Industrial Heat Electrification

This brief owns the heat layer of industrial decarbonisation treated as one system: high-temperature heat pumps, electric resistance and electrode boilers, electrothermal storage (the Rondo and Antora class of “heat batteries”), the steam systems all of them must feed, and the tariff structures that decide when any of them runs. The scope cut against the neighbouring briefs is explicit. What happens inside the processes — steel chemistry, cement chemistry, the green premium nobody pays — belongs to zero-carbon industrial systems. What a stored megawatt-hour costs a grid, and how storage fleets are dispatched and paid, belongs to energy storage revolutions, which covers the same Rondo and Antora plants from the other side of the meter. Bulk delivery of the electricity belongs to energy corridors. The arbitration rule where the briefs meet: if the question is what a stored kilowatt-hour costs and earns, it belongs to the storage brief; if the question is what a tonne of steam costs, at what temperature and pressure, and what machine made it, it belongs here.

Established Heat is the largest single use of energy in industry, and temperature is the organising variable. Process heat is roughly two-thirds of industrial final energy use, and industry roughly two-fifths of the world total, so industrial heat is on the order of a fifth to a quarter of global final energy. Every claim in this subject is indexed to a sink temperature, and any statement about “industrial heat” that does not carry one is not yet a claim.

Established The one-sentence version of what follows. Below about 100 °C, electric heat pumps are mature and often already cheaper to run than gas. Between 100 and about 165 °C — the bottom of the steam raster — the machines exist commercially but thinly, with measured coefficients of performance between about 2 and 3.5 depending on temperature lift. Above that, electrification today means a resistor: electrode boilers to 70 MW are decades old, and refractory-brick and solid-carbon storage now holds heat at 1,000–2,400 °C in first commercial plants. At no band is device physics the binding constraint. The binding constraints are the ratio of industrial electricity to gas prices, the size and timing of the grid connection, and the near-total absence of independently measured operating data.

Frontier The evidence base is asymmetric, and this brief says so before using it. The heat-pump record is published but small; the electric-boiler record is long but mostly Nordic and rarely written down; the thermal-battery record is one to two years old and comes almost entirely from vendors and trade press. Where a number below is a company's own, it is marked. This pack's web access was cut off mid-research; sources it could not fetch are cited by name.

Frontier The demand structure: roughly a quarter to a third of industrial heat is below 200 °C, another quarter between 200 and 500 °C, and 40–50% above 500 °C. The decomposition varies by author and boundary — the European survey work of Naegler and colleagues (2015) and the Agora Industry / Fraunhofer ISI assessment of 2024 draw the bands differently, and the United States mix is lighter in primary metals and heavier in food, paper and chemicals — but every serious decomposition agrees on the shape: a very large low-temperature block dominated by steam and drying, and a high-temperature block concentrated in a few sectors that the industrial-systems brief owns. This brief's centre of gravity is the block below about 400 °C, which is most of the addressable heat and nearly all of the near-term market.

Established Below about 100 °C the heat pump case is closed as engineering. Ammonia and transcritical CO2 machines supply hot water at 90–100 °C at coefficients of performance of 3–6 in dairies, breweries and food plants, in thousands of installations. Mechanical vapour recompression — a heat pump using the process vapour itself as working fluid — has run for decades in evaporation and distillation with effective COPs of 10–30 at lifts of 5–20 K, and is the quiet proof that electric compression heat is not exotic: it is already standard wherever the lift is small.

Frontier Between 100 and 200 °C the machines exist but the measured record is thin. The IEA Heat Pumping Technologies programme's Annex 58 inventory (2023) lists roughly three dozen commercial or near-market high-temperature heat pumps with sink temperatures above 100 °C, thinning sharply above 140 °C, with the highest commercial sinks clustering near 165–185 °C. The best-documented demonstrations are European: the DryFiciency project ran industrial-scale machines at sink temperatures up to 160 °C in a starch plant and a brickworks, with reported COPs in the 2.2–3.3 range depending on lift. Frontier Manufacturer datasheets go further — Kobe Steel's steam-generating machines quote a COP near 3.5 for 120 °C steam and near 2.5 for 165 °C steam from a 65–70 °C source (vendor figures) — and no independent programme systematically meters installed fleets against them. Established The physics sets the envelope honestly: an ideal machine lifting heat from 40 °C to 140 °C has a Carnot COP of about 4.1, and real machines deliver 40–50% of ideal, so a COP of about 2 at 100 K lift and about 3 at 60 K is what thermodynamics itself predicts. The reported numbers are therefore plausible; what is missing is not credibility but coverage.

Established Above the heat-pump envelope, the electric boiler is old technology with a long operating record. Electrode boilers convert electricity to steam or hot water at better than 99% in single units from a few megawatts to about 70 MW, connect at medium voltage, and have run for decades in Norwegian and Swedish pulp and paper mills as interruptible load soaking up hydro surplus — the original flexible industrial electro-heat, running when power is cheap and standing down when it is not. A COP of 1 is not a defect awaiting a fix; it is the price of temperature reach and capital simplicity, and it moves the entire question onto the tariff.

Established Electrothermal storage crossed from pilot to first commercial scale in 2025–26, and the delivered examples are specific. Rondo Energy's 100 MWh refractory-brick battery at Holmes Western Oil in California entered commercial operation on 16 October 2025, charged by on-site solar, storing heat above 1,000 °C and delivering steam alongside the site's existing gas boilers, following a 2 MWh demonstrator in 2023. Antora Energy's solid-carbon plant at Big Stone City, South Dakota — 50 MW and 5 GWh, carbon blocks heated toward 2,400 °C — was built in under twelve months, financed by a single external investor, and delivers process heat for bioethanol production rather than electricity to a grid. Rondo and Covestro broke ground on a further 100 MWh unit at Brunsbüttel in Germany, financed with €75 million from a philanthropic climate fund and the European public investment bank, targeted at end-2026 commissioning and about 10% of the site's steam. Frontier The state of the fleet is thinner than the coverage suggests: Rondo's operating fleet totalled about 132 MWh when the Covestro project broke ground; Antora's own 19 May 2026 status for Big Stone was “delivering energy” with full operation expected later that year; and no plant in the class has published round-trip efficiency or delivered-energy data measured by anyone but the vendor. Handwave Rondo's claim of above-97% round-trip efficiency is an electricity-to-heat number from a company release with no cost figure attached; quoted beside electricity-to-electricity storage efficiencies, as it routinely is, it is a category error. Behind the two leaders sit a dozen variants — crushed-rock, molten-salt and electrified-firebrick systems from Brenmiller, Kyoto Group, Electrified Thermal Solutions and others — at demonstrator scale or below.

Established The operating economics reduce to one ratio. A gas boiler at 90% efficiency is beaten on running cost when the coefficient of performance exceeds 0.9 times the ratio of electricity price to gas price per unit of energy. At a ratio of 3 the breakeven COP is 2.7, which 120 °C-class machines meet; at a ratio of 5 it is 4.5, which nothing beyond trivial lift meets, and an electrode boiler or heat battery at COP near 1 then needs cheap hours, not cheap averages. Frontier The ratios themselves: recent United States industrial averages sit near 5–6 (electricity around $80/MWh against gas near $15/MWh); Germany's mid-size industrial ratio has run nearer 3.5–4.5 including levies; the Nordic ratio has historically sat below 2.5, which is exactly where the electric boilers are. These move yearly and are stated as bands, not points. Established The cheap hours the flexible machines need are measurably multiplying: the IEA's Electricity 2026 price work records France, Germany, the Netherlands and Spain each near 6% of hours at negative prices in 2025, up from 3–5% in 2024 — and also records negative-price frequency in California and Texas falling as battery fleets charge through midday, a warning that the resource electrothermal storage is designed to harvest is contested and self-eroding.

Frontier The United States ran the largest deliberate experiment on this subject and then partially cancelled it, and both halves are evidence. The Department of Energy's Industrial Demonstrations Program selected 33 projects in March 2024 for up to $6.3 billion in federal cost share, including a band of process-heat electrification awards; the largest food-sector example was Kraft Heinz at up to $170.9 million for heat pumps, electric boilers and related systems across ten plants. In May 2025 the department terminated 24 awards worth about $3.7 billion, with press reporting placing the Kraft Heinz award among them; in October 2025 a broader termination wave swept the demonstration portfolio, including all seven regional hydrogen hubs and the $8 billion behind them, the department itself hedging that no further determinations had been made. By April 2026 some terminations elsewhere in the portfolio had been confirmed reversed — direct-air-capture hubs and five hydrogen projects, $1.2 billion restored. The final state of the heat-specific awards is genuinely unresolved on the evidence this pack could reach, and this brief says so rather than assuming either way. The programme's project-level failure record — Cleveland-Cliffs repurposing a $500 million award, Sublime Systems' layoffs, Heidelberg's cancelled capture plant — is examined in zero-carbon industrial systems and not re-argued here.

The open questions are measurement questions before they are invention questions, and each one is stated with what would answer it.

Frontier What COP does a 150–200 °C steam-generating heat pump deliver over a year of real duty? Every published number is a commissioning point or a campaign average; industrial steam demand cycles daily, waste-heat sources drift seasonally, and defrost, standby and part-load penalties live precisely in the gap between datasheet and year. No fleet-metering programme exists anywhere, which means the single most decision-relevant number in the subject — annual delivered COP at the header, by band — is currently unknowable from public sources.

Frontier What do the heat batteries actually deliver? Realised round-trip efficiency under commercial duty, standby losses per day, availability against the host's schedule, and — the commercially decisive one — the realised charging price: what fraction of megawatt-hours were actually bought in the cheap tail the business model assumes. Vendors state the first and none of the rest.

Frontier How much flexibility can a steam host really sell? A plant that stands its boiler down for the evening peak is selling its process margin; the interruptible record from Nordic pulp mills suggests the answer is “a lot, at a price”, but no modern study measures forgone-production cost curves for steam hosts outside that niche.

Speculative Can compression close the 200–400 °C gap? Water itself (R718) as the working fluid, in multi-stage turbo-compression, is the standing proposal; laboratory and pilot machines sit near 180–200 °C, and materials, lubrication and compressor duty above that are unsolved at industrial scale. It is an open question whether the band is won by compression at COP 1.5–2 or simply conceded to resistive heat at COP 1 plus storage, and the answer is economic as much as thermodynamic.

Frontier Does the Dakotas thermal rider generalise? The tariff innovation that made Big Stone chargeable — a regulator-approved rider under which utility and developer exchange day-ahead data so charging tracks surplus — exists in three adjacent states. Whether an equivalent can be written inside organised wholesale markets, where the same flexibility is already monetised by others, is untested.

Established The spark gap is the master bottleneck. Wherever the delivered ratio of electricity to gas price exceeds about 3, every technology in this brief except low-lift heat pumps loses on running cost, and the ratio is set less by generation than by levies, network charges and taxation stacked on the industrial power price. This is a policy artefact, not a physical constant — the Nordic countries prove a ratio near 2 is compatible with a modern grid — but it binds exactly like physics until changed.

Established Demand charges and ratchets punish exactly the load shape electro-heat needs. United States demand charges commonly run $10–25 per kilowatt-month, assessed on a single peak and often ratcheted for the following eleven months; a 25 MW electrode boiler that touches the system peak once can incur several hundred thousand dollars a month in capacity charges regardless of energy taken. Flexible charging is the remedy in principle and the victim in practice, because the tariff sees the four-times-oversized draw, not the off-peak intent.

Established The interconnection is usually the critical path. A thermal battery charging in a six-hour daily window draws roughly four times the average power of the continuous boiler it displaces, so the feeder, the transformer and the interconnection agreement are sized to the window — and utility connection queues for tens-of-megawatts loads now run years in most Western systems. Rondo's first commercial unit ducked the problem entirely by charging off-grid from dedicated solar, an elegant proof and an admission at once.

Frontier Working-fluid regulation squeezes the heat-pump band from both sides. The leading synthetic fluids for 120–165 °C sinks are HFOs caught in the European F-gas phase-down and the proposed PFAS restriction; the natural alternatives — ammonia, hydrocarbons, CO2, water — each carry pressure, flammability or lift limits. The compressor and fluid supply chain for sinks above 160 °C is a handful of firms.

Established Finance is still concessionary. The flagship European thermal-battery project is funded by philanthropy and a public investment bank, not project finance, and the reason is circular: without independent operating data there is nothing to lend against, and without lending the fleets stay too small to generate the data. Breaking that loop is cheaper than any hardware programme in this brief.

Established The heat layer imports its fuel price from the storage and grid layers. The cheap-hour resource that electrode boilers and heat batteries harvest is produced by renewable overbuild and shaped by the battery fleets documented in energy storage revolutions; the same brief's finding that midday charging is already erasing negative-price hours in California and Texas is a direct constraint on this one's economics. Bulk delivery of new industrial load depends on the transmission build examined in energy corridors, and the demand side — which plants survive to buy electric heat at all — is decided in zero-carbon industrial systems.

Frontier Component dependencies are narrow rather than deep. High-temperature compressors and their working fluids; grid-scale rectification and resistive element supply for multi-hundred-megawatt-hour bricks; refractory and containment materials already standard in steel and glass. Nothing here waits on a scientific discovery. Frontier The one genuine research dependency is thermometric and institutional at once: a standard for measuring and publishing delivered-heat performance — the heat analogue of the settlement data grids publish as a by-product of markets — without which every other dependency is priced blind. Established Tariff design is a dependency in its own right: every operating success in this brief, from Nordic interruptible boilers to the Dakotas rider, was preceded by a tariff written for it.

This subject is unusually testable, and the ranking is clean.

Frontier The decisive result is a year of independently metered operating data from one of the 100 MWh-class thermal batteries now running under commercial duty: tonnes of steam delivered, the realised electricity price paid to charge, round-trip efficiency as operated, and availability against the host plant's schedule. That natural experiment is already underway at Holmes Western Oil and Big Stone City; what is missing is the instrument, because no institution collects delivered-heat data from these plants and no vendor has published any. A single honest year would settle more than every roadmap in the reading list: either it converts vendor claims into bankable numbers, at which point project finance replaces philanthropy, or it shows where the losses and outages live. The asymmetry with grid storage is stark — battery fleets publish dispatch and revenue through market operators as a by-product of settlement; heat delivered behind a fence settles privately and reaches nobody.

Frontier Second: a fleet-metering campaign for high-temperature heat pumps. Fifty installed machines across the 100–165 °C band, metered at the steam header for a year, published by band and lift. The IEA's Annex 58 built the inventory; nobody has funded the meters. This is a single-digit-millions programme that would replace the entire vendor-datasheet evidence base.

Frontier Third: a tariff experiment. One organised-market jurisdiction writing an interruptible electro-heat rate with a demand-charge holiday for verified off-peak charging, run against matched control sites. The Minnesota–Dakotas thermal rider is the uncontrolled pilot; the controlled version would separate tariff effect from technology effect, which no dataset can currently do.

Frontier Fourth, already running whether anyone wants it or not: the Industrial Demonstrations Program as a natural experiment in policy risk. Thirty-three awards made, a tranche terminated, some restorations elsewhere in the portfolio confirmed by April 2026 — the surviving and cancelled heat projects form treatment and control arms for the question of what federal cost-share actually changes. Speculative The experiment nobody has proposed: a paired-site trial holding process and product constant — one plant on a heat battery, its twin on gas — long enough to price reliability, the one variable industrial buyers say they fear and no demonstration has isolated.

Established Integration is a steam-system problem before it is an electrical one. Industrial steam lives on a raster of pressure classes, roughly 4 to 40 bar and 150 to 250 °C saturated. A heat pump that makes 120 °C water does not serve a 10-bar header; it serves feedwater preheat, deaerator duty or a low-pressure branch, and the honest unit of engineering is pinch analysis of the whole site — a method settled since the 1980s — not the device datasheet. Steam-generating machines and vapour recompression tie into the header directly but demand a usable warm source stream within pipework distance, which is a plant-by-plant fact no market study captures.

Established Every first-of-class deployment so far keeps the gas boilers. Rondo at Holmes Western delivers steam alongside the existing fired plant; the Covestro unit is sized to about 10% of site steam. Hybrid operation is not timidity: it is how a site caps demand charges, rides price spikes, and protects uptime while new hardware earns trust. Turndown behaviour, condensate return chemistry, and control handover between electric and fired plant are the real engineering content, all of them disciplines fired plant already practises.

Frontier The battery's discharge side is a fired-boiler impersonation. Brick and carbon systems discharge by driving air or an inert working gas through the hot mass into a heat-recovery steam generator; holding outlet temperature and turndown across a discharging thermal gradient is the vendor-differentiating art, and the 2,400 °C carbon systems add inert-atmosphere containment borrowed from graphite furnaces. None of this is exotic; all of it is proprietary, which is part of why the performance record is closed.

Established The electrical works are sized to the charging window, not the load. Medium-voltage connection, rectification or thyristor control, and a transformer rated at roughly four times average draw for a six-hour-window charger; the civil footprint of a 100 MWh brick unit is a small warehouse. The engineering is unremarkable, which is the point — the schedule risk lives at the utility interface, not inside the fence.

Established The nearest neighbours share hardware, not questions. The identical Rondo and Antora plants appear in energy storage revolutions as long-duration storage assets and here as steam suppliers; the storage brief's unit-of-account discipline (system cost per kilowatt-hour, with technology, duration, market and year attached) is the right template for the delivered-heat metric this field lacks. Energy corridors carries the wires any large-scale electrification of heat presupposes, and zero-carbon industrial systems owns the process-emissions half this brief deliberately excludes, plus the demand-side finding that shadows everything here: buyers will not pay a green premium above roughly 10%.

Established Hydrogen is the counterfactual heat carrier, and the measured comparison is not close. The hydrogen economies brief puts hydrogen heating near 70% system efficiency against 278% for a heat pump, four to six times the primary energy per unit of heat — which is why hydrogen retreats in that brief to feedstock and the highest-temperature niches, and why this brief treats it as a boundary rather than a competitor below 400 °C.

Frontier Two competitors work the same bands from outside the electric system. Nuclear process heat — small reactors co-sited with industry, covered in advanced fission — competes above the heat-pump envelope if it is ever cheap and licensable at industrial sites; and post-combustion capture, covered in carbon capture at scale, competes wherever keeping the flame and burying the flue is cheaper than replacing the burner. Cogeneration cuts the other way: electrifying steam strands the economics of on-site CHP, a large installed base with its own constituency.

Established The most important institutional artefact in the subject is a tariff, not a programme. The thermal market energy pricing rider approved by regulators in Minnesota, North Dakota and South Dakota — under which utility and developer exchange day-ahead data so that battery charging tracks surplus renewable output — is what made the Big Stone plant chargeable; its commissioning was regulatory approval as much as construction. Nordic interruptible-boiler tariffs did the same work half a century earlier. The pattern is consistent: everywhere industrial electro-heat operates economically, a tariff was written for it first.

Frontier Germany is testing the subsidy alternative. The Klimaschutzverträge carbon contracts for difference — first auction decided in 2024, on the order of fifteen projects and €2.8 billion over fifteen-year terms — pay the operating-cost gap rather than the capital, which is the economically literate instrument for a technology whose problem is the spark gap. Results are years away and the figures here are from memory of the award announcements, not fetched documents.

Frontier The American instrument was capital cost-share, and its whiplash is now data. Selection in March 2024, terminations in May and October 2025, partial restorations elsewhere in the portfolio by April 2026: whatever the final ledger, the demonstrated possibility that a signed federal award can be unsigned has entered every boardroom model, and the sibling briefs record the same lesson from the hydrogen hubs and capture hubs. Established The missing institution is the same one the storage brief names: nobody collects operating data. Market operators publish settlement; no equivalent exists for heat delivered behind a fence, and until one does, public money is buying demonstrations whose results stay private.

Established The distributional question is who pays for the grid the factories will use. Industrial electrification at scale is paid for partly through network charges spread across all customers, including households, while the fuel-cost savings accrue to the plant; every jurisdiction that discounts industrial network charges to hold the spark gap down is making a distributional choice it rarely states. The defensible version states it: cheap industrial power in exchange for verified flexibility that lowers system cost for everyone.

Frontier The local environmental case is stronger than the climate case and less often made. Replacing a fired boiler removes on-site NOx, CO and particulates along with the CO2; Rondo's release makes the point in permitting terms — no air permits required — which is vendor framing of a real regulatory fact. For fenceline communities the combustion is the harm they can measure.

Established Workforce effects are modest and tractable. Steamfitting, controls and high-voltage work map closely onto existing industrial trades; the transition risk concentrates instead in fuel-supply chains and in regions whose industrial identity is bound to cheap gas — a political-economy question this brief only flags.

Established Heat is the largest remaining coupling between industry and the flame. Electrifying it roughly doubles to triples a typical plant's electricity demand and, aggregated, adds a load comparable to whole national grids — the demand-side twin of the generation build the energy briefs describe.

Established Cheap heat migrates industry; it always has. Aluminium smelting settled wherever hydropower was stranded a century ago; the same gravity is already visible in thermal-battery siting toward wind-rich, transmission-poor interiors — Big Stone City is the pattern's first data point. A world that electrifies heat is a world whose industrial geography drifts toward its cheapest clean kilowatt-hours, with the trade and security consequences the corridors brief examines.

Frontier The carbon arithmetic is hour-shaped. Electro-heat is only as clean as its charging hours, and flexible charging can be cleaner than the grid average — or dirtier, where the cheap hours are coal-set. The same flexibility that fixes the economics fixes the emissions accounting, which is why tariff design and carbon accounting converge on the same requirement: measure when the electrons were bought.

Horizons track delivered, measured heat — not announcements, and not selections.

• 10 yr: Frontier Heat pumps default for new low-temperature process heat in markets with price ratios below 3; the 100 MWh-class thermal batteries either publish (or leak) a real operating record and reach project finance, or remain a philanthropically funded niche — the fork is evidentiary, not technical. The surviving Industrial Demonstrations awards commission and report, settling what the cost-share bought.
• 25 yr: Speculative Most heat below 400 °C contestable on running cost across the OECD, contingent on spark-gap reform and interconnection reform, neither of which any major economy has yet done deliberately; electrothermal storage a standard utility-tariff category if the rider model generalises.
• 50 yr: Speculative Resistive and stored electro-heat standard up through the 1,000 °C class wherever process chemistry permits, with combustion persisting only where it is feedstock or flame-contact process; industrial siting visibly reorganised around clean-power basins.
• 100 / 250+ yr: Handwave Full electrification of industrial heat, including the process-chemistry bands, is asserted in long-range scenarios by extrapolation; nothing in the measured record either compels or forbids it, and the honest statement is that the 100-year question is owned by the process briefs, not the heat layer.

• Depends on Nothing on this map blocks it in the strict sense of waiting on another brief's result, and this brief says so rather than inventing an edge. Its economics are downstream of price formation described in energy storage revolutions (the negative-hour resource, and the battery fleets already eroding it), its delivery at scale presupposes the transmission examined in energy corridors, and its buyers survive or fail by the demand-side findings of zero-carbon industrial systems — but each of those is a price, a wire or a market fact, not a pending discovery.
• Requires (not on this map) A delivered industrial electricity price below about three times gas per unit energy, the ratio at which the commercial heat-pump envelope wins outright and above which everything here needs special hours; tariff structures that price interruptible charging, of which the Minnesota–Dakotas thermal rider is the only industrial-heat example this research found; independently measured delivered-heat data from the 100 MWh-class plants now operating, which nobody currently collects or publishes; grid interconnection for flexible loads that draw roughly four times their average power, through utility queues currently measured in years; and a compressor and working-fluid supply chain rated for sink temperatures above 160 °C, where the commercial inventory thins to a handful of firms under refrigerant-regulation pressure.
• Enables The energy-emissions half of heavy-industry decarbonisation — the roughly 40% of cement CO2 and the analogous fuel shares elsewhere that are combustion rather than chemistry — and, on the grid side, a gigawatt-class flexible load resource that storage-rich systems will need as their own price signals flatten.
• Adjacent Hydrogen economies (the counterfactual carrier, measured and beaten below 400 °C), advanced fission (the non-electric route to the same temperatures), and carbon capture at scale (the keep-the-flame alternative wherever capture beats replacement).

The claims in circulation, stated and answered.

Handwave “Heat batteries are 97% efficient — better than lithium.” The comparison circulates widely and is a category error: the figure is electricity-to-heat, lithium's is electricity-to-electricity, and a resistor is also “100% efficient” by the first metric. The honest comparison is delivered cost per tonne of steam against a gas boiler, and no public data yet supports it either way.

Established “Heat pumps cannot do industrial temperatures.” False below 165 °C, where commercial machines exist and demonstrations have measured COPs above 2; true today above about 200 °C, where no commercial machine operates. Both halves matter, and each is routinely quoted without the other.

Established “Industrial heat is hydrogen's market.” The measured comparison in the hydrogen brief — roughly 70% against 278% system efficiency, four to six times the energy input — answers this below 400 °C. Hydrogen's real industrial role is feedstock, a different brief's argument.

Frontier “Thermal batteries are proven.” Delivered, yes — two 100 MWh-class plants operating is real and recent. Proven, no: no third-party performance data exists, the flagship European project is philanthropically financed, and the vendors' own status language (“delivering energy”, full operation expected) is commissioning language. The gap between delivered and proven is exactly one published year of data wide.

Frontier “Electric heat is clean heat.” Only as clean as the charging hours. In coal-heavy systems a resistive boiler charged at the margin can emit more than the gas boiler it replaced; the same plant charged flexibly in a renewables-rich system beats it decisively. The claim is not wrong; it is unfinished until it names the hours.

Frontier “The DOE cancellations killed industrial electrification.” Overdrawn in both directions. The terminations were real and partially reversed elsewhere in the portfolio; meanwhile the two largest thermal-battery plants were financed without federal cost-share — one by a single private investor, one by philanthropy and a public bank — and the tariff innovation that matters most cost nothing. Policy risk is now priced in; the sector is not dead, and neither claim should be flattered.

Established “COP is a property of the machine.” It is a property of the machine and the lift together; quoting a COP without source and sink temperatures is the field's most common unit error, and it is load-bearing in most optimistic cost claims. Every COP in this brief carries its temperatures.

Speculative “Efficiency and waste-heat recovery will shrink the problem away.” Recovery and pinch discipline genuinely cut demand — that is established — but the claim that they remove the need for new heat supply extrapolates single-site results to a sector; the measured economy-wide record shows efficiency slowing, not reversing, industrial heat demand growth.