Asked in this order, every time
The objections, answered plainly
A reader who builds hardware for a living raises the same seven questions. Here they are, with the answers we would give in a room rather than the ones that sound best.
Questions
Seven, in the order they arrive
01
Is this a replacement for silicon?
No, and the sheets never say it is. Silicon logic is the finest substrate ever built and should stay exactly where it is. The claim is about one narrow job beside it — holding analog weights — where the manufacturing route, not the physics, is the advantage.
02
Protonic conduction in anodic alumina depends on humidity. Isn't that fatal?
It is the first real engineering risk, and we do not hide it. Sealing and packaging are part of the device, not an afterthought. If a sealed layer cannot hold its hydration state, retention (E3) fails and the sheets narrow the claim accordingly.
03
What is actually new here?
Not the material and not the mechanism — both are decades old and cited as such. What is untested is whether a bath-grown layer can be engineered into a repeatable analog-state device read electrically. That is what the three experiments settle.
04
Why does the manufacturing route matter so much?
Because leading-edge logic is capital-rationed and allocated rather than sold. Anodising does not compete for a wafer slot, an advanced-packaging slot or a memory allocation. Sidestepping the queue is valuable precisely because it is not trying to be what is queueing.
05
One per cent of data-centre electricity is small. Why bother?
Because it is the smallest number still worth the arithmetic, and because it is honest. At sector growth rates, one per cent is replaced in weeks — which is the argument for a structurally different device rather than a better version of the same one, not against it.
06
What would change your mind?
A negative species result closes the compute programme permanently. No repeatable switching drops the analog-weight thesis. A competing device doing this job better on a comparable route is also a result — the goal was the one per cent, not the oxide.
07
Is any of this a product today?
No. Every compute statement on this ledger is conditional on experiments that have not been run. Nothing here is a product specification, a performance guarantee, or an offer of securities.