Agora puts Germany's avoidable grid build-out at 66 billion euros

Germany's larger distribution network operators had to file a new round of expansion plans with the regulator on 31 October. The day before, Agora Energiewende published a costing of the alternative: about 66 billion euros less to build by 2045 if heat pumps, electric vehicles and batteries are planned in as flexible load instead of as peaks to reinforce against. In the same seven days to 3 November, Ember put numbers on how much relief data centres can actually hand back to the grids they are straining, and a listed British battery fund worth close to 600 million pounds put itself up for sale while its peers started pricing the second round of capital spending on assets that are already running.

Agora costed the German grid build-out that flexible demand would avoid

The filing is routine and the timing was not. Under paragraph 14d of the German energy act, distribution operators with at least 100,000 connected customers submit expansion plans every two years, which covers about 80 of the more than 850 network operators in the country. The 2024 round added up to 237 billion euros of planned spending to 2045, and it mapped the high-voltage network almost completely while reaching only around 30 percent of the low-voltage network in residential streets. There is no public consultation on these plans, and the Bundesnetzagentur does not confirm them, although it can demand changes.

Agora Energiewende published its own numbers the day before the deadline. The consultancy Consentec and the research institute FfE modelled the German distribution system using representative low, medium and high-voltage networks. Solving every overload with new lines, transformers and substations costs about 248 billion euros to 2045, with the annual bill climbing from 7 billion euros in 2030 to 16.1 billion in 2045. Applying the most effective of the measures studied brings the total to roughly 182 billion, 27 percent lower. Agora puts the saving at around 66 billion euros, worth at least 2 billion a year from 2030 and 5 billion a year from 2035. For scale, the federal government spent 6.5 billion euros in 2026 alone holding network charges down.

The saving splits in a way that matters for anyone planning a connection. About 22.5 billion sits on the generation side and arrives early: connecting more generation than the line can export and curtailing it in the rare full-output hours cuts the expansion needed for new feed-in by 59 percent in medium voltage and 31 percent in high voltage by 2030, and trimming only the midday solar peak cuts the medium-voltage need by a quarter in 2035. The other 43.5 billion sits on the load side and arrives from the mid-2030s, when electric vehicles, heat pumps, electric trucks and electrified industrial processes become the reason to build. One number shows how wide the gap already is: at the eastern operator E.dis, wind and solar developers have applied for connections totalling 110 gigawatts, against a highest load of 2.3 gigawatts across the whole network area.

None of this is law yet. Agora asks the government and the Bundesnetzagentur for price signals that track local network utilisation, for connections sized smaller where flexibility on site carries the peak instead of the network, and for a targeted emergency curtailment power, with large high-voltage batteries paying network tariffs that rise and fall with utilisation. For an asset owner the practical reading is that a smaller connection is turning into a commercial option instead of a penalty. If a site can carry its own peak, that is now worth saying in the connection application, and the curtailment hours being accepted are worth pricing properly rather than assuming them away. A battery sitting on a high-voltage connection should be modelled against a tariff that moves, because that is the direction the regulator is being pushed.

Fusebox replaces the patchwork of site control, portfolio tooling and TSO workflows with one operating layer, which is what makes a tighter connection workable: assets are onboarded faster, real-time dispatch is standardised, and market entry repeats without rebuilding the stack. The platform is hardware-agnostic and connects mixed fleets without forcing a rip-and-replace, and the aggregation layer turns diverse assets into market-ready capacity while the owner keeps strategy and compliance. Existing tools and partners plug in, and modules can be swapped without re-platforming.

Talk to Fusebox about what your flexibility is worth to the network →

Ember found data centres can give the grid more by generating than by shifting

Ember published a different question on 29 October: once a large load is connected, how much of its demand can it actually hand back. Across nine markets, Texas, California and New York in the United States, Ireland, the United Kingdom and Germany in Europe, and Malaysia, the Philippines and Singapore in Southeast Asia, shifting computing workloads would cut peak demand by an estimated 0.1 to 0.6 percent in seven of them. Only about 3 percent of global data centre capacity will be able to take part by 2030, because roughly 87 percent of that capacity runs cloud services, enterprise IT and AI inference, work that has to answer users immediately.

Ireland is both the European exception and the European warning. Data centres there already account for about 19 percent of peak demand, and when the system peaked at 6 gigawatts in January 2025 an estimated 1.1 gigawatts of that was data centre load. Ember projects 1.6 gigawatts against a 6.5 gigawatt peak by 2030, or 24 percent. Even at that share, load shifting is worth about 100 megawatts, 1.5 percent of the projected peak. The timing mismatch is the underlying problem: a large data centre needs 200 to 500 megawatts and can be built in 18 to 24 months, while a conventional grid expansion takes four to ten years.

Generating or discharging on site is the larger lever. For two Texas data centres with a combined projected load of 5.6 gigawatts, Ember finds that on-site capacity could cut grid demand by up to 2.7 gigawatts during a stress event, against about 0.55 gigawatts from load shifting, nearly five times as much. Elisabeth Cremona, energy infrastructures lead at Ember, said that switching to clean on-site generation eases grid pressure without adding to consumer bills. The report also names two ways it goes wrong: if the on-site plant is gas, it locks fuel into the system for decades, and if grid operators ignore behind-the-meter capacity a site already holds when they procure, two sets of assets get built for one block of demand. Ireland has already moved: after consultation, an initial flexibility requirement on data centres was replaced with a requirement that facilities of 10 megawatts or more bring generation or storage capacity of their own.

If you operate C&I loads, on-site PV, batteries or backup generation, the useful lesson is that the regulator is starting to count what is behind your meter. Equipment bought for resilience and never dispatched is capacity you are paying for twice, once in capex and once in the network charges that fund the reinforcement it could have avoided. Metering it, declaring it and putting it under automated control is what turns it from a standby cost into a position. Exploratory questions about how that works in a given market can also go to our knowledge bot, Fusie.

Tell us what is sitting behind your meter →

Gresham House put its listed British battery fund up for sale

On 3 November the board of Gresham House Energy Storage Fund, which trades in London as GRID, instructed its financial advisers to invite offers for the entire issued ordinary share capital, 569,064,139 shares and a market capitalisation of nearly 600 million pounds. The shares rose 10 percent on the announcement. Chair John Leggate said the decision followed a wide range of feedback from shareholders and came alongside continued growth in earnings and a 15.8 percent uplift in net asset value over the first six months of the year. Listed British storage vehicles have traded poorly since revenues fell in 2023 as ancillary services saturated, sometimes below net asset value, and higher interest rates left them unable to raise capital to grow out of it. Peer Harmony Energy Income Trust was sold to Foresight last year, and Gore Street shareholders have pushed for a wind-down.

A day earlier the same publication reported from the Battery Asset Management Summit in London on the bill the operating fleet still has in front of it. Pablo Zalvide, structured finance and M&A lead for UK battery storage at Fotowatio Renewable Ventures, expects a wave of refinancings: even where debt was sized against a 15-year warranty duration, most financing in place today is mini-perm structured at around seven years, so a cohort of projects will come back to lenders with duration assumptions that have aged, and a good share of them will augment at that point. David Harris, senior asset manager at Gore Street Capital, described what that does to operations. A part-augmented site runs in two halves, where the new half can be dispatched at full power for a full hour while the older, more degraded half may manage only 50 minutes, so the energy management system has to telemeter the right values and discharge units at different rates to hold the declared level.

For owners and portfolio managers the takeaway is that the second half of an asset's life is now a financing question as much as a technical one. Build an augmentation assumption into the model at financial close instead of discovering it at refinancing, and check that the control layer can run a site whose halves have different state of health, because a declared capacity the asset cannot hold is a penalty waiting to happen. The same logic applies to a portfolio assembled over several years from several suppliers.

Let's look at how your fleet is dispatched as it ages →

What to watch

• Germany: the new distribution expansion plans were due on 31 October and go onto the shared VNBdigital platform, and the Bundesnetzagentur can require changes to them (cleanthinking).
• Germany: whether the government and the regulator adopt the framework Agora asks for, including locational price signals and smaller connections backed by on-site flexibility (Agora Energiewende).
• Ireland: how the requirement that data centres of 10 megawatts or more bring their own generation or storage is applied to new connection applications (pv magazine).
• Europe: whether system operators start counting behind-the-meter capacity when they procure, which Ember warns is the difference between one set of assets and two (Ember).
• Great Britain: the outcome of the GRID formal sale process, and the refinancing wave that lands when seven-year mini-perm debt meets 15-year warranties (Solar Power Portal).