Data centres, water, and the one test

This is article 1 of 8 in the Data Centre series.

I spent forty years putting numbers on things that were hard to measure, almost always for people who had to bet a great deal of money on the answer, and in all that time the single most useful idea I have come across did not arrive from a cost model, or a boardroom, or a fab running at three in the morning. It came out of immunology, and it begins with a fact about your own body that almost nobody stops to consider.

Right now, inside you, there are more organisms that are not genetically you than there are cells that are. They are in your gut, on your skin, in your mouth, trillions of them across thousands of species, carrying their own DNA, descended from lineages that branched away from ours long before there were animals to branch away from. By any definition a customs officer would recognise, every one of them is foreign.

Your body has taken all of them in. More than taken them in, it has come to depend on them, so that if you stripped them out you would not get a cleaner human being, you would get a sick one.

Now take a heart out of your own perfect DNA match, a good heart, young, correctly sized, matched as closely as medicine knows how to match, and put it in your chest. There is a real chance your body will destroy it inside a week, and nothing at all will be wrong with the heart.

Two facts, and the distance between them is where forty years of building and running companies at scale finally taught me something.

Your body does not run a test on origin. It runs a test on fit.

Fit, not origin. Fit.

One question, asked of everything, forever

The immune system cannot tell a gift from an invasion, does not weigh intent, does not read the donor card, and has no interest whatsoever in how much you needed the thing that just arrived. It asks one question of every cell it meets, over and over, for the whole of your life: is this me, or is this not me? Not where did you come from. Not who sent you. Only asks: Can I recognise you? Can I take you in, and will you be compatible?

Everything that clears that question stays, including trillions of organisms with no human DNA at all. Everything that does not clear it gets dealt with, however flawless it may be by every other measure we know how to apply.

What took me the better part of a career to see is the reason for writing this series and the reason I am writing it for operators rather than for biologists.

The living world, Earth's body, if you will, runs that same test on everything we build.

It does so via chemistry. Every material we put into circulation is submitted to a system that will either metabolise it, break it down and fold it into the next living thing, or fail to, in which case it accumulates, and that system holds no opinion about our intentions and observes no quarterly calendar. It is simply the sum of what four billion years of biochemistry can and cannot process.

This series calls that system, that makes up the Earth's body, the BiosVerse™, and the name is boldly doing one specific job. It is there to stop you from hearing "the environment", which is a word that lands on a board slide as scenery, a place we visit on weekends and ought to be somewhat nicer to. The living world is not scenery, and it is not a stakeholder, and it is not a category of risk sitting in a register. It behaves like a body: interconnected, self-regulating, running a recognition test on everything that enters it. We are not beside it. We are inside it, the way your gut flora are inside you, which is to say everything we create is entirely dependent on being recognised.

The rejection is the mercy

Consider the parallels between a failed graft and a failed material; the difference is the point of looking closer at what is going on in the real world, where people and living things live, not the metaverse where only pixels exist.

When a body destroys a mismatched organ, it does so with violence, with fever and inflammation and a full mobilisation, because it has met proteins like these before and knows precisely which alarm to ring. That violence looks like the tragedy and it is not. The violence is the mercy. It is fast, unmistakable, and everyone involved knows within days that the graft has failed; while nobody would call it a good outcome, it is an outcome you can still act on.

The BiosVerse does not have that mercy available for everything, and the reason is almost unbearably simple. A molecule with no precedent in four billion years of biochemistry trips no alarm, because no alarm was ever wired for it. So there is no fast rejection. There is no rejection at all, not at first.

The thing is simply admitted. It sits in soil, rides groundwater, settles into fat and bone and breast milk, and turns up in a blood sample drawn from a place that has never had a factory making or using it within a thousand miles. It goes on doing that for years, for decades, for generations, long before anyone can agree on what to call or how to cost the illness it is causing.

A body under attack tells you immediately that something is wrong. A body that has failed to recognise an invader tells you nothing until the bill arrives, and by then the bill has compounded well past the point where anyone can trace it back to a single Tuesday in a single laboratory. In thirty-five years of cost models I have never found a cleaner statement of where the money actually goes. Yet the body failed.

We are rarely undone by our failures. Failures are loud, they announce themselves, and we route around them. What undoes us is the thing that works, the material that passes every test we know how to run and goes out into the world looking like a gift, only to be read by the one system whose reading is final: this is not me. The forever chemical, so named because it does not leave. The refrigerant that turned out to eat the sky, sixty years after the man who invented it demonstrated its safety by inhaling a lungful in front of a lecture hall. The coating that will not break down, because not breaking down was precisely what we made and paid it to do.

Every one of them a foreign organ, shipped by the billion, by careful people, in good faith.

So there must be a number

If the test is real, and it is as real as the chemistry of a carbon-fluorine bond, then it has a cost. It also has a credit, and the credit is the half that almost nobody ever reaches for.

Every project any of us has signed, whether we know it at the time or not, carries three accounts, not just the two we build human accounting and economics on. The first is what it costs to build or acquire, which we count to the dollar and defend line by line. The second is what it costs to run, which we count monthly and argue about in committee. Those two are the entire grammar of modern management, the first thing on the board slide and the last thing anyone fights over, and they are not going anywhere, nor should they.

The third runs in megalitres and megawatt-hours and tonnes and acres; it has been running the whole time, and we have never once agreed how to write it down. Not because anyone is wicked. Because nobody ever recognised the ledger and built the account line to write it on.

That account observes no fiscal calendar and accepts no write-off. You can close the year, retire the asset, walk the site, sell the permits, rename the subsidiary, and make the whole thing vanish from your balance sheet in a single journal entry, and none of it settles anything, because what actually happened is the only entry the account has ever held.

And it carries one property that no ledger of ours has ever managed.

It never makes an error in your favour.

Which brings us to the most resented neighbour in America

The Third Number test is not being run in a laboratory somewhere. It is being run this quarter, at a scale with no recent precedent, in specific American counties with specific water tables, on a machine most people have never seen the inside of yet touch several hundred times a day.

A data centre arrives in a town with a jobs number and a tax base that will finally pay for the school roof. A few years later that same town is fighting over water rights, grid capacity, and whose electricity bill went up to subsidise somebody else's cloud storage.

In one of those meetings something happens that ought to be impossible, which is that both sides are telling the truth. The company's figures are real; it does recycle most of its cooling water and it does hold a power purchase agreement. The woman at the back who has farmed the same section for thirty years is also right, because she knows what a dry decade does to that water table. She carries a sense she cannot put into the language the company is using, that something is being drawn out of a system which took a very long time to build.

They are measuring the same ten- to twenty-five-year decision with two instruments that cannot read each other, and so the argument degrades into adjectives, greedy on one side and hysterical on the other, both untrue and both entirely predictable, because the real disagreement is arithmetical and nobody in the room has the arithmetic.

And months before that meeting, in a different room with no doubt better chairs, a board had already decided it.

They were choosing between two designs for a plant they would operate for as many as twenty-five years or more. Design A built the ordinary way, a hundred million dollars to build and ten million a year to run. Design B cooling without drawing the river down, putting its own waste heat to work, specifying materials the living world can take back, at a hundred and thirty million to build and the same ten to run. The premium was thirty million dollars, real money, sitting right there on the page where everyone could see it.

That board was operating on a rule I have watched work correctly for four decades: a decision is sound when the numbers on the page are sound. Scrutinise what is in front of you, defend every line, do not approve what you cannot defend. It is the discipline that separates operators from enthusiasts, and I have spent a career teaching it.

It has one failure mode, and only one. It cannot see an omission. A soft assumption gets caught, a missing row does not, because a missing row looks exactly like solid ground. So they approved Design A, and everyone voting did their job correctly on the information in front of them, but that information provided only two-thirds of the decision.

Put the third number on that page, and their decision comes apart. Design A specifies a material that cannot be taken back in and fails on closing its loops, so post what that costs across generational time, and post it in the right order, which means never a dollar first. The physical entries first: the water consumption it will reach, the tonnage nobody can recover on any timescale that matters. Those ride the account for a generation, and only then does a directional dollar figure follow, and that figure dwarfs the thirty million Design A saved at the build.

The Third Number bill lands roughly a dozen years after the vote, all at once, on a successor who was not a part of the decision-making. And the meeting at the back of the school gym is where people who never saw the slide argue about it.

The half that turns this from a warning into a business

Everything to this point is the debit column, and a ledger carrying only debits is an indictment, and nobody has ever built an industry based on indictments.

The account credits exactly as readily as it debits, because physical facts run in both directions. An aquifer that recharges is as real as an aquifer that draws down; ground handed back with topsoil rebuilding posts a credit, just as stripped ground posts a liability.

And what reorganises the politics of every one of those town meetings is not the size of the entry but who carries it. When a thing cannot be taken back in, the liability posts to everyone at once: to the operator who saved thirty million, to the community that hosted it, to the watershed that absorbed it. When a design can be taken back in, the credit posts the same way, to the same three parties.

Which means the compatible design stops being the operator's sacrifice and the community's charity, and becomes, given enough generational time, the one design where all three ledgers clear on the same side of the page and all stakeholders can see it.

The waste heat that used to go up into the sky becomes warmth for the town or food in a greenhouse in February, posting twice, once as megawatt-hours not burned again and once as revenue where a cooling cost used to sit. The water returns to the basin cleaner than it came out, posting twice. The material that was a disposal line comes back as feedstock, and its cost per unit falls with every turn of the loop, posting twice again. Three separate ledgers stop being three. What we have been calling a supply chain, with its single path and its weakest link, starts behaving instead like a possibility network, with more than one route between any two points and value, like mycelium networks, moving along all of them.

And credits compound, which is the fact almost nobody has priced. Every operator in the world understands that a deferred liability grows quietly in the dark; far fewer have noticed that a credit posted early compounds in exactly the same way, at the same steadiness, in the same direction, on your side of the page. Thirty years of that is not a rounding error on a report. It is a different industry.

This is an arithmetic problem wearing a moral costume

I am not a biologist and I am not an environmentalist. I am an operator and a futurist, and what I keep running into in these decision-making discussions is not a shortage of goodwill, which is abundant and useless, but a shortage of arithmetic and critical thinking.

We built an economy that prices two-thirds of any consequential decision with extraordinary precision and the remaining third not at all. Everything in that last third number and ledger then gets handled the only way an unpriced thing can be handled: by argument, by regulation arriving decades late (e.g. compliance), and by people standing up at the back of rooms with no units and no arithmetic with which to understand and argue.

That is not capitalism working badly. That is capitalism working exactly as designed for a two-number system but built on an incomplete input.

Give it the third input, the units and the maths, and the behaviour changes without anyone having to change their character, which is the only kind of change that scales. A firm with a Third-number ledger that closes its loops, sits in the right climate, returns water cleaner than it drew it, and leaves ground better than it found it currently receives nothing for any of it beyond a paragraph in a report no lender will underwrite. Hand that same firm a number a bank can lend against, and every one of those choices stops being a cost imposed on the project and becomes the now-recognised generational liabilities and credits and the reason the compatible project wins the bid.

What comes next

There is a phrase I keep returning to, and it's meant to be read as engineering rather than poetry: nature returning home. Not a repaired planet, which is on nobody's table and which anyone promising it is selling you. Something narrower and far more achievable: a built world the living world can recognise, take in, and keep metabolising long after every one of us has stopped making generational decisions and compatible things.

The instruments for that recognition are further along than most people realise and less settled than any of us working on them would like. The BiosVerse-Compatibility Index and the Generational Cost of Ownership model both exist. What is not yet settled is who holds the standard, what counts as evidence rather than assertion, and who has the standing to fail a project. Those are not questions a laboratory answers. They get answered by the people who make the drawings, sign the cheques, run the towns, and farm the section next door, which is an uncomfortable way to build an institution and yet it is the only way anyone ever has.

So I am not going to close this with a tidy sentence, because the thing is not tidy and the interval in which we get to make it happen is shorter than it looks.

Your own body has been answering one question about every cell inside you, every second, since before you were born. The living world has been answering the same question about everything we build, and settling the account in units that do not negotiate, since long before any of us learned to keep books.

The verdicts were never missing. The ledger was.

So what gets counted on your next generational decision, and when the account finally settles, do your third numbers post as credits, or as liabilities, and whose name is on the door the day they land?

What we are asking for

If you have sat on either side of one of those decisions, you are already holding something this work needs. The BiosVerse-Compatibility Index (BCI) scores whether a design fits. The Generational Cost of Ownership (GCOO) model prices what the fit or the misfit costs across generations. Both exist and both run today. What is not settled is how they are governed: who holds the standard, what counts as evidence, and who has the standing to fail a project. That is the work of Biomimetics26, on November 9 and 10 at the Minnesota Landscape Arboretum, and it is open. Bring what you wish you could have put on the page the night those decisions were being made, and bring the evidence that would have made the room believe it.

Next: a hundred megawatts of computing is also, at the same instant, a hundred megawatts of heating. Almost nobody is heating anything with it, and the reason turns out to be a calendar.

illuminem Voices is a democratic space presenting the opinions of leading Sustainability Thought Leaders, their views do not necessarily represent those of illuminem.

The world needs sustainability knowledge. At illuminem, no interest group or shareholder can influence our work. Thank you for supporting our mission to make high-quality and independent sustainability information free for all. Every contribution helps. Thank you for donating today.