Climate & Energy · Wednesday, 3 June 2026
01 · Briefing · what happened
The grid becomes the climate story: cheap clean power, nowhere to put it
Brazil is throwing away clean power it can't use, the EU is racing to rein in data-centre demand, and a record-warm May reminds everyone why the wires matter.
Key takeaways
- The grid, not generation, is now the climate story: Brazil is throwing away clean power because the wires can't carry it to where it's needed.
- The EU is racing to rein in data-centre power demand, and where demand wins, old fossil plants come back online.
- A record-warm May is the reminder of why moving cheap clean power actually matters.
title: “The grid becomes the climate story: cheap clean power, nowhere to put it” summary: “Brazil is throwing away clean power it can’t use, the EU is racing to rein in data-centre demand, and a record-warm May reminds everyone why the wires matter.”
The cheapest part of the energy transition is now building the clean power. The hardest part is the wires, switches and storage that move it. Three stories today — from Brazil, Brussels and a nuclear plant in Pennsylvania — are really one story: electricity supply and demand are both surging, and the grid in the middle is the bottleneck. A record-warm May across western Europe is the backdrop, a reminder of why all this is happening at all.
The clean power Brazil is throwing away
Atlas Renewable Energy, one of South America’s largest clean-power generators, has frozen $1 billion of planned investment in Brazil — about 1.5 gigawatts of solar and wind it had lined up to build
This is curtailment: when a grid operator switches off working solar or wind because it can’t move or store the electricity right then. The power is there; the wires to carry it aren’t. It’s clean electricity wasted, and a signal the grid needs more lines, more storage, or smarter timing.
Atlas says curtailment hit 15-25% on its existing Brazilian projects in the spring quarter
Brazil isn’t alone. The same bottleneck is biting in Australia, Japan, India and Chile
Brussels goes after the demand side
If Brazil shows supply outrunning the grid, the European Union is now worried about demand doing the same. On Wednesday the EU said it will set minimum energy-efficiency rules for data centres — the warehouses of computers behind cloud services and AI
The numbers explain the alarm. EU data-centre capacity is expected to more than double, from 12 gigawatts last year to 28 GW by 2030, pushing their share of Europe’s electricity past today’s 2.5%
The plan includes a sustainability label forcing big data centres to disclose their water use and how clean their power is
When demand wins, old plants come back
What happens when demand growth wins the race? You restart things you were about to shut down. Two American stories this week show it.
Constellation Energy got a boost on Monday for its plan to restart the Crane nuclear unit — the reactor formerly known as Three Mile Island Unit 1
The other approach is to use the grid you already have more cleverly. Google is funding a 100-megawatt virtual power plant in the PJM region, which runs the grid across 13 mid-Atlantic and Midwest states
For anyone with an electricity bill, these two moves are the visible edge of the same squeeze: the grid is tight, so utilities are paying to keep old plants alive and to recruit your appliances into balancing the system.
The wind fight goes to court
Not every part of the transition is moving forward. Seven states sued the Trump administration on Tuesday over a deal to kill a major offshore wind project
The reason all of this matters: a record-warm May
Behind the grid scramble sits the thing driving it. Jersey, in the Channel Islands, recorded its warmest May since records began in 1894 — the monthly average 0.7C above the old record, with a 34.2C day on 26 May, the hottest May or June day the island has ever logged
A quieter story makes the heat harder to track. A long-running US ocean observatory that gathers climate data is going dark under federal funding cuts, costing scientists a continuous record they can’t recreate
02 · Lesson · why it matters
The bottleneck always moves — solve one, and the constraint jumps somewhere else
When you fix the slowest part of a system, the system doesn't get fast — a new slowest part takes over, and that's where the next fight happens.
A strange kind of waste
In Brazil this spring, a company built solar farms that worked, then watched the grid refuse to take 15 to 25 percent of the power they made. Not because nobody wanted it. Because the wires couldn’t carry it that moment. So it was switched off — clean electricity, made and thrown away.
That should feel odd. We spent a decade making solar and wind cheap, and the result is power so abundant in places that the grid chokes on it. The thing we worked hardest to fix — the cost of clean generation — stopped being the problem. Something else became the problem instead.
This is the pattern worth carrying: in any system that moves something from one end to the other, there is always a narrowest point. The whole system runs at the speed of that point. And the moment you widen it, the bottleneck doesn’t vanish. It moves.
Why fixing the slow part doesn’t make things fast
People expect systems to add up. Make the generation cheaper, the system gets cheaper. Make one step faster, the whole thing speeds up.
Systems don’t work that way when the parts depend on each other in a line. Power has to be made, then carried, then used — each step feeding the next. The line can only run as fast as its slowest step. For years the slow step was making the power. We poured effort there, and we won: solar undercuts new gas across most of the world.
But the carrying step didn’t get any wider while we worked. So the instant generation stopped being the limit, the wires became the limit. The effort moved; the constraint sat still and waited to take over. Brazil didn’t fail to build solar. It succeeded so well that it slammed into the next wall.
The same wall, four ways
Watch the same story repeat across today’s news, each actor hitting it from a different side.
The European Union looks at data centres — the warehouses of computers behind cloud services and AI — and sees demand about to double by 2030. Their worry isn’t generation or even wires. It’s that demand will outrun everything, so they’re writing rules to slow the demand down. They’re trying to manage the bottleneck before it forms.
In Pennsylvania, a company is restarting an old nuclear reactor, and a utility is paying to keep a plant alive past its retirement, because demand is winning the race and they need raw supply fast. Google, meanwhile, is funding a “virtual power plant” — thousands of home batteries and thermostats linked by software to act as one plant. That’s not building a new link. It’s getting more out of the link you already have.
Four moves: slow the demand, add supply, stretch the existing network. All four are arguments about where the narrowest point is and who should pay to widen it.
The constraint is where the money and the fights live
Notice where the pain landed. Not on the people making power — on the ones whose power got rejected. A Brazilian generator that gets curtailed still has to honour its contracts, so it buys replacement power at twice what it agreed to sell for. The bottleneck didn’t just waste electricity. It moved the financial risk onto whoever sits downstream of it.
This is why the bottleneck is always where the conflict is. The narrowest point is the most valuable point — control it and you control the whole flow. The lawsuit over the cancelled offshore wind project is a fight over a future bottleneck: who gets to build the supply, and whether public money can be spent to stop them. Seven states didn’t sue over a price. They sued over who decides what gets built.
When you see a system under strain, follow it to its narrowest point. That’s where the value concentrates, where the failures show first, and where the people with power spend their effort — either widening it or guarding it.
What you now see
The next time something you care about gets faster, cheaper, or more abundant, don’t expect the whole system to follow. Ask what the new slowest part is, because there is one. Cheap solar revealed the grid. A wider grid will reveal storage. More storage will reveal something after that.
The work is never finished, because finishing one constraint promotes the next. That isn’t failure. It’s how systems made of dependent parts behave — and once you see it, you stop being surprised that the cure for one shortage is the cause of the next.
03 · Lab · your turn
Where's the Bottleneck
Rehearse spending a limited budget to widen a power pipeline, and feel the bottleneck jump to a new stage each time you fix the slow one.