A grid fails the way a bridge fails: not from average loads, but from concentrated stress at weak joints. Canada is aligning those stresses. Extreme heat is no longer a meteorological nuisance; it is now a balance sheet liability for utilities and provinces. Reliability planners warn of “profound change” unfolding faster than their models assumed. Meanwhile, demand from AI data centers and electric vehicles stacks new peak loads on top of air conditioners, while hydropower — the country’s historical shock absorber — is thinning under drought. Add trade tensions that make cross-border electricity flows political, and the system looks less like a diversified portfolio and more like a leveraged bet on good weather and smooth relations. Markets price electricity as if past stability guarantees future uptime. That is a fragile assumption.
The old grid was designed for a winter-peaking Canada and predictable industrial loads. That world is gone. Air conditioning adoption has climbed for years, pushing summer peaks higher. Layer on electrification and the hunger of data centers training and serving AI models, and the physics becomes harder. Utilities can meet average energy needs and still fail on the peak. The North American Electric Reliability Corporation — a dull name for the people who police these stress tests — has been blunt: we are living through rapid change, with the pace accelerating. Their projections point to a double-digit jump in peak summer demand over the next decade, driven by new loads that do not sleep when the thermometer spikes. EV charging can be shifted, but it does not always shift. Data centers promise demand flexibility, but their real product is low latency and uptime, not altruism. When temperatures soar, air conditioners, EV chargers, and servers tend to draw power in the same hours. The portfolio benefits of diversity erode. What looks manageable in an annual energy budget turns dangerous in a two-hour ramp. Investors and policymakers who fixate on average consumption statistics miss the point. The grid’s job is to survive the worst hour, not serve the average day.
Canada’s ace in the hole has been hydropower — flexible, dispatchable, clean. In reliability terms, large reservoirs are a battery you can count on when wind slumps and solar fades. But hydropower is water plus gravity; when the water is absent or restricted, the firm turns soft. Prolonged drought across key watersheds has cut hydro output and narrowed the export surplus. Last year, electricity exports to the U.S. fell sharply, while some provinces increased imports during tight periods. This is not merely an energy problem; it is a capacity problem that arrives exactly when it hurts most. In probabilistic terms, the variables we pray are independent — temperature, wind speeds, river inflows, wildfire risk — are moving together. Call it tail dependence: the bad tails line up. Smoke reduces solar output. Heat arrives with stagnant wind. Drought lowers reservoirs and raises wildfire risk that threatens long transmission corridors. The 1996 Western blackout cascaded after hot weather drove heavy flows, trees hit lines, and protection systems did what they were designed to do — but the network could not absorb multiple hits. The modern network has more automation, but also tighter margins. When hydropower’s buffer shrinks, imports become the backstop. And if neighbors face their own heat wave at the same hour, that backstop is an empty promise.
Reliability planning assumes that interties behave like firm insurance. But interconnections are contracts wrapped in politics. Trade tensions have escalated, with new tariffs and the threat of retaliation, including talk of surcharges and even cutting electricity exports. Ontario’s leadership signaled it would participate if the federal government retaliates. This is not theater. Several U.S. states rely on Canadian power to keep peak prices in check and lights on during stress. Quebec and Manitoba hydropower smooth volatility in New England and the Midwest. When tariffs become bargaining chips, electricity becomes collateral. Game theory offers a simple lesson: in repeated games, credibility matters. Threats that were once unthinkable start to look plausible when other levers are exhausted. The equilibrium shifts from cooperation to tit-for-tat. The risk is not that Canada or the U.S. permanently sever grid ties; it is that, in a hot week, deliveries get curtailed or scheduled flows face new friction. Reliability coordinators can model N-1 contingencies — the loss of a generator or line — but they cannot easily model N-1-Politics. The outcome is a quieter kind of fragility: more reliance on local peaking assets, higher scarcity prices, and a thinner safety net just as climate variability widens.
The financial imagination tends to favor efficiency: run assets hot, defer capex, lean on imports, and count on the smoothing magic of markets. That mindset works until the distribution shifts. The grid, like a balance sheet, needs liquidity. In power-speak, that means firm capacity, geographic diversity, and reserves that matter in the specific hours when mortality risks go up. Solar is valuable but fades at sunset when heat lingers. Wind is cheap but can be becalmed during high-pressure heat domes. Gas turbines are fast, but gas pipelines can bottleneck in extreme conditions. Batteries can bridge ramps but do not fix week-long drought. The solution set is plural and uncomfortable: overbuild some resources, pay more for verified demand response, harden transmission corridors, and accept that reliability costs real money. Scarcity pricing has to be allowed to bite to attract the optionality the system needs. That is not a “tip,” it is option theory: reliability comes from paying an insurance premium before the accident. Investors chasing narratives should also revisit what counts as defensive. Utilities with diversified portfolios, credible peak programs, and honest intertie exposure may look boring but will matter more when the fat tail shows up. Data centers boasting green power purchase agreements still need physical redundancy and grid services; glossy certificates will not cool a server hall at 6 p.m.
A resilient grid is less like a glass sculpture and more like a well-designed ship: compartmentalized, redundant, and able to survive hull breaches. That implies more islanding capability, microgrids around critical services, and protection schemes that localize faults instead of broadcasting them. It also implies planning that respects uncertainty. Weather normals from the last century are not a guide to the next decade. Reliability standards built around N-1 need to reflect N-1-1 events — multiple hits in compressed time — and the real-world correlation that climate change is introducing. Canada’s dependence on hydropower and cross-border flows is not a weakness by itself; it becomes one when the rest of the stack assumes smooth water and friendly neighbors. The market can price this if regulators let it. Transparent scarcity signals, capacity constructs that reward performance in extreme hours, and contracts that enforce deliverability across borders even in tense times would push the system toward antifragility. None of this is glamorous. All of it is cheaper than cascading outages and political blame cycles.
Electricity has lulled us with its invisibility. We flip a switch, and an unimaginably complex machine responds in milliseconds. That machine is entering a rougher regime. Heat waves now come with heavier loads, deeper droughts, and stickier geopolitics. The rational stance is not panic; it is humility. Assume stress will arrive when resources correlate the wrong way. Buy slack ahead of the peak. Treat interconnections as valuable but contingent. And remember what history already taught: grids break where we optimize too tightly and diversify too little. If the lights dim this summer, it will not be a surprise. It will be the bill for confusing efficiency with resilience.
