Data centre in the shed reduces energy bills to £40

Terrence Bridges says he "can't fault the heating system", which captures heat from more than 500 mini-computers processing data

ByBen Schofield

BBC East, political correspondent

• Published

  16 November 2025, 03:10 GMT

Updated 7 hours ago

An Essex couple have become the first people in the country to trial a scheme that sees them heat their home using a data centre in their garden shed.

Terrence and Lesley Bridges have seen their energy bills drop dramatically, from £375 a month down to as low as £40, since they swapped their gas boiler for a HeatHub – a small data centre containing more than 500 computers.

https://www.bbc.co.uk/news/articles/c0rpy7envr5o

Interesting approach.

A few thoughts/reservations as the quoted figures sound too good to be true.

500 'computer chips'.

I don't know what chips or how much power they draw, but looking at current processor chips say they draw between 0.05 and 0.1 kw each.

That is a draw of 25 to 50 kw.

So where does this come from ?

The average house supply in the UK is 55kw.

The average gas boiler puts out between 15kw and 25 kw. But that is intermittent.

Presumably these chips are on all the time.

And what happens in the summer when home heating is not required ?

Presumably the installation of a data center would include installing the power and data transfer infrastructure.

The computers are Raspberry pi processors, which draw a few watts each.

https://www.solarpowerportal.co.uk/colocation/could-modular-data-centres-heat-britain-s-homes-a-look-at-thermify-s-heathub

“The HeatHub includes 450 Raspberry Pi processors, which collectively sit in a box the size of a combi boiler. The heat generated by undertaking cloud computing tasks contracted to the company will be used to heat a water tank over the course of about two hours.”

I imagine part of the motivation is saving on real estate costs

I have to wonder if any large-scale data centers could run turbines off of the waste heat and defray a portion of their electricity costs, or put these (or larger) setups where there is already heat distribution infrastructure.

35 minutes ago, swansont said:

Presumably the installation of a data center would include installing the power and data transfer infrastructure.

The computers are Raspberry pi processors, which draw a few watts each.

https://www.solarpowerportal.co.uk/colocation/could-modular-data-centres-heat-britain-s-homes-a-look-at-thermify-s-heathub

“The HeatHub includes 450 Raspberry Pi processors, which collectively sit in a box the size of a combi boiler. The heat generated by undertaking cloud computing tasks contracted to the company will be used to heat a water tank over the course of about two hours.”

I imagine part of the motivation is saving on real estate costs

I have to wonder if any large-scale data centers could run turbines off of the waste heat and defray a portion of their electricity costs, or put these (or larger) setups where there is already heat distribution infrastructure.

Not turbines, surely? The temperature of the waste heat won't be high enough. Space heating is about all it would be good for, I should have thought. But certainly district heating would be feasible.

47 minutes ago, exchemist said:

Not turbines, surely? The temperature of the waste heat won't be high enough. Space heating is about all it would be good for, I should have thought. But certainly district heating would be feasible.

Not with water, but a closed-loop system can use something with a lower boiling point. There are a number of possibilities.

9 minutes ago, swansont said:

Not with water, but a closed-loop system can use something with a lower boiling point. There are a number of possibilities.

OK, fair enough. But still the temperature difference above ambient won't be that much so it can't be very efficient.

1 hour ago, swansont said:

The computers are Raspberry pi processors, which draw a few watts each.

So if we go to 5 watts each, and stick with the 500 that is 2.5 kwatts continuous.

That is certainly not enough to heat my home in winter.

The BBC article also said that the heat is transferred by an oil.

1 hour ago, studiot said:

So if we go to 5 watts each, and stick with the 500 that is 2.5 kwatts continuous.

That is certainly not enough to heat my home in winter.

Which would explain why their bill didn’t drop to zero.

10 minutes ago, swansont said:

Which would explain why their bill didn’t drop to zero.

Not sure I follow.

A UK domestic gas or electricity bill will never be zero due to the standing charge (which is unconscionably high IMHO).

Added to which, £40 per month is good for a summer month for someone who doesn't have gas since that powers cooker, fridge, freezer, washing machine, drier, domestic hot water and uncle tom cobbly.

I remain curious as to the temperature of the heat this system can provide. I don’t know but would guess computers would run at a maximum of about 50C. That’s barely enough to run radiators, let alone generate domestic hot water. There don’t seem to be any temperatures quoted. Can computer processors run at 70-80C?

3 hours ago, exchemist said:

I remain curious as to the temperature of the heat this system can provide. I don’t know but would guess computers would run at a maximum of about 50C. That’s barely enough to run radiators, let alone generate domestic hot water. There don’t seem to be any temperatures quoted. Can computer processors run at 70-80C?

There seem to be a lot of things they haven't told us.

Like heat pump suppliers don't tell you you need bigger radiators because they run at lower water temperatures.

They may also have super insulated the premises in the article.

2 hours ago, studiot said:

There seem to be a lot of things they haven't told us.

Like heat pump suppliers don't tell you you need bigger radiators because they run at lower water temperatures.

They may also have super insulated the premises in the article.

Actually, after a bit of digging I came across this: https://electronics.stackexchange.com/questions/60273/max-operating-temp-of-ics. which suggests the chips themselves may be able to operate up to 125C. So then it's a question of what is the max temp of the coolant that carries the heat away from the chip. I assumed this would be air ventilation and that the liquid coolant would be used to cool the air, but maybe I'm wrong and the chips are in an oil bath or something. (Deep fried?😁)

35 minutes ago, exchemist said:

Actually, after a bit of digging I came across this: https://electronics.stackexchange.com/questions/60273/max-operating-temp-of-ics. which suggests the chips themselves may be able to operate up to 125C. So then it's a question of what is the max temp of the coolant that carries the heat away from the chip. I assumed this would be air ventilation and that the liquid coolant would be used to cool the air, but maybe I'm wrong and the chips are in an oil bath or something. (Deep fried?😁)

It's a long time since I did any work on chip cooling efficiency, but then the raspberyy pi is getting long in the tooth itself.

Yes chips will work at 100^o +, however the hotter they are the shorter their service life, especially at the high end (It's a non linear degradation curve)

I remember that the most efficinet chip cooler were not air cooled but of the 'heatpipe' variety.

A suitably volatile liquid is sealed in a tube or pipe directly connected to the chip as metal to case.

This carries the heat rapidly away to a secondary heatsink thay will be (possibly force) air cooled, by evaporation and condensation.

The cooled liquid then runs back inside the tube to carry on the cycle.

7 hours ago, studiot said:

It's a long time since I did any work on chip cooling efficiency, but then the raspberyy pi is getting long in the tooth itself.

Yes chips will work at 100^o +, however the hotter they are the shorter their service life, especially at the high end (It's a non linear degradation curve)

I remember that the most efficinet chip cooler were not air cooled but of the 'heatpipe' variety.

A suitably volatile liquid is sealed in a tube or pipe directly connected to the chip as metal to case.

This carries the heat rapidly away to a secondary heatsink thay will be (possibly force) air cooled, by evaporation and condensation.

The cooled liquid then runs back inside the tube to carry on the cycle.

Maybe this volatile liquid is what @swansont had in mind with his suggestion about turbines. I would agree that if one can get a working fluid at ~100C, say, extracting heat from a chip at 120C, there is some chance of getting it to do some useful work, though perhaps not much better than the old atmospheric steam engines, pre superheating. One could certainly run a domestic hot water and central heating system if heat at 100C is available.

This isn't something I foresee having widespread application. Whilst efficiency gains - using waste heat that would otherwise go to waste - are a significant element of decarbonising it is a stretch to call it 'clean, green'; that will depend on the sources of energy powering the data centre. Decarbonising the primary energy inputs needs to remain a priority.

Ideally we want ever more energy efficient chips and data storage, ones that don't shed so much heat. Maintaining long running dependence on ongoing data centre inefficiency for the sake of the waste heat for other uses would be a mistake to my mind.

Even district heating systems, that should be built with very long working life, will need to avoid dependence on waste industrial heat from sources that themselves need to transition to energy efficient low emissions alternatives.

12 minutes ago, Ken Fabian said:

This isn't something I foresee having widespread application. Whilst efficiency gains - using waste heat that would otherwise go to waste - are a significant element of decarbonising it is a stretch to call it 'clean, green'; that will depend on the sources of energy powering the data centre. Decarbonising the primary energy inputs needs to remain a priority.

Ideally we want ever more energy efficient chips and data storage, ones that don't shed so much heat. Maintaining long running dependence on ongoing data centre inefficiency for the sake of the waste heat for other uses would be a mistake to my mind.

Even district heating systems, that should be built with very long working life, will need to avoid dependence on waste industrial heat from sources that themselves need to transition to energy efficient low emissions alternatives.

Yes I suppose that's true. Exhibit A might be the Pimlico District Heating Undertaking in London, which was built to use waste heat from Battersea Power Station, just across the river. When that closed, they had to build a separate boiler (coal-fired!) to run the system.

Thinking further about this subject a number of questions occur to me.

1) Does running several kw 24/7 present a fire risk ?

2) Does this constitute commercial use of the land for planning or local tax purposes ?

3) Does this in any way affect the buildings insurance ?

3) If this is now part commercial use, how is capital gains tax affected ?

4) Again if this is now part commercial use does the property have any restrictive covenants against this ?

I say this because UK law is heavily restrictive against the ordinary citizen in many respects.

For instance if you have a stream/river in your garden as the 'riparian owner' your are generally liable to maintain the banks. But you require a licence to place an electricity generator in the water stream, which is genereally not granted.
By contrast in Europe if you want to do this you can actually get a grant towards the cost.

2 hours ago, studiot said:

Thinking further about this subject a number of questions occur to me.

1) Does running several kw 24/7 present a fire risk ?

2) Does this constitute commercial use of the land for planning or local tax purposes ?

3) Does this in any way affect the buildings insurance ?

3) If this is now part commercial use, how is capital gains tax affected ?

4) Again if this is now part commercial use does the property have any restrictive covenants against this ?

I say this because UK law is heavily restrictive against the ordinary citizen in many respects.

For instance if you have a stream/river in your garden as the 'riparian owner' your are generally liable to maintain the banks. But you require a licence to place an electricity generator in the water stream, which is genereally not granted.
By contrast in Europe if you want to do this you can actually get a grant towards the cost.

You mean the EU, presumably. The UK is in Europe, after all 😉.

(I submit herewith my entry for today’s pedantry prize.)

On 11/17/2025 at 2:33 PM, Ken Fabian said:

Ideally we want ever more energy efficient chips and data storage, ones that don't shed so much heat. Maintaining long running dependence on ongoing data centre inefficiency for the sake of the waste heat for other uses would be a mistake to my mind.

That was my first thought on seeing this thread. Second was the expense and energy consumption of moving fairly diffuse heat somewhere useful - the net gain seemed possibly tiny. Paging Maxwell's demon...

On 11/16/2025 at 9:58 PM, exchemist said:

Not turbines, surely? The temperature of the waste heat won't be high enough. Space heating is about all it would be good for, I should have thought. But certainly district heating would be feasible.

Given the sort of temperatures available, T_H~375K, T_C~300K the Carnot efficiency would be a mere 20%. After losses, they'd struggle to achieve a nett positive power output.

Reading between the lines, my guess would be that the data centre is cooled by a heat pump that discharges into a hot oil circuit that in turn supplies the domestic heating.

Maybe wrong, but that's the approach I'd go for.