My first instinct was that Calgary doesn’t need more solar.

If someone proposed building another utility-scale solar farm here, the conventional read would stop the conversation quickly. We don’t need the incremental power, and the price is too low to justify it. Alberta’s own market makes the case. The average pool price fell roughly 53 per cent in 2024, to about $63 per megawatt-hour, and the grid saw 376 hours of zero-dollar power across the year. The previous annual record was 47. That is a market signalling, loudly, that it already has enough.

Then the second thought arrived and undercut the first. If you have the capital to build solar now, why wouldn’t you, as a hedge against future price escalation? And if you keep pulling that thread, you arrive somewhere genuinely uncomfortable for anyone in this industry. Overbuilding solar may not be the mistake our instincts say it is. It may be the natural equilibrium that cheap solar produces on its own. The interesting question is not whether we overbuild. It is who pays for what happens when we do.

Why solar wins the middle of the day

The first angle to consider is dispatch logic, because that part isn’t really up for debate. Electricity markets dispatch the cheapest available generation first, working up the cost stack until supply meets demand. This ordering is called the merit order. The last and most expensive unit needed to meet demand sets the market clearing price, the single price every dispatched generator receives regardless of what it bid. In Alberta, that is the pool price referenced earlier. Solar sits at the very bottom of that stack because its fuel is free, so its marginal cost is close to zero. Whenever the sun is up, solar clears, and it pushes more expensive gas generation down and out of the money. That is not a glitch. That is the merit order working exactly as designed.

Push enough solar onto a grid and the predictable happens. In the middle of the day you generate more than the system can use. California has lived this for years. In just the first four months of 2025, its grid operator curtailed more than 738,000 megawatt-hours of mostly solar generation. Curtailment is simply the deliberate dialling-back of available output when there is nowhere for it to go. Midday net demand (total electricity demand minus wind and solar production at that moment) in California has fallen by roughly 45 per cent since 2020, resulting in the famous duck curve deepening toward a canyon.

Here is the first place our instincts mislead us. We treat curtailment as waste, almost as a failure of planning. In a world of increasing renewable energy, it’s worth questioning whether that reflex still holds. Curtailing surplus energy that costs nothing to produce is the system behaving rationally. So, the real question isn’t how to eliminate curtailment. It’s who carries its cost, and that turns out to be a question about contracts and market design rather than about physics.

The catch the pitch skips

This is where the hedge logic I started with comes apart a little.

Winning dispatch is not the same as earning revenue. In an energy-only market, which Alberta runs and intends to keep running under its Restructured Energy Market, the marginal unit sets the price for everyone. When solar output is abundant at midday, the price-setting unit is often solar itself, the market floor, or even a negative-priced offer. A new solar farm may win the right to generate, but it is generating into a price environment that its own cohort has helped drive toward zero.

The industry term for this is cannibalisation. In an energy-only market it is the very mechanism that disciplines overbuild. Each incremental solar project erodes the value of the hours it depends on for revenue. A merchant project, one that sells its power at the going market price rather than under a fixed-price contract, sees its economics deteriorate as midday saturates. Under this paradigm, building stops making sense the moment the price it actually captures falls below what it costs to build.

Notice what does the disciplining. It is not concern for the gas plant down the road. An energy-only market does not ask a solar developer to weigh the effect on other generators, any more than it asks a gas plant to weigh its effect on a competitor. What disciplines the build is the developer’s own collapsing capture price. Each builder simply responds to its own economics, at precisely the hour it is generating most.

Alberta is already showing the early innings of this. Installed capacity jumped 11.3 per cent in 2024, with nearly 1,900 megawatts of new gas and roughly 1,370 megawatts of new wind and solar, and prices fell accordingly as supply outran demand. We are not theorising about California. We are watching a milder version of the same story unfold in our own province.

So why build at all?

Two answers, and together they are the whole game.

The first is storage. Add a battery, whether co-located with a solar project or stand-alone on the grid, and you are no longer forced to sell into the glutted hour. You charge when prices are low and discharge into the evening peak, lifting your average capture price back into viable territory. This is precisely why California’s battery fleet grew from about 4 gigawatts in late 2022 to more than 11 gigawatts by mid-2024. That build-out did not chase a subsidy. It chased the curtailment. The market routed around its own cannibalisation, and storage was the detour.

The second answer is subtler, and I think more important. It comes down to who holds the price risk. A merchant developer bears cannibalisation directly, so merchant overbuild tends to self-correct. A developer operating under a fixed-price offtake agreement (for example, a power purchase agreement, or PPA) is shielded from spot-market price collapses. When prices crater, the loss is borne instead by the PPA counterparty, whether that is a corporate buyer, a utility and its ratepayers, or, in some cases, taxpayers.

Once that link between generation and market price is weakened, investment and dispatch decisions change. A contracted developer can continue adding capacity and may even bid negative prices to remain online, because the downside is no longer theirs to bear. That dynamic is a major reason heavily contracted or subsidised solar fleets can push wholesale power prices below zero.

Which leads to a reframe I find hard to shake. A PPA is best understood not as a price but as an allocation of risk. The fixed-price buyer is wagering that prices will climb. They overpay if prices stay low, and they are protected if prices spike. The developer signing that contract gives up the upside of a high-price future in exchange for protection against a cannibalised one. The number the two parties settle on is really a negotiation over which future each believes in, and which of them is more afraid of being wrong. Strip away the legal language and a long-term PPA is a shared bet on the structure of the grid a decade, or more, out.

We price the future by assuming the present

That bet rests on something most financial models never say out loud, and it is worth dragging into the light.

In a gas-heavy market, electricity is not really priced so much as it is derived. A long-dated power forward, the price you can lock in today for power delivered years from now, is effectively the market heat rate multiplied by the gas forward curve, plus a carbon adder. The heat rate is just the efficiency with which a gas plant converts fuel into power, so it translates a gas price into an electricity price. Because gas is the marginal, price-setting unit for most hours, and because there is almost no trading liquidity in power a decade or more out, the long end of the power curve is literally constructed from the gas curve. When a developer signs a twenty-year PPA or an analyst runs a project-finance model, they are not really pricing electricity. They are holding a leveraged position on natural gas, wrapped around an unstated assumption that gas stays marginal.

That assumption is invisible precisely because everyone shares it. Regular readers will recognise the question I keep returning to, the one I think of as the legacy-paradigm test. Is this still true, or is it simply how we have always done it? Pull the assumption out and look at it. What happens when gas becomes a minor player on the grid? The pricing method does not merely shift. It breaks. With no fuel cost to anchor to, the price splits in two, sitting near zero when the sun is up and rising toward scarcity levels, set by storage opportunity cost, when it is not. The smooth, gas-linked line becomes jagged, and the thing being priced migrates from energy, measured in megawatt-hours, toward flexibility and firmness, measured in the ability to show up on command. A forward curve built on gas cannot represent that world at all.

Now add the other great uncertainty, which is demand. The cleanest way I have found to hold both unknowns at once is a simple two-by-two. Put the supply regime on one axis and demand growth on the other.

	Modest demand growth	Explosive demand growth (e.g. data centres)
Gas still marginal	Today’s world. Gas-linked curves roughly hold; merchant overbuild self-limits through cannibalisation.	Scarcity returns, prices rise, overbuild is rational again, and thermal plants regain their value.
Solar and storage marginal	The danger zone. Chronic midday zero or negative prices, deep cannibalisation, stranded solar. Storage becomes essential, not optional.	The transition success case. Flexible demand absorbs the midday surplus, capture prices stabilise, and overbuild, storage and responsive load all make sense at once.

Each quadrant produces a different forward curve, a different answer to whether you should overbuild, and a different party left holding a mispriced contract. And notice what large new demand does. A sizeable, even partly flexible load acts as a sink for surplus power. It fills in the belly of the duck, lifts capture prices, and can rescue both solar economics and thermal value at the same time. The most dangerous forecast to build a project on is not the most pessimistic one. It is “modest demand on a gas grid,” because that is both the consensus assumption and the quadrant that punishes overbuild hardest if reality lands somewhere else.

Who pays to turn the sun off?

Which brings us back to the title, and to policy, because the government is simply another player allocating the same risk, with larger instruments.

Every incentive structure is, underneath, an answer to the question of who bears the cost. Line them up by who holds the risk and the picture clarifies. With a merchant project, the developer holds everything, and overbuild self-corrects. A production tax credit, paid per megawatt-hour generated regardless of price, rewards generating no matter what the grid actually wants, which is one reason it can push prices negative. It pays a plant to keep running at the very moment it should ease off. A feed-in tariff sits at the far end of the spectrum, a guaranteed, administratively set price where the ratepayer absorbs almost all of the risk. It is the most overbuild-inducing tool ever deployed. It built Germany’s solar boom, and closer to home it built Ontario’s under the 2009 Green Energy Act, then left ratepayers carrying long-dated, above-market contracts for years afterward. A contract for difference, by contrast, can be designed to be price-aware, paying the gap up to a strike price but withdrawing the reward when prices turn negative, and clawing back the upside when prices run high.

That single distinction, between price-blind support and price-aware support, is the whole lesson. It is not how much we subsidise that determines whether we get useful overbuild or wasteful overbuild. It is how we design the subsidy.

Then there is the hardest version of “who pays,” which concerns the thermal plants being pushed out of the money. Their owners are not wrong to see a stranded-asset risk, and it is worth being clear-eyed and fair about what those plants represent. These are the assets that have kept Alberta’s lights on through every cold, windless January night, and the people who built and run them carry exactly the operating discipline this transition will continue to need. The honest question is how to repurpose them on terms that are fair to the capital already committed and to the ratepayers who would fund any bridge.

One option is a standby payment, paying a gas plant to remain available without burning fuel, a managed decline rather than a cliff edge. China did roughly this for its coal fleet in 2024, introducing a monthly capacity payment to ease plants from baseload duty into a balancing role. The cautionary evidence is just as real. Capacity payments have a long history of quietly keeping uneconomic assets on the books and handing the bill to ratepayers, the stranded-cost problem that has dogged utility transitions elsewhere. And it is worth saying plainly that Alberta looked hard at a capacity market for exactly this purpose and chose against it, keeping an energy-only market with a narrow thirty-minute reserve product instead. Anyone proposing standby payments here is arguing for something the province has deliberately declined, which is not a reason to dismiss the idea but a reason to make the case honestly.

So, who pays to turn the sun off? In an energy-only market, the honest answer is whoever guessed wrong about the future. The merchant developer who overbuilt into cannibalisation. The offtaker who locked a price on the wrong side of the regime shift. The ratepayer funding a price-blind subsidy, or a standby payment that outlived its usefulness. The forward curve quietly assumes one particular future, and someone is left holding the cost when a different one arrives.

The components exist; the design is the hard part

There is one more move that changes the whole picture, and it deserves its own article rather than a paragraph here. Demand does not have to stay still. The duck curve exists today largely because we have bolted a brand-new, variable supply side onto an old and inflexible demand side. But why would anyone charge an electric vehicle overnight if midday power is nearly free? As smart devices let demand chase supply instead of the other way round, the midday trough fills in, the curve flattens, and decades of “shift your load to overnight” logic begin to invert. That is where I want to go next.

For now the takeaway is narrower, and I think more useful. Anyone quoting you a confident power price for 2040 is extrapolating a market structure that may not survive the decade, in either direction. That is not an argument for paralysis. It is the argument for optionality. Storage, flexibility, and contracts that do not stake an entire project on a single quadrant are how you stay solvent when the regime shifts under you.

The components already exist. Cheap solar, capable storage, flexible demand, and a reasonable set of policy instruments are all on the table. The hard part, as it so often is in this business, is the design. The future value of electricity is not a number waiting on a curve. It is a question about who is willing to be wrong, and who can afford to be.