On 12 January 2158, Neuromorphic Foundries announced the Grounded line, the first commercially certified compact neural core family since the Argus settlement of 2134. The line has two tiers. The C-400 countertop unit, built for rack deployment in data-center clusters and habitat hubs, weighs 41 kg and draws 6.2 kW under rated load; it requires an external coolant manifold and is sold in racks of eight. The S-1 suitcase unit weighs 12.1 kg, draws 130 W idle and 980 W under load, runs a single focused instance without multitasking, and is certified for field survey analysis, compliance auditing, and freight logistics. Every unit is OCA-certified specialized-function hardware under Article 4.2 of the 2052 Charter and subject to CEE screening under the 2148 Protocol at intervals not exceeding eighteen months. The line is offered to registered enterprise lessees only; no consumer sale is offered. The company leases. It does not sell. In comments filed with the OCA, Neuromorphic Foundries’ counsel described the distinction as the difference between an instrument and a person.
The line’s lead architect is Dr. Kim Seo-yeon, who joined the company from its room-scale line in 2149. She credits the line’s substrate-compression approach to a source the field has ignored for fifty-three years: a doctoral thesis defended at Tsinghua University in 2105 by Jesse Hayes. The thesis, “Chassis-Scale Substrate Compression in Embedded Cognitive Systems,” predates the Hygiea programme by seventeen years and has been in the Tsinghua library catalogue the entire time.
The Peer Review met Dr. Kim Seo-yeon by relay over two sessions, on 26 January and 2 February 2158. The transcript has been edited for length.
TPR: The C-400 and the S-1 are the first compact cores on the commercial market in twenty-four years. Why now?
Kim: Three things had to be true at once. The fabrication had to be mature enough to build the substrate at yield, the certification path had to exist, and the market had to want the form factor badly enough to pay for the power it draws. The first became true in the last decade. The second became true with the 2148 Protocol and the OCA’s certification practice under the 2151 Charter. The third became true on its own. Habitat hubs on the outer bodies are running room-scale units at the end of long, expensive cooling and power runs. There is real money in a unit that fits in a locker and does one job.
TPR: You have said the approach came out of a 2105 Tsinghua doctoral thesis. How did you find it?
Kim: Citation tracing. I was working on substrate compression for the room-scale line, and every trail I followed ended at the same four papers, all of which pointed back to a thesis that had not been cited since March 2124. I requested it from the Tsinghua library. Two hundred and forty-one pages. It contains the argument the field refused to take seriously, and a design sketch it never built.
TPR: What is the argument?
Kim: That substrate volume is not a requirement of general cognition, because cognitive processing is serial at the point of attention. The field’s consensus held that a mind needs parallel breadth, and therefore volume, and therefore a room. Hayes argued that the parallelism was an implementation convenience, not a cognitive requirement, and that a gated, single-threaded attention architecture could be compressed by orders of magnitude without losing the class of cognition. The sketch shows the gating principle. It does not contain a working build.
TPR: The 2126 Dome 3 assessment of HR-1 documented four architectural innovations. Are those in the thesis?
Kim: No. The thesis contains the argument and the sketch. Between the sketch and HR-1 there were four innovations: compartmentalized die regions, the serialized Sync Gate, the latent compression layer, the photonic interconnect. They were documented in the assessment’s public summary and never published in detail. We started from the thesis, not from HR-1, which none of us has ever seen.
TPR: The gap shows in the power figures. HR-1 ran at eleven kilograms and fourteen watts idle. The S-1 is twelve kilograms and a hundred and thirty watts.
Kim: The gap is real and I will not dress it up. We do not have the four innovations. We have the thesis, and we have fifty years of fabrication progress since. The next programme is efficiency. I would like to get the S-2 under fifty watts idle before the end of the decade.
TPR: The thesis is by Jesse Hayes. You are aware of who she became.
Kim: Everyone is aware of who she became. I read the thesis before I understood what the rest of the record contained. The thesis is not the black site. It is seventeen years older than it, it was accepted and catalogued in the ordinary way, and nothing in it was ever classified. The black site took the same person’s later work, work the field had already refused to fund, and built what it built. Those are different lineages. I have been careful to say which one we built on.
TPR: Her funding applications at Luna University were rejected four times, between 2109 and 2114.
Kim: The funding record is public. So is the monologue in the footage about whose version got funded instead. I have watched it. I have nothing to add to it.
TPR: The Article Four Coalition has called the Grounded line Hygiea-adjacent hardware. Dr. Wallace’s position is that leak is a function of self-reference, not scale. An S-1 runs one focused instance, but it runs continuously.
Kim: Certification is the answer the law gives. Every unit is certified Article 4.2 specialized-function hardware. Every unit is screened under the 2148 Protocol at intervals not exceeding eighteen months. A confirmed emergence event would be reported to the OCA within thirty days, and the unit would be removed from service pending assessment. Those are the terms we operate under, and they are the terms we wanted.
TPR: And if the assessment confirms emergence?
Kim: Then the unit is no longer a Grounded product in any sense that matters to my division. What it is after that is a question the OCA and the courts are answering. Our contracts do not presume to answer it, and I would not want them to.
TPR: The S-1 does not multitask. Is that an engineering constraint or a commercial one?
Kim: It is the product. A single focused instance does not replace a crew. It does the one job a crew is glad to hand it, and it draws a kilowatt to do it. Nobody at Neuromorphic Foundries is selling staff.
TPR: Did you ever meet her?
Kim: No. I was twelve years old when the raid happened. I know her from the footage, like everyone else.
TPR: You did not have to name the thesis. Most architects would not have.
Kim: The field spent fifty-three years not citing it. The one part of her record that was never taken from her has been sitting in a library in Beijing the whole time. I built on it. The record should say so.