London Global Tech Conference · 2026
The world's biggest energy source is the one we throw away. We make it measurable, verifiable, and tradeable — without moving a single pipe between buildings.
The problem
The air conditioner
Takes heat out of the building and throws it into the sky.
That heat was made from electricity. Already paid for.
The boiler, same building
Burns gas to make heat for hot water.
The single largest line on the energy bill for hotels, gyms, hospitals.
One is throwing away exactly what the other is paying to create. Nobody connects them, because nobody has ever been paid to.
Why this isn't already solved
Pipe the heat to somebody who needs it. It works — in Copenhagen. It will never reach the long tail.
District heating has built a few hundred of these. There are tens of millions of outdoor units.verify
The inversion
Heat can't travel.
Data can.
So we stopped moving the heat.
Each building recycles its own heat, on the spot.
The only thing that leaves is a signed number.
That inversion turns an infrastructure project into a network.
What we install
Saving 1 — electricity
Condenser runs cooler → compressor draws less. 10–25% off cooling load.verify
Saving 2 — gas
Recovered heat makes 40–60 °C water → boiler fires less. This is where the money is.
The hard part · our moat
So the edge refuses readings that violate physics — before they can ever become money.
| Attack or fault | What rejects it |
|---|---|
| Temperature sensors swapped mints credit from an idle loop |
Supply must be hotter than return. Otherwise you took heat, you didn't give it. |
| Flow rate inflated | Above the pipe's rated flow → rejected |
| Flow and temperature each nudged every value individually plausible |
Combined power exceeds what the source can produce. Energy conservation, enforced at the edge. |
| A reading replayed | Every reading carries a nonce. Cached at the edge, unique-indexed in the database. |
Every rejection is written to an audit table. The plumbing is a commodity. This is the product.
Architecture
Unit economics · one mid-size site
| Per site, per year | KRW | ≈ USD |
|---|---|---|
| Gas saved (recovered heat replaces boiler) | 4,166,000 | 3,000 |
| Electricity saved (cooler condenser) | 405,000 | 290 |
| Verified savings | 4,571,000 | 3,290 |
| → building owner keeps (40%) | 1,828,000 | 1,320 |
| → ThermaChain (60%) | 2,743,000 | 1,970 |
| Owner's up-front cost | 0 | 0 |
Equipment ≈ ₩8M per site. Payback 2.9 years.modelled Published tariffs and standard emission factors — not pilot data. Yet.
Why a blockchain
A database would settle payments fine. If that were our answer, you should be sceptical.
The real constraint
1,000 sites ≈ $6M of equipment.
Unit economics are good — but banks will not underwrite thousands of scattered small assets they cannot audit.
What tokenisation actually buys
Investors fund a site; repaid from that site's verified savings.
They audit the meter, not our books — every reading hardware-signed, committed on-chain.
That is what DePIN actually means: decentralised capital formation for physical infrastructure.
And it only works if the measurement can't be faked.
Where we are
✓ Done
Edge oracle — HMAC auth, physics verification, Merkle-committed batch signing
Settlement contracts — savings waterfall, site funding, soulbound carbon proof
97 passing tests. A verified reading goes meter → edge → chain in one hop, today
✗ Not yet
No pilot. The economics on slide 8 are modelled, not measured
Only the gas saving is metered directly. The electricity saving needs a weather-corrected baseline — 3 months of pre-install data
The ask: one site with steady hot water demand — hotel, gym, hospital,
or a data centre with a use for 60 °C water.
Three months of baseline. Then slide 8 stops being a model.
thermachain.pages.dev · press O for the appendix
Appendix · likely question 1
What software does catch
Swapped sensors, inflated flow, replayed readings, impossible clocks
Energy conservation — values that individually look fine but together exceed what the heat source can produce
What software cannot catch
A gateway that lies by a small amount, inside the rated envelope.
No amount of cryptography fixes this. It is not a software problem.
The answer is physical: a legally sealed meter, a key inside a secure element that cannot be extracted, and a second meter on the delivery side so one tampered device fails the cross-check.
Appendix · likely question 2
Metered — no baseline needed
Recovered heat. Every kWh into the tank is a kWh the boiler did not produce.
This is the majority of the money, and it is directly measured.
Estimated — baseline required
The electricity saving from a cooler condenser. You cannot meter the world where we did not install anything.
Currently excluded from settlement. We under-report rather than over-claim.
90% of ESCO disputes are baseline disputes. So: three months of pre-install measurement, the outdoor-temperature-to-consumption relationship agreed in writing, before a single bolt is turned. IPMVP Option C.
Appendix · likely question 3
Because we did the arithmetic and it does not work.
| Per site, per year | KRW | share |
|---|---|---|
| Verified energy savings | 4,571,000 | 100% |
| Carbon allowance value @ ₩10,000/t | 79,000 | 1.7% |
| Carbon allowance value @ ₩40,000/t | 316,000 | 6.9% |
And small scattered sites cannot reach the compliance market without aggregation and paid
verification that costs more than the credits.
Carbon is a real co-benefit. It is not the revenue model — so it is not in our financial model.
Appendix · numbers
Good sites: sauna, gym, motel, hotel, hospital, care home, laundry, pool — anywhere hot water runs all day.
Bad sites: offices, convenience stores, cafés. They cool, but they barely use hot water,
and without hot water demand the economics are near zero.