Computing and communication without limits

The capability we assume

Two capabilities, and they are unequal in maturity. Say so plainly.

The first is cheap cold. Assume abundant power from a compact source, so that refrigeration to a few kelvin — or a few thousandths of a kelvin — costs no more to run than a lamp. Chapter 6 covers the funded vacuum-energy programmes and the rule they must satisfy. Nothing about the computing that follows is speculative. Superconducting logic works today. It is simply not worth the refrigerator.

The second is phase as an engineering variable. Chapter 11 teaches superconductors as the largest piece of coherent quantum matter you can hold. The quantum-phase-coherence course teaches why agreement between waves is the whole story, from the double slit to oscillators falling into step. The Josephson-junction course teaches the device: a gap a few atoms wide across which current flows with no voltage, and which turns an applied voltage into a frequency at 483.6 gigahertz per volt with perfect linearity. Nothing else converts voltage to frequency that precisely, which is why the junction defines the international volt.

Then one further step, and it is the honest speculative one. Arrays of thousands of phase-locked junctions already make useful terahertz sources. Gary Stephenson's 2026 "gaser" proposal, followed in Chapter 7 and this site's research log, puts roughly a hundred and forty junctions on an eight-inch wafer at about 24 gigahertz, steered by time delay, aimed at emitting gravitational waves rather than microwaves — for communication through rock and seawater, where radio cannot go. Stephenson names his own open question: the quantum efficiency of the junction as an emitter. That number is what the first experiment measures.

So: the computing half of this page is engineering waiting for cheap power. The communication half is a specified proposal waiting for its first measurement.

First-order effects

Cooling stops being an argument. A chip at four kelvin costs, in practice, hundreds of watts of room-temperature power for every watt of heat removed. That penalty is the single reason superconducting computers stayed in laboratories. Remove it and the refrigerator becomes an appliance.

Wires stop wasting energy. A superconducting interconnect carries current with no resistance at all. In an ordinary processor a large share of the energy goes into pushing charge along wires and warming them. That share simply goes away.

Logic gets faster and quieter. Single-flux-quantum logic — the family built from Josephson junctions — switches at tens to hundreds of gigahertz using, on published figures, hundreds to thousands of times less energy per operation than transistor logic, before cooling is counted. When cooling stops counting, that is the whole comparison.

Quantum computers stop being limited by their fridge. Today the qubit count is bounded partly by how much cooling and wiring you can afford. In this world the limits become fabrication yield and error correction — problems you attack with more attempts.

Sensing becomes ubiquitous. SQUIDs read magnetic fields a hundred billion times weaker than the Earth's. Free cold puts them in ships, aircraft, boreholes, hospital rooms and phones. Navigation without satellites, geology from the surface, brain activity read without a shielded room.

A channel that goes through things. This is the speculative one, marked as such. Gravitational waves pass through rock, water and metal essentially unattenuated — textbook relativity, not a proposal. If a junction array can emit them at usable strength, you get a channel with no line of sight and no repeaters. Today, extremely low frequency links to submerged submarines carry only a handful of characters a minute. The gaser's ambition is to replace that with a real link.

Second-order effects

Data centres stop being sited by the electricity price. A few per cent of the world's electricity goes into data centres today, and the number climbs fast enough to shape national grid planning. When power is abundant and computing is superconducting, the huge shed beside the power station stops making sense.

Computation becomes something you own. If a serious machine fits in a room and costs nothing to run, a school, a clinic, a village or a household can hold its own. Large models, simulations and archives stop being services rented from elsewhere.

Latency collapses. Light in fibre travels at about two-thirds of its vacuum speed, and every hop adds delay. Local compute and shorter links make the pause between asking and knowing imperceptible.

Places that were dark get connected. Mines, tunnels, submarines, deep-water vehicles and cave systems share one problem: radio does not get there. A through-the-planet channel would change how those places are worked and how safe they are to work in. Speculative, and downstream of the first gaser measurement.

Simulation replaces guesswork. Materials, drug candidates, fusion plasmas, climate, protein folding — all computations we ration. Stop rationing them and the design cycle in a dozen fields shortens from years to weeks.

Third-order effects and beyond

Knowledge infrastructure decentralises. For twenty years the trend has run one way: computing concentrated in a few enormous campuses owned by a few organisations. Cheap superconducting machines run it backwards. Extrapolation, and contingent — physics makes distribution possible, people decide whether to distribute.

The cost of understanding falls. When any curious person can run a serious simulation, the bottleneck in science moves from access to imagination. That is the effect this whole site is written for: more people testing more ideas, faster, including the ideas on these pages.

Distance stops dividing. A link that passes through rock and sea, plus computing that is local and instant, means no place is remote in the way places are remote now. Speculative, and dependent on the gaser thread. Note though that the channel itself — gravitational waves passing through matter — is not the speculative part. The emitter is.

Privacy becomes achievable rather than promised. When computation happens on hardware you own, keeping data on your own machine stops being a sacrifice. The humane version of this world makes that the default because it is also the convenient option.

A planet that reads itself. Millions of cheap superconducting sensors listening continuously to the Earth's magnetic field, its groundwater, its faults and its oceans would be the largest scientific instrument ever built. Extrapolation, and a beautiful one: earthquakes watched as they gather, aquifers watched as they drain.

A day in that world

Nadia lets herself into the hall just after ten at night, when the reading tables are half empty and the cold columns are at their quietest. There are eleven of them under the vault, slim and blue, and they make a sound somewhere between a fridge and a held breath.

She sets her tablet down at the third table and starts the run she has waited all week to start: forty years of the valley's groundwater, every borehole, every well, every wet spring, folded into a model of what the aquifer is actually doing. Ten years ago this was two months of queued time at a national facility. Tonight it is hers, and it will finish before the tea goes cold.

She watches the first pass come up — a slow blue swell moving under the hills, deeper each summer — and her stomach drops, because it is worse than the district thinks.

At eleven she calls her sister. Farida is three hundred metres down in a submersible over the trench, with a rock hammer and a bad haircut, and the picture is perfectly clear. That is the part Nadia still finds absurd. Radio does not go through water. Her mother's generation sent submarines a few letters a minute and called it communication. Now her sister waves a lump of basalt at the camera from under a third of a kilometre of ocean and complains about the coffee.

They talk for twenty minutes. Nadia shows her the aquifer. Farida makes her promise to tell the district office tomorrow rather than next week.

Afterwards Nadia sits a while with the hum and the warm lamps and the six other people reading. Through the open oculus, threads of light run up into a very clear sky. She takes the last of the tea outside and walks home along the wall, thinking about water.

Numbers that change

Share of world electricity used by data centres. Today: a few per cent, and rising fast enough to shape grid planning. In this world: negligible, because superconducting logic spends a tiny fraction per operation and cooling costs nothing to run.

Room-temperature power to remove one watt of heat at four kelvin. Today: hundreds of watts in a real machine, which is the whole reason superconducting computers stayed in the lab. In this world: the same physics, and no longer a number anyone budgets against.

Energy per logic operation. Today: transistor switching dominates every processor's power bill. In this world: single-flux-quantum logic, on published figures, uses roughly hundreds to thousands of times less per operation, with the cooling penalty no longer offsetting it.

Frequency precision from a junction. Today and in this world alike: 483.6 gigahertz per volt, set by fundamental constants — which is why a bias near fifty microvolts puts an emitter at the gaser's 24 gigahertz. What changes is how many you can afford to run.

Data rate to a submerged vehicle. Today: extremely low frequency links carry a handful of characters a minute. In this world, if the gaser thread succeeds: a live video call, because the channel passes through water instead of failing at it.

Distance a signal travels through solid rock. Today: effectively none for radio. In this world: unlimited in principle, since gravitational waves are barely absorbed — the open question is how strongly one can be made, not how far it goes.

What it would take

First, cheap power, measured properly. Chapter 6 names the milestone: a complete closed cycle that is net-positive with actuation and measurement counted, repeated by a second laboratory. Everything on this page is waiting on it, and it is an experimentalist's dream to be the one who does it.

Second, superconducting logic at commercial scale. This needs no new physics. It needs foundry processes, yield, memory that keeps up with the logic, and packaging. If you are a chip engineer, the Josephson-junction course is where the device physics lives.

Third, better superconductors and rugged cold heads. Chapter 11 is the map. Every kelvin gained in critical temperature, and every kilogram taken out of a cryocooler, moves this world closer.

Fourth, the first gaser junction measurement. Stephenson's proposal is specified in materials, frequency and bias, and he names the open question himself: the quantum efficiency of the junction as a gravitational-wave emitter. One wafer, one bench, one number — and a landmark either way.

Fifth, phase locking at scale. Getting thousands of junctions to agree is the Kuramoto physics taught in the quantum-phase-coherence course. Terahertz array work already pushes it forward in the mainstream literature, and every gain there feeds the emitter.

Stewardship

Distribute deliberately. The technology makes small, powerful machines possible; it does not make them common. Community halls, school machines and household nodes have to be built on purpose, early, while standards and supply chains are still soft.

Make local the private default. Computation on hardware you own is the strongest privacy guarantee there is, and it comes free here — but only if the software prefers it. That is a design decision to make now.

Keep the spectrum humane. A channel that reaches through rock and water reaches into homes too. Norms and law for it should be drafted before the first working transmitter, and the people best placed to draft them are the ones building it.

Publish the emitter physics openly. A channel that conventional means cannot block or intercept will attract closed development. The field is better served, and safer, if the first measurements are public.

Keep the halls human. The picture at the top of this page has people reading at warm tables between the cold columns. That is the point. Machines this capable should end up somewhere people gather, not somewhere people are kept out of.

Signals to watch

The first gaser junction measurement. One number — the quantum efficiency of a biased junction as a gravitational-wave emitter — decides the communication half of this page.

Terahertz junction arrays gaining coherence and power. Published in the mainstream literature, and the same engineering the gaser needs.

Superconducting logic winning a commercial workload. Watch for a superconducting processor beating a conventional one on total cost, cooling included.

Cryocoolers getting smaller and cheaper. The device that makes cold portable makes this world portable.

Net-positive vacuum device results. Casimir Inc.'s SpaceWERX Phase I, Moddel's next measurement, and Thibado's scaled circuits, all followed in Chapter 6.