Use What You Own — Interview V

Q 01 · The Register

Germany published a national register of industrial waste heat. What is it, and why does it matter?

You have been working directly with a government dataset most people have never heard of. Start there. What is the Plattform für Abwärme, where does it come from, and why does its existence change the conversation?

It comes from law. § 17 of the Energieeffizienzgesetz — the Energy Efficiency Act — requires any company consuming more than 2.5 gigawatt-hours of energy a year to declare its waste heat: where it is, how hot it is, how much there is, when it is available across the year. The Bundesstelle für Energieeffizienz collects those declarations and publishes them on a platform called the PfA. It is public. No registration, no fee. You can download the whole table this afternoon. The data is licensed CC BY-NC 4.0 — open for non-commercial use with attribution.

The scale is the part people do not expect. More than 24,000 individual sources, from more than 3,000 companies, with named addresses down to the postal code and street, declared temperatures, declared thermal power, and monthly availability profiles. The aggregate is large — on the order of 205 to 254 terawatt-hours a year, depending on the snapshot of the register.

Why it matters is almost philosophical. You cannot connect what you cannot see. For decades, industrial waste heat was invisible — every plant knew it had it, no neighbour knew it was there, and no planner could count it. Germany made a deliberate decision to make the invisible visible, by statute. The register is the first act of any infrastructure: drawing the map. Most countries have never drawn it. Germany did.

You cannot connect what you cannot see. § 17 made the invisible visible — by law. The register is the first act of infrastructure: drawing the map.

Q 02 · The Classification

You classified all 24,000 sources by Thermal Plug compatibility. What does the data show?

You did not just read the register — you ran the whole thing through the Thermal Plug classification. What were you sorting for, and what came out?

The primary divide is not temperature. It is medium. A source that already carries its heat in a liquid — cooling water, process water, steam, condensate, a refrigeration condenser circuit — is the easy connection. You install a heat exchanger and a metering boundary on the outside of the existing pipe. The plant does not change. I call that class TP_DIRECT. A source that carries its heat in gas or air — flue gas, exhaust air, a dryer's hot stream — needs a conversion step first, a gas-to-liquid heat exchanger, before it can hand off to a water-based network. That is TP_INDIRECT. Harder, but still defined.

Of 24,215 valid records: 27 percent are TP_DIRECT — 6,637 sources, 13,160 megawatts of declared thermal power. 30 percent are TP_INDIRECT — 7,151 sources. The remaining 43 percent I labelled TP_UNKNOWN, because their names alone — "BHKW", "compressor", "boiler" — genuinely do not say whether the heat leaves as water or as gas. That label is not a failure. It is the honest limit of free text. Force a guess there and you publish a wrong number.

The single most important finding is in the language of the register itself. A large number of sources are flagged by the companies as nicht nachrüstbar — not retrofittable. But read what that means. It is the company describing its own internal plant: there is no easy way for them to re-plumb their process to recover the heat internally. The Thermal Plug does not ask them to. It is an external interface on a pipe that already runs. The "not retrofittable" label is a statement about the inside of the factory. It says nothing about whether a neighbour can take the heat from the outside. The accessible opportunity is larger than the register's own labels suggest.

The companies that reported their heat as inaccessible were describing their own internal view. The interface doesn't touch their internals.

Industrial plant with piping and cooling infrastructure

[PLACEHOLDER IMAGE — replace before publish.] More than 24,000 industrial sources are declared in Germany's public PfA register. 27 percent carry their heat in a liquid — the simplest connection: an interface on the outside of an existing pipe.

Q 03 · The Grid Is Already There

Germany has roughly 36,900 km of district heating pipe. The consumers are connected. Why isn't the waste heat flowing?

So the heat is documented. But heat needs somewhere to go. What does the receiving side look like — and if it already exists, what is the actual blockage?

The receiving side is enormous and it is already built. Germany has on the order of 36,900 kilometres of Fernwärme — district heating — pipe, across roughly 3,800 separate networks, serving about one in six dwellings in the country. [VERIFY: the 36,900 km figure, and where/how the existing thousands of km were actually built — route, ownership, build era. Cross-check AGFW/BDEW.] The consumers are already connected. The pipes are already in the ground. And here is the pressure point: around 70 percent of the heat in those networks is still generated by burning fossil fuel. The operators are under a legal and commercial mandate to decarbonize. They want the heat. The demand is not hypothetical — it is contractually committed and politically required.

So the blockage is not availability. It is not willingness. It is not the pipes. It is that every single connection between a waste heat source and a network is, today, a custom engineering project. Bespoke pipe sizing, bespoke metering arrangement, bespoke contract, bespoke control logic. Each one starts from a blank sheet. The first connection in a city costs the same to engineer as if no connection had ever been made anywhere. There is no repeatable interface specification — so there is no learning curve, no falling cost, no standard part. The heat is declared, the grid is built, the buyer is mandated — and the connection is still hand-made every time.

I want to be honest about what comes next, because it is not as simple as bolting a part to a pipe. The temperatures have to match, or you raise them with a heat pump. The pressures have to match. The metering has to be fair to both sides. The contract has to hold for twenty years. The series works through each of those in turn — that is the whole point of doing it as a series rather than a slogan. This interview fixes only the starting point: that the connection itself should be a standard, repeatable part, not a fresh invention every single time. Everything else, we solve step by step.

The Register, Classified — PfA Snapshot (24,215 valid records, classified 2026-05-29)

Class	Sources	Share	Declared thermal power	What it means
TP_DIRECT (liquid)	6,637	27.4%	13,160 MW	Water / steam / condensate / refrigeration circuit. Interface on the outside of an existing pipe — no internal modification.
TP_INDIRECT (gas/air)	7,151	29.5%	8,768 MW	Flue gas / exhaust air / dryer. Needs a gas-to-liquid conversion exchanger first, then connects.
TP_UNKNOWN	10,427	43.1%	15,838 MW	Medium not determinable from the source name alone (CHP, compressors, generic "Abwärme"). Honest "insufficient data", not a no.
Total	24,215	100%	37,766 MW	57% classifiable from the declared name alone.

Source: Plattform für Abwärme (BfEE / BAFA), dataset of 2026-05-29, licensed CC BY-NC 4.0. Classification: thermal-plug medium classifier v0.1. Power figures are declared maximum thermal power summed by class. Annual-energy aggregation withheld pending a data-pipeline correction.

"The map already exists. Twenty-four thousand sources, named and located, published by law. The grid already exists — thirty-seven thousand kilometres of it, still burning fossil fuel for want of an alternative. What is missing is not the heat. It is the socket."

Dimitri Wolf

Q 04 · The Proof Cases

There are already live projects — Norderstedt, Berlin, Frankfurt. What do they prove?

This is not all theory. Several connections are running today. What do the real ones actually demonstrate — and what do they quietly reveal about the cost of doing it the current way?

They prove the physics works, the economics work, and the demand is real — and they reveal exactly what makes it slow. Take three. Norderstedt: the municipal utility built its own data center and fed roughly 6.8 gigawatt-hours into its district network, live since April 2024, warming on the order of 350 households. It worked smoothly for one reason — the same organization owned both sides. The shortcut was organizational, not technical. There was nobody to negotiate with.

Berlin, the Pallasseum: Deutsche Telekom's data center now heats a 514-apartment building in Schöneberg, live since December 2025, covering a large share of its heating demand. [In contact with engineers to verify: is the 7,030 MWh/a figure total building consumption or heat-pump output? Decides ~65% vs ~100%.] It is a genuine success — and it took years of negotiation for a 140-metre pipe to a single building. Frankfurt/Langen: a data center connecting to 1,300 apartments, under construction now, due 2027.

That is the tell. Every one of these is a triumph of individual effort over a missing standard. Norderstedt worked because ownership removed the negotiation. Berlin worked because someone spent years pushing one connection through. None of them is repeatable as-is. What standardization does is turn Norderstedt's success into a template instead of a one-off — so the hundredth connection looks like the first, instead of starting from zero every time.

Q 05 · The Missing Interface

The heat exists, the grid exists, the demand exists, the proof cases exist. What is still missing?

If everything is in place, the gap should be nameable in one sentence. What is it — and what, precisely, does the Thermal Plug define that closes it?

One standardized specification that defines the interface — once, for everyone. That is the whole of it. The Thermal Plug defines the things that are the same at every connection and should never be re-engineered: the pipe size and flange class, the metering boundary where ownership of the heat changes hands, the data exchange points and the parameters that cross them, the fault and status signalling, the redundancy class. It deliberately does not define what happens on either side of the interface — not the price, not the internal process, not which control protocol a given operator prefers. It standardizes the connection, not the parties.

The analogy is exact, and it is the one in the name of this whole series. Before USB, every peripheral had its own connector, and every device-to-device link was a custom engineering problem. USB did not change the printer or the computer. It changed the interface between them — and the moment it was standard, the hundredth connection cost nothing to design. The device is the commodity. The socket is the infrastructure.

This is what "use what you own" means in practice. A company does not need to rebuild its plant, change its process, or fund a research programme. It already owns the heat — the register proves it is there. It already has the pipe carrying it. What it has lacked is a defined, bankable point at which that existing, owned heat becomes something a neighbour can plug into. You do not build a new asset. You put a standard socket on the one you already have.

You already own the heat. You already own the pipe. What is missing is the standard socket that turns what you own into something a neighbour can plug into.

Q 06 · Germany as a Test Case

Is Germany special here — or is this a pattern that repeats everywhere?

You have built the whole argument on a German dataset. Does that make this a German story, or is Germany just the first place the data became visible?

Germany is special in exactly one way, and it is not the one people assume. It is not special in having the heat. Every industrialized country has tens of thousands of waste heat sources — steel, chemicals, glass, cement, data centers, food processing. The heat is universal. What makes Germany special is that it passed a law forcing the heat to be declared. The § 17 disclosure obligation is rare; almost no other country has a public waste heat register at this scale. Germany did not create more waste heat than anyone else. It made its own visible.

That makes the country a clean test case for two propositions at once. The PfA register is the proof of concept for what the data looks like when you require disclosure: a national map appears, and it is bigger than anyone guessed. The Thermal Plug applied to that register is the proof of concept for what the connections look like when you standardize the interface: a large, immediately-addressable fraction sorts itself out of the noise. Germany is where you can see both, because Germany drew the map first.

And the model exports precisely because the heat is not special. Once one country demonstrates that disclosure plus a standard interface turns invisible industrial heat into a connectable, bankable resource, the argument travels to every other industrialized economy unchanged. The data obligation is policy any country can copy. The interface is a specification any country can adopt. The Thermal Plug is designed in Germany, against German data and German standards — but nothing about the physics or the logic stops at the border.

Sidebar · The Small-Scale Cousin

What about putting servers inside homes to recycle the heat — is that the same idea, smaller?

It is the same instinct at the bottom of the scale, and it is instructive precisely because it mostly failed. Several companies tried to turn computers into domestic heaters — Nerdalize went bankrupt in 2019; Cloud&Heat closed its residential product at the end of 2024, in its own words because "the business model proved unsuccessful." The survivors — Qarnot, Heata, Leafcloud — all fled upward, to building and city scale.

It fails on its own terms, at every layer at once. A server makes too little heat (around half a kilowatt per node against an eight-kilowatt house — roughly fifteen nodes to heat one home), too cool (45–55 °C, below the 60 °C Legionella threshold — pre-heat only), at the wrong time of year (a third to 40 percent of the heat is dumped in summer), in a place you cannot physically secure (compute hardware sitting in a stranger's home — a cold-boot, side-channel and data-protection problem that rules out sensitive workloads). The one case that touches this series is Qarnot now injecting roughly 16 gigawatt-hours a year into Brescia's district heating — but that is a purpose-built datacenter feeding a network, not a box in a living room.

The cousin doesn't need a Thermal Plug, because there is no handover. The Thermal Plug starts exactly where this ends — at the scale and temperature where a heat source is large enough and hot enough to hand off to a network.

"Germany did not create more waste heat than anyone else. It made its own visible. The map is the rare part. The heat is everywhere — and so, once one country proves it, is the model."

Dimitri Wolf