Two gates stand between the glut and the demand that drains it

Change 1 · the second gate

The price gate rations by cost and clears continuously. The connection gate rations by permit and by waiting time. A project that passes the first can fail the second, and no amount of surplus power changes that, because the binding resource is transmission capacity rather than energy.

Outright refusal is the rare form. Energimarknadsinspektionen examined Svenska kraftnät and six regional network companies on 11 June 2026 and found connection times routinely exceeding the two-year statutory limit, particularly for large projects. It documented delay rather than denial. Delay is still a filter, because it selects for whoever can hold capital idle longest.

The clean case is Swedish. Power2Earth, the Lantmännen, Fertiberia and Nordion fossil-free fertiliser plant, was shelved in autumn 2025 after Svenska kraftnät refused the 4 to 5 TWh it needed. Per Arfvidsson of Lantmännen described the refusal as concerning "inkopplingen och kapaciteten i det svenska elnätet, alltså trängsel i infrastrukturen".

Change 2 · where the glut is deepest

Luleå sits in SE1, which spent 65.6% of 2026 so far below the €54 electric-boiler breakeven, the highest share of any bidding zone in the EU and Norway. SE2 follows at 64.3%. On the price gate these are the most favourable conditions in Europe.

That is what makes the case structural rather than anecdotal. If abundant cheap power were sufficient, this plant would be under construction. The constraint the model needs to represent is therefore not scarcity of energy but scarcity of the right to draw it.

Change 3 · data centres act on the queue

Version 4 drew data centres crowding out firm power in general. The mechanism is more specific: they consume connection-queue capacity, which is the input to gate 2. Svenska kraftnät took 43 applications in 2025 covering over 9,000 MW of withdrawal, about half of it data centres, and granted 3,695 MW of consumption across 22 positive decisions. Roughly two applied megawatts in five got through.

Data centres hold 5,000 to 6,000 MW in the queue with about 80% of it in SE3. Set against roughly 7,000 MW of new industrial demand expected across the whole of 2026 to 2030, a single customer category now matches five years of industrial growth.

Change 4 · the price channel, and why it points south

The queue is not the only way data centres crowd out industry. Because a data centre is capital-intensive, electricity is a small share of its costs, so it can carry a price that would close an industrial plant. Pär Holmberg of Sweden's IFN makes the point directly: the build-out continues even as it raises the price, and the pressure falls on southern Sweden, where broadband draws the data centres and there is no surplus to absorb them, rather than on the cheap north. This is the same channel visible in PJM, where data centres drove 63% of the 2025/26 capacity-price increase.

The two channels therefore have two geographies. The connection gate bound in the north, at Luleå, where the glut is deepest and the queue is the constraint. The price channel binds in the south, where the data centres actually want to sit and the surplus that would blunt them does not exist. A model that drew only one would miss half the map.

Change 5 · the queue has an allocation rule, and it is a lever

Gate 1 has policy inputs; gate 2 needed one too. Sweden allocated connections first come, first served until 2024, which rewards whoever can apply earliest rather than whoever produces most value per MWh. Under that rule a category of new load with committed capital is structurally first in line, because nothing else is being ranked.

The maturity requirements introduced in 2024 replaced it. They alone shrank the queue by 14,300 MW, and about 20,000 MW of speculative bookings have now been cleared. Svenska kraftnät is going further with capacity zones and business matching, announced 4 May 2026.

Whether this removes the capital bias or formalises it is an open question rather than a finding. Documented milestones, permit progress and a detailed energy analysis are easier to produce for a financed hyperscaler than for a municipal heat utility or a farmer-owned joint venture. The rule now selects for project maturity, and maturity is partly a function of access to capital.

Change 6 · Germany is a completed transition

Energy-intensive production ran 17.8% below 2021 in 2025 and has stayed there for four years. Version 3 and version 4 described this as an ongoing shock being absorbed, which understates it. R1 should be read as having already run once to completion in one large economy, not as a risk under discussion.

Ireland shows the policy response converging on the same instrument from the other side. The moratorium that began in November 2021 was lifted in December 2025 and replaced by a requirement that new connections self-firm with on-site generation or storage, which is a connection gate with a price attached rather than a refusal.

The same instrument is converging across jurisdictions. Holmberg proposes that southern Swedish data centres be required to sign long-term contracts for new unsubsidised generation, and the US administration is encouraging data centres to fund their own supply. Add capacity in step with load and the price channel closes on its own.

Change 7 · the queue is a biased sample

Data centres are the fastest-growing load by a wide margin. The IEA puts their consumption growth at around 15% a year to 2030, "more than four times faster than the growth of total electricity consumption from all other sectors". They are still only about 10% of European electricity demand growth, against nearly half in the United States. Growing fastest and constituting most of the growth are different claims, and in Europe only the first holds.

What accelerated in Sweden is the queue rather than the meter. SE3 applications for large data centres went from about 1,310 MW in 2024 to 6,692 MW in 2025. Consumption sits near 4 TWh, and 2030 forecasts run from RISE's ~5 TWh and Energimyndigheten's 4.4–5.0 TWh to 14–15 TWh derived from the Nordic operators' 7 → 25 TWh path.

So half the queue and a tenth of demand growth are both true, because the queue only sees loads arriving as discrete blocks needing new firm capacity. Heat pumps and boiler retrofits grow behind existing connections and never enter it. Gate 2 therefore selects for lumpiness before anyone acts strategically, which no allocation rule can correct. The mechanism is not Swedish either: EU connection waits run two to ten years, and seven to ten in the FLAP-D hubs.

Change 8 · scarce flexibility does not mean cheap power

The model has been open to an obvious objection: if cannibalization pushes prices down, that is a transfer to buyers rather than a loss. Geis, Lindner and Brown close it. Modelling the German sector-coupled system to 2045, their constrained-flexibility case produces a very high count of zero-price hours and an average electricity price 14–17 €/MWh above the base case, because backup must still cover the peak. Their high-flexibility case leaves few zero-price hours (7%) and the lowest average price, 3–5 €/MWh below base, with system costs 3.3% lower and 22 GW less backup. Throttling B2 widens the distribution at both ends rather than making power cheap.

The same paper qualifies whom the model casts in B2. In their 2045 base case daily flexibility comes from stationary batteries (38%) and flexible EV charging (30%), with resistive heaters at 7% and electrolysis 6%. A cost-optimal system does not reach for industrial heat first.

Heat enters on a condition. Heat pumps and resistive heaters without thermal storage raise winter flexibility needs; in district heating, where storage exists, resistive heaters provide flexibility year-round. Storage is what turns heat from a consumer of flexibility into a supplier of it. Note also the inversion: the low-flexibility scenarios build more resistive heaters, as cheap substitute flexibility, while the high-flexibility one builds fewer and runs more heat pumps, which are efficient but less flexible.

Change 9 · a flat tax puts a floor under the glut

Gate 1 has treated the electricity tax as something that lowers the breakeven. Fernqvist, Broberg, Torén and Svensson identify the sharper property: Sweden's electricity tax is static, "a set price per kWh (0.36 SEK/kWh in June 2022), independent of the electricity price". A per-kWh charge that does not move with the market puts a floor under what a district-heating operator pays. However deep the glut goes the delivered cost stops falling, and negative prices, the strongest possible signal to consume, are invisible at the meter. The tax does not only shift the threshold; it truncates the signal that B2 runs on.

Their capacity figures set the scale. Of Sweden's 305 largest district-heating systems, 72 have heat pumps or electric boilers, totalling 1,207 MW of heat pumps and 1,151 MW of electric boilers, a combined maximum electrical draw of 1,540 MW. Between 2017 and 2021 those assets supplied an average of 7.6% of heat production. The estimated power-to-heat potential is 0.2 to 8.6 TWh, a range wide enough to be an admission that nobody knows.

The storage that would let the plant run on the market's schedule is also small and short. Swedish district heating held roughly 150 TJ in 2016, about 96 TJ of it available for additional load variation, which is around 42 and 27 GWh. That is a day, not a season. The authors state plainly that against Denmark and Germany the opportunity is unexploited, and that the literature does not support an accurate national flexibility estimate.

Change 10 · the gate was open once, and the wedge is measurable

Sweden ran this experiment already. Power Circle records about 130 electric boilers in the district-heating system, roughly 1,300 MW installed, and puts their peak at 1990, when 6.3 TWh of electricity went into district heat. Today they run at 0.1 to 0.4 TWh a year. B2's load did not fail to appear; it was built, and then it stopped.

Change 9's static tax can now be priced. An extra MWh into an electric boiler costs 1,264 SEK/MWh once spot price, energy tax and network fee are counted, and 694 SEK/MWh with the tax removed. The wedge is about 570 SEK/MWh, close to half the delivered cost, and it does not shrink when the spot price falls.

Of that 1,300 MW, some 30 MW is prequalified to the ancillary-service markets. Heat pumps contribute about 300 MW, "främst till FCR". The resource is installed, price-responsive and almost entirely unused.

The exception proves the mechanism. Swedavia's aquifer store at Arlanda has run since summer 2009 and cuts purchased energy by 19 GWh a year, 4 GWh of electricity and about 15 GWh of district heat, equal to 2,000 houses. It cools terminals, melts snow on the stands and pre-heats ventilation air by moving water between seasons rather than by converting electricity into heat. A store that barely draws power is not standing at gate 1 at all.

Change 11 · the affordability loop

R1 sent cheap power to industry through flexibility alone. The older claim, that cheap energy is the backbone of an industrial economy, is a second and more direct route, and the diagram above draws it. A new quantity, the industrial power price, is the all-in delivered cost a smelter pays, spot plus tax plus tariff, and it is not the capture price: the same glut lowers what the producer earns and what the buyer pays at once.

The loop is balancing. A deeper glut cheapens the industrial price, which keeps industry viable, which sustains value-added, which funds the flexibility that drains the glut. The surplus has a second absorber, the industrial economy its own cheapness keeps alive.

What the drawing adds is the condition. B3 runs through the same two gates as B2: the flat electricity tax floors the industrial price so it stops falling with the glut, and the queue throttles the return leg. Germany is the loop run backward once, energy-intensive output at 82.2% of 2021 after the gas shock; Sweden is the loop held open but not conducting, cheap power that never reaches the load. The affordability link is strong only where power is a large share of cost, which is why it attaches to industry and not to the data centres of the fork.