
A single hot week did what years of engineering warnings could not: it proved that a weather-dependent grid cannot survive a heatwave. Below — why incumbent generation fails under heat, and what a water-fuelled, weather-independent platform changes.
Recent extreme heat showed it plainly: a “green” grid does not survive a hot week. In the recent British case, wind output collapsed from the usual ~30% to 6–7% — a fivefold drop — because calm sets in once temperature exceeds the multi-year average by about 5°C. At the same moment solar panels lost efficiency (up to 10% per +5°C above 25°C), gas turbines de-rated (their nameplate is defined at 15°C air), and imported French nuclear had to be curtailed because river-cooled reactors throttle when the water runs warm. All of this happened exactly when air-conditioning demand peaked. A system that has zeroed out coal and leans a third of its generation on “is the wind blowing?” balances only through import cables. One hot anticyclone, and the safety margin is gone.
This is not a British peculiarity. It is a structural disease of any model in which generation depends on weather, season, cooling-water temperature and the political goodwill of neighbours.
The problem: demand scales faster than generation can be built
Two factors that did not exist before now break the familiar picture. First, energy has again become an instrument of pressure rather than a commodity. Second, the explosive electricity demand of data centres: global data-centre consumption is estimated at about 460 TWh in 2024 and is projected to exceed 1,000 TWh by 2030 — comparable to the annual consumption of a large country. In Europe alone, roughly 170 GW of new data-centre projects have been announced — about one third of the continent’s current consumption.
The central issue is rate, not volume. Computing capacity scales in months; generation scales in years and decades. A nuclear block takes 10–15 years and costs $10,000–11,000 per kW, with routine overruns of 2–4×; small modular reactors do not change the arithmetic — a 200 MW load implies $2–2.2 billion in capital cost. And weather-dependent renewables run into the physics of availability: solar stops at night, wind stops in calm, the real service life of an onshore turbine is 10–15 years rather than the nameplate 25, and unsubsidised payback does not arrive within the equipment’s life.
Why storage does not rescue the model
The core technical dead end of the “green” model is that once intermittent generation exceeds about 15% of a unified grid, the whole network is stressed: you cannot predict whether the power will be there when it is needed. The proposed answer is storage. But storage does not generate energy — it only shifts it in time, while degrading, requiring replacement every 7–12 years, and depending on lithium, nickel, graphite and rare-earth supply chains against a lithium-ion recycling rate of about 5%. These systems do not live without continuous state support.
In other words, weather dependence is simply swapped for dependence on scarce materials. The problem is not solved — it is relocated.
The Arbok answer: LongBattery (ALB) — electricity from water
Arbok LongBattery is a modular power plant in a standard 20-ft container whose working medium is water — fresh, sea, or treated municipal wastewater. Inside the unit, water is decomposed and the resulting gases perform mechanical work that is converted to electric current through electromagnetic induction; the process runs continuously. Hydrogen, a by-product, passes through an in-house fuel cell (50–60% efficiency) and raises output, while oxygen is recovered as a marketable industrial commodity.
The decisive difference from everything that “collapses in the heat”: the process runs inside the container and depends on neither wind, nor sun, nor air temperature, nor wars, nor radiation. There is no calm that erases a third of output. No night that switches off panels. No warm river water that forces reactors offline. The “fuel” is water — available everywhere.
Per module (calculations and prototype data):
- base power ~1 MW, up to 2 MW continuous with the fuel cell, 24/7;
- capacity factor ~90% (for comparison, solar is 15–18%), annual output 15.5–15.8 million kWh;
- water consumption ~500 l/h (1 tonne per 4 MWh) — genuinely consumed, which is why it counts as fuel;
- footprint ~14 m² (two parking spaces), service life ~25 years, noise 40 dB;
- emissions CO₂ < 0.1 kg/t, zero SOₓ, no liquid tailings, no radiation; grid connection optional.
By areal energy density ALB reaches about 1,100,000 kWh/m² per year — roughly 4,000–5,000× solar and 10–20× nuclear including facility footprint. A 2.5 GW plant fits on a footprint smaller than a soccer field, on-site. A 200 MW data-centre baseline needs about 100 containers placed next to the load: minimum transmission losses, maximum reliability.
And a direct answer to the “heat” problem specifically. ALB co-generates heat (~185 kW per MW of electrical capacity), usable for district heating in winter. In summer the solid-state TELORISE module converts that heat back into electricity — so extra output arrives exactly when air-conditioning drives the peak. Where a conventional system folds in the heat, this one gains power instead.
Economics: zero client capital cost
The commercial model is BOOM under a Power Purchase Agreement (PPA): client capital cost is zero, and the client pays only per kilowatt-hour at a fixed regional price (5-year fixation, then a 15-year horizon). No transmission lines, substations, exclusion zones or nuclear licensing are required. First modules deploy in 6 months, a 200 MW plant in 1–3 years. On-site installation takes 1–2 days with no infrastructure rebuild.
Each megawatt of ALB displaces 1.6–2.46 million m³ of natural gas per year; a 1 GW cluster displaces €160–740 million of gas annually. By Arbok’s estimates, savings for the client reach €3–5 million over 15 years per unit. The price of energy is fixed for the whole term — no oil spikes, lockdowns or external shocks.
Where it changes the rules
Begin where the demand-versus-build-rate gap is sharpest — with data centres: they need continuous 24/7/365 power with no storage, and they need it in 2026–2027, not by 2040. The second entry point is utilities and district heating: a 200 MW plant yields on the order of 250 million kWh of heat per year — enough for a settlement the size of Monaco. Then critical infrastructure: hospitals, airports, rail hubs, military bases, water supply, and remote sites where the grid is weak or absent.
There is a strategic conclusion, too. A modular architecture means no single point of failure: a fleet of a hundred containers keeps running when individual units fail, and induction-based generation carries no explosion risk — resilience against fuel-price exposure and against drone attack alike. And when energy can be produced locally, modularly and from water, the very rationale for resource-control conflicts weakens: energy shifts from a scarce geopolitical resource to a distributed industrial product.
The choice
The heatwave showed the honest picture: you can keep building systems that fall to zero in a calm, choke on warm river water and hang on an import cable — and pray to the weather every hot week. Or you can bring a container to the site, connect it to water, and get a steady megawatt independent of wind, sun, temperature, wars and radiation.
The right picture is not the one that looks good on a nameplate-capacity chart. The right one delivers the same kilowatt-hours at 15:00 on 15 January and at noon on 15 July.
👉 Writing about the future of energy, building a data centre, or responsible for critical infrastructure? Message me — ARBOK is ready for joint ALB pilots under a PPA model.
