- Lesson 5: renewables, carbon accounting and power purchase agreements.
- Lesson 6: energy storage and firm clean power portfolios for hourly matching.
- Lesson 7: on-site generation with gas turbines, engines, fuel cells and batteries.
- Lesson 8: natural gas supply, pipelines, price risk and winter reliability.
- Lesson 9: electric service agreements, large-load tariffs and PPA terms.
- Lesson 10: nuclear, geothermal and other firm clean sources.
- Lesson 11: comparing the economics of power options, including time to power.
- Lesson 12: building a staged power strategy for a large campus.
Common mistakes
MistakeComparing jurisdictions by energy price alone.
InsteadInclude demand, transmission, capacity and peak-based charges, which can rival energy costs.
MistakeIgnoring the Global Adjustment and ICI rules in Ontario.
InsteadPeak-based charges reward predicting and reducing load in a few hours a year.
MistakeAssuming rules will stay as they are for the facility's life.
InsteadLarge-load rules and rates are changing rapidly; build in flexibility and monitor proceedings.
MistakeOverlooking the costs of taking transmission-level service.
InsteadThe owner must fund, build and maintain the substation and associated equipment.
MistakeTreating curtailment for peak avoidance as free.
InsteadRunning backup generators has fuel, wear, emissions and permit implications.
Summary and key terms
- Data centres are among the largest electricity customers; supply structure shapes cost, timing and carbon.
- Vertically integrated utilities plan centrally under cost-of-service regulation; restructured systems use wholesale markets.
- Electricity is provincial: Québec and BC have Crown hydro utilities, Ontario a hybrid IESO market, Alberta a competitive energy-only market.
- Ontario's Global Adjustment and ICI allocate large costs by contribution to five peak hours.
- Alberta's pool price is volatile; its 2026 regulation favours data centres with their own generation.
- Bills combine energy, demand, delivery, capacity, riders and power factor charges.
- High load factor spreads demand charges, lowering average cost per MWh.
- Availability, price, carbon, reliability and regulatory direction decide where to build.
- Vertically integrated utility
- A utility owning generation, transmission, distribution and retail.
- Crown corporation
- A company owned by a Canadian government, such as Hydro-Québec or BC Hydro.
- IESO
- Ontario's Independent Electricity System Operator.
- AESO
- The Alberta Electric System Operator.
- Global Adjustment
- Ontario charge recovering contracted and regulated generation costs above market prices.
- Industrial Conservation Initiative
- Ontario program allocating Global Adjustment by share of the five peak hours.
- Pool price
- Alberta's hourly wholesale electricity price.
- Energy-only market
- A market paying only for energy produced, without capacity payments.
- Demand charge
- A charge based on the highest demand in a billing period.
- Load factor
- Average load divided by peak load.
- Demand ratchet
- A minimum billed demand based on past peaks.
Practice problems
Work each problem on paper first, then reveal the answer to check.
How much energy does a 100 MW campus use in a year at 90% load factor?
Show answer
100 × 8760 × 0.9 = 788,400 MWh, about 0.79 TWh.
What is the main difference between vertically integrated and restructured systems?
Show answer
Integrated utilities own all functions under cost-of-service regulation; restructured systems separate them, with competitive generation in a wholesale market.
Why does Ontario's ICI encourage load reduction at peak times?
Show answer
Class A customers pay Global Adjustment in proportion to their demand in the five highest system peak hours.
In Alberta, what does an energy-only market mean for prices?
Show answer
Generators earn revenue only from energy, so prices rise very high during scarcity to reward available capacity.
A 60 MW peak load pays $10/kW-month demand. Annual demand charge?
Show answer
60,000 × 10 × 12 = $7.2 million.
Why do high-load-factor customers pay less per MWh?
Show answer
Fixed demand charges are spread over more energy.
What is a demand ratchet?
Show answer
A provision billing a minimum percentage of the highest demand in a prior period, often 11 months.
Give two costs of taking service at transmission voltage.
Show answer
Building and maintaining the substation, protection and transformers; responsibility for equipment failures.
Name two provinces that now allocate limited power to data centres through government-controlled processes.
Show answer
British Columbia (competitive 400 MW process) and Québec (government authorization); Alberta and Ontario also impose conditions.
Why should data centre developers monitor regulatory proceedings?
Show answer
Large-load rates, connection rules and curtailment obligations are changing quickly and can alter project economics.
Lesson quiz · 5 questions
Score 60% or more to complete this lesson.
Create a free account to save your progress, take quizzes and earn a certificate.