DSC-TR-2026-011FrameworkDraft for Comment
Neural Data Preservation Framework (NDPF), Version 0.9 — Preliminary
Directorate of Neural Continuity
Research program · NCRP
Neural Continuity Research Program
A staged, multi-decade research program investigating whether human cognition could eventually be represented, reconstructed and sustained on computational systems, beginning with acquisition standards, small-organism emulation benchmarks, a research ethics framework and the Neural Data Preservation Framework.
- Lead office
- Directorate of Neural Continuity
- Established
- 9 June 2026
- Partner offices
- Office of Advanced Artificial Intelligence; Continuity Infrastructure Administration; Office of Long-Term Systems Research
Program overview
The Neural Continuity Research Program (NCRP) is the Department’s principal long-term research initiative. Chartered on 9 June 2026 and led by the Directorate of Neural Continuity, it investigates whether human cognition could eventually be represented, reconstructed and sustained on computational systems, and what scientific, engineering, ethical and institutional conditions would have to be met for that to occur.
The Program’s long-term objective is stated in its charter plainly: to determine whether human minds could be preserved in purpose-built computational infrastructure and, if they could, to establish the conditions under which such preservation might responsibly be undertaken. That objective is aspirational. No method for achieving it exists. No whole-brain emulation of any vertebrate has been demonstrated, and the Department does not know whether the objective is achievable in principle, achievable with foreseeable technology, or achievable at all. The Program is designed so that this question is answered by evidence accumulated in sequence rather than assumed at the outset.
The Program follows from the Department’s mission, which includes the preservation of human knowledge and cognition. The Interagency Task Force on Civilizational Continuity observed in its final report (ITFCC-2025-07) that the information held in physical archives can, in principle, be copied and protected, while the knowledge held in human minds currently cannot, and that no federal research effort was examining whether that asymmetry is permanent. NCRP is the Department’s response to that observation. Its timescales extend well beyond the 2035 horizon of the Civilizational Systems Outlook, and the Department does not regard the Program as a contingency measure for any projected outcome.
Work is organized in four Program Phases. Phase I (2026–2028) establishes foundations: acquisition standards, small-organism emulation benchmarks, a research ethics framework and the Neural Data Preservation Framework (NDPF). Phase II (2028–2031) addresses mammalian-scale reconstruction, including a full mouse connectome with functional validation and primate cortical volumes. Phase III (2031–2034) examines human-scale feasibility. Phase IV (2034 onward) would study the conditions for continuity deployment and will begin only if Phase III results and independent review support it. Each transition is a formal gate at which the Program may be continued, re-scoped or concluded.
The Program draws on all of the Department’s research divisions. The Office of Advanced Artificial Intelligence supports segmentation, model evaluation and simulation methods; the Continuity Infrastructure Administration provides compute at Facility Alpha, where the NCRP emulation benchmark suite runs on ALPHA-1, and long-duration data custody at Facility Bravo and Facility Charlie; and the Office of Long-Term Systems Research contributes work on identity, institutional design and multi-decade governance. Every NCRP research protocol is reviewed by the Research Ethics Review Board (RERB) before it begins.
Scope of the Program
NCRP is a research program. It does not offer, and is not developing, any service to the public. No current or scheduled phase includes an attempt to emulate a human individual. The Department maintains no register of individuals for preservation and does not accept requests for enrollment, tissue donation or data submission through this site.
Technical documentation is released as it is completed. Current Program documents include Connectomic Data Acquisition: Technical Requirements for Synaptic-Resolution Brain Mapping, the preliminary Neural Data Preservation Framework, Version 0.9 (open for comment until 15 December 2026), Energy and Compute Requirements for Large-Scale Neural Emulation, and the working paper Criteria for Identity Continuity in Emulated Cognitive Systems.
Research objectives
- Establish open, common technical standards for synaptic-resolution acquisition of neural tissue, covering specimen preparation, imaging, registration, segmentation and quality metrics, building on DSC-TR-2026-009.
- Build and maintain the NCRP emulation benchmark suite on ALPHA-1 and demonstrate independent, reproducible EFL-3 emulation of the C. elegans nervous system and of published Drosophila circuit datasets.
- Finalize the Neural Data Preservation Framework as Version 1.0 following public comment on Version 0.9, and apply it to all Program data from acquisition onward.
- Develop validation methods that relate structural reconstructions to measured neural activity and behavior, and define quantitative acceptance criteria for each Emulation Fidelity Level.
- Produce, in Phase II, a complete mouse connectome with functional validation, together with reconstructed primate cortical volumes sufficient to test cross-species scaling assumptions.
- Quantify the acquisition throughput, storage, compute and energy that human-scale reconstruction and emulation would require, and maintain those estimates as technology changes (DSC-TR-2026-008).
- Establish, with the RERB, a research ethics framework covering consent, the moral status of emulations, identity, equity of access and the conditions under which research must pause.
- Advance the scientific and philosophical understanding of identity continuity, treating it as an open question (DSC-WP-2026-006).
- At each phase gate, determine on the evidence whether the Program should continue, be re-scoped or conclude, and publish the basis for that determination.
- Preserve all Program data, methods and decision records so that results remain usable and auditable beyond the service life of current technologies and institutions.
Current research areas
Select a research area to read a summary of current work.
Connectomics
Reconstruction of the complete wiring diagram of a nervous system at the resolution of individual synapses. NCRP connectomics work concentrates on reproducibility rather than on single landmark datasets: common specimen-preparation protocols, shared reference volumes for comparing segmentation pipelines, and error models that state how many synapses in a reconstruction are likely to be missing or spurious. Near-term work uses invertebrate and mouse tissue; Phase II extends this to a whole mouse brain and to selected primate cortical volumes.
Whole-brain emulation
The construction of computational models intended to reproduce the activity of an entire nervous system from its measured structure, rather than to approximate behavior statistically. The Program treats whole-brain emulation as a research hypothesis with a staged test plan. Present work is confined to organisms whose connectomes are fully mapped, and to the question of how much of an animal’s measured behavior an emulation reproduces when it is given only structural data and generic biophysical parameters.
Neural architecture modeling
Models of how cell types, local circuits and long-range projections are organized across brain regions and species. This work provides the priors that reconstruction and emulation depend on: expected connection probabilities, cell-type distributions and laminar structure. It also identifies which architectural features are conserved between mouse, primate and human cortex, and therefore which Phase II results can reasonably inform Phase III planning and which cannot.
Computational consciousness research
Assessment of theories that propose measurable correlates of conscious experience, and of whether any of them can be applied to computational systems. The Program does not assume that an emulation would or would not be conscious. Its interest is practical: the RERB requires a defensible method for deciding when an emulated system might have morally relevant states. This area evaluates candidate indicators, documents their limitations, and supports the provisional assessment protocols used in ethics review.
Neural data acquisition
Development of the instruments and pipelines that convert tissue into data: fixation and staining, serial sectioning or block-face imaging, multi-beam electron microscopy, expansion and light-sheet methods, and molecular labeling. The governing constraint is throughput. At current rates, imaging a single cubic millimeter of tissue at synaptic resolution occupies an instrument for months. The Program funds methods that reduce time, cost and error per unit volume, and evaluates them against the requirements set out in DSC-TR-2026-009.
Memory representation
Research into how long-term memories are physically encoded — in synaptic strengths, structural spine changes, molecular states, or combinations of these — and which of those encodings survive tissue preservation and can be measured. The question bears directly on fidelity: if memory depends substantially on molecular state that current methods do not capture, structural reconstruction alone would be insufficient. Studies in this area use model organisms with well-characterized learned behaviors so that reconstructed memory traces can be checked against behavior.
Cognitive reconstruction
Methods for inferring functional parameters that cannot be measured directly from a preserved specimen, such as ion-channel densities, synaptic efficacies and neuromodulator sensitivities, by combining structural data with cell-type libraries, molecular markers and recorded activity from comparable tissue. Reconstruction is evaluated on held-out data: a reconstructed circuit must predict responses it was not fitted to. This area is where the gap between a wiring diagram and a working model is measured most directly.
Digital identity continuity
Joint work with OLSR on whether, and under what criteria, an emulated cognitive system could be said to continue the identity of the person whose brain it was derived from. The area addresses copying, branching, gradual versus discontinuous transfer, and the relationship between psychological continuity and physical substrate. It does not assume an answer. The working criteria published in DSC-WP-2026-006 are a framework for structured disagreement, not a resolution.
Advanced simulation systems
Software and hardware for running large spiking and biophysical network models efficiently: event-driven simulators, mixed-precision numerical methods, partitioning of networks across thousands of accelerators, and checkpointing of very large model states. Work is conducted on ALPHA-1 at Facility Alpha and informs the design of the human-scale compute demonstrator planned for Phase III. Simulation results are reproducible by design: every benchmark run records software versions, seeds and input data hashes.
Long-term neural information preservation
Custody of neural datasets over decades and potentially centuries. Program data are managed under the Neural Data Preservation Framework, which defines tiers of retention, fixity checking, format migration, provenance records and access control for data derived from human donors. Working copies are held at Facility Alpha; preservation copies are held at Facility Bravo and, for designated reference datasets, in long-duration media at Facility Charlie, in coordination with the Long-Duration Information Preservation Program.
Technical challenges
The obstacles between present capability and the Program’s long-term objective are large, and most of them are not yet well quantified. This section summarizes the principal challenges as the Department currently understands them. Throughout, progress is described using the Emulation Fidelity Levels (EFL) defined below, which all Departmental documents use to state what a given model does and does not represent.
| Level | Definition | What the model represents | Current status | NCRP phase target |
|---|---|---|---|---|
| EFL-0 | Statistical or behavioral model | Input–output behavior only; no internal correspondence to neural structure | Routinely achieved; not considered emulation | Reference baseline only |
| EFL-1 | Regional network model | Brain regions as nodes, with connectivity estimated from imaging | Achieved for human and animal brains at coarse resolution | Reference baseline only |
| EFL-2 | Cell-type population model | Populations of cell types per region with characteristic dynamics | Achieved for portions of rodent cortex | Used for Phase II comparison |
| EFL-3 | Neuron-level connectome model | Each neuron and its synaptic connections, with generic biophysical parameters | Demonstrated for C. elegans and for partial Drosophila circuits | Phase I (invertebrate); Phase II (mouse) |
| EFL-4 | Synapse-level model with plasticity | Individual synaptic strengths and the rules by which they change | Not demonstrated for any complete nervous system | Selected mouse circuits, Phase II |
| EFL-5 | Molecular-state-informed model | Adds receptor, channel and neuromodulatory state measured from tissue | Not demonstrated; acquisition methods do not exist at scale | Feasibility question, Phase III |
| EFL-6 | Validated individual-equivalent | A model shown to be functionally equivalent to a specific individual | Hypothetical; no known path; validation criteria undefined | No target set |
Acquisition scale
Synaptic-resolution reconstruction requires imaging tissue at voxel sizes of a few nanometers laterally and tens of nanometers axially. At throughputs representative of current multi-beam electron microscopy, a single cubic millimeter of cortex requires months of instrument time. A mouse brain occupies roughly 500 cubic millimeters; a human brain roughly 1.2 million. The gap between present capability and human-scale acquisition is therefore about six orders of magnitude in volume, before accounting for sectioning losses, re-imaging and quality control.
The Program does not expect a single technique to close that gap. Its Phase I work compares electron-microscopy, expansion-microscopy and hybrid approaches against common reference volumes, so that improvements in throughput can be measured against a fixed standard of accuracy rather than reported on incompatible terms.
Data volume
Raw image data scale with tissue volume. On the order of one to two petabytes are generated per cubic millimeter at synaptic resolution, before compression. A whole mouse brain would produce on the order of an exabyte; a human brain, on the order of a few zettabytes. For comparison, the tape archive at Facility Alpha holds 1.6 exabytes in total.
Derived products — segmentations, synapse tables and connectivity graphs — are several orders of magnitude smaller than raw imagery, but the Program cannot yet assume that raw data may be discarded once processed: segmentation methods continue to improve, and re-processing has repeatedly recovered structure that earlier pipelines missed. The NDPF therefore defines retention tiers that distinguish raw data held for re-processing from derived data held for long-term reference.
Functional validation
A reconstruction can be checked for structural accuracy against manual annotation. Whether a model built from it behaves like the original tissue is a separate and harder question. The Program’s approach is to record neural activity and behavior from an animal before its tissue is preserved and reconstructed, then test whether the resulting model reproduces activity it was not fitted to. This places strict demands on experimental design: the recorded neurons must be matched to their reconstructed counterparts, and the tests must be specified before the model is run.
Validation criteria for EFL-3 in invertebrates are being finalized in Phase I. Criteria for mammalian tissue will be set before Phase II acquisition begins, not afterwards.
The structure-to-function gap
The C. elegans connectome has been known in substantial detail since the 1980s, yet models built from it reproduce the animal’s behavior only partially. The example is instructive. A wiring diagram specifies which neurons connect; it does not by itself specify the sign and strength of each connection, the electrical properties of each cell, or the influence of signals that do not travel through synapses. Some of these properties can be inferred from cell type and morphology. Others may need to be measured directly, and for some there is no current method of measuring them in preserved tissue.
The size of this gap for mammalian brains is unknown, and estimating it is one of the main purposes of Phase II. If structural data together with cell-type information prove sufficient to predict function at EFL-3 or EFL-4, the path to later phases is clearer. If they do not, the Program will report that result as a primary finding.
Plasticity and neuromodulation
Brains change continually. Synaptic strengths shift with experience, new connections form and old ones are removed, and neuromodulators such as dopamine and serotonin alter how entire circuits respond over seconds to days. A model that captures a fixed snapshot of connectivity but none of these processes would not learn, and might not remain stable over time. EFL-4 requires plasticity rules to be represented explicitly; EFL-5 additionally requires the molecular state that governs them.
Plasticity rules are known in outline for some synapse types and poorly or not at all for many others. Neuromodulatory state is largely lost at fixation. The Program’s memory-representation and cognitive-reconstruction research areas address these questions directly, and the Department regards them as among the least tractable in the Program.
Compute and energy
Estimates of the computation needed to emulate a human brain span many orders of magnitude, depending chiefly on the fidelity level assumed. DSC-TR-2026-008 sets out the Department’s working estimates and their assumptions. Under its central assumptions, real-time emulation at EFL-4 would exceed the sustained capability of ALPHA-1 — approximately 1.1 exaFLOPS FP64 peak and 38 exaFLOPS low-precision peak — by a wide margin, and memory capacity and inter-node communication, rather than arithmetic throughput, are likely to be the binding constraints.
Energy is a separate constraint. The human brain operates on about 20 watts. Facility Alpha is designed for 30 megawatts of IT load. Any long-term objective that involves sustaining emulations over extended periods would require an improvement in energy efficiency of several orders of magnitude, through specialized hardware or methods not yet developed, and would also require power and cooling arrangements designed to operate over decades. These questions are pursued jointly with the Continuity Infrastructure Initiative.
Verification of fidelity
Each EFL is defined together with the evidence needed to assign it. For EFL-3 and EFL-4, that evidence consists of structural error rates within stated bounds and predictive performance on held-out activity and behavior. For EFL-5 the Program has proposed, but not adopted, criteria based on predicting molecular perturbation effects.
EFL-6 has no adopted verification procedure, and the Department is not confident that one can be devised. Demonstrating that a model is functionally equivalent to a particular person would require comparisons across an open-ended range of behavior and inner states, some of which may be inaccessible to external measurement. The Program therefore treats EFL-6 as a conceptual upper bound on the scale, not as a planning target, and no Program milestone depends on reaching it.
Program documents assign an EFL only to models that have passed the verification procedure for that level. Claims of fidelity made without that evidence are recorded as unverified.
Ethical considerations
Research directed toward the preservation of human cognition raises questions that are not settled by technical progress and may not be settled at all. The Department’s position is that these questions must be addressed in advance of the work they concern, not after it. The considerations below are reflected in the NCRP research ethics framework now in development, and every Program protocol is reviewed against them by the Research Ethics Review Board (RERB).
Consent
No NCRP protocol uses human tissue in Phase I or Phase II. When human tissue acquisition is proposed in Phase III, it will be limited to tissue donated after death with documented prior consent, and to tissue removed during clinically indicated surgery and donated with consent. The Program will not acquire tissue from any living person for research purposes alone, and no clinical decision may be influenced by a donor’s participation.
Consent for neural data differs from consent for ordinary biological samples. A synaptic-resolution dataset may, in principle, contain information about memory and personality that the donor never disclosed, and it may be retained for longer than any consent process can meaningfully anticipate. The ethics framework will therefore require consent that addresses duration of retention, permitted uses, prohibited uses — including any attempt at individual emulation — and the procedures that apply if the donor’s wishes cannot be ascertained.
Moral status of emulations
If an emulation were sufficiently faithful, it is possible that it would have experiences, interests or the capacity to suffer. The Department does not know at what level of fidelity, if any, this would occur, and no current theory provides a reliable test. The Program adopts a precautionary posture: emulations of vertebrate nervous systems at EFL-3 or above require a provisional moral-status assessment before they are run, and conditions limiting run duration, input regimes and termination procedures may be imposed.
Research involving animals, including the mouse and primate work planned for Phase II, is subject to established animal-welfare oversight in addition to RERB review. The Program records the number of animals used and reports it annually.
Identity
Whether an emulation derived from a person would be that person, a new person with that person’s memories, or something else is unresolved. The question is not academic: it determines what consent can authorize, what obligations would be owed to an emulation, and whether the Program’s long-term objective describes preservation at all. A reconstruction that can be run can also be copied, which raises the further question of how identity would apply to two simultaneous instances.
The Department does not take a position on these questions. DSC-WP-2026-006 sets out working criteria that allow competing positions to be stated precisely and tested where possible. Program communications avoid language that presupposes an answer.
Equity of access
If preservation of human cognition ever became feasible, its cost and capacity would probably be limited for a long period. Who would have access, and on what basis, would then be among the most consequential decisions a society could make. The Department has not adopted, and is not developing, any criteria for selecting individuals. Its position is that access must not be determined by wealth, status or proximity to the institutions conducting the research, and that any allocation principles should be set through public deliberation rather than by a research program.
OLSR is studying the institutional arrangements under which such decisions could be made legitimately and sustained over long periods. That work is published as it is completed.
Governance by the RERB and ESAB
The Research Ethics Review Board is a standing body that reviews every NCRP protocol before work begins and at least annually thereafter. It may approve, require modification of, suspend or terminate a protocol, and its suspension decisions take effect immediately. The External Scientific Advisory Board (ESAB), with 14 members meeting quarterly, reviews the Program’s scientific progress, its estimates of feasibility and the evidence presented at each phase gate. The Chief Ethics and Research Integrity Officer is responsible for the RERB’s independence and resourcing. The Office of the Inspector General may audit any aspect of the Program.
Research moratoria conditions
The NCRP charter defines conditions under which affected research pauses automatically pending review. A pause does not imply that misconduct has occurred; it ensures that decisions about how to proceed are made deliberately and on the record.
- Any emulation meets the threshold for concern under the provisional moral-status assessment, or exhibits behavior that the assessment did not anticipate.
- Any proposal to emulate an identifiable human individual, at any fidelity level, is made in any phase.
- Any breach, loss or unauthorized access affecting human-derived data held under the NDPF.
- Any finding by the RERB that consent documentation for human tissue is deficient.
- A recommendation from ESAB that the scientific basis for a phase no longer supports continuing it.
- A determination by the Chief Ethics and Research Integrity Officer or the Inspector General that Program governance procedures have not been followed.
Research paused under these conditions resumes only with written RERB approval and, where a phase gate is affected, ESAB concurrence. Pauses and their resolution are reported in the Program’s annual report.
Program governance
NCRP is led by the Directorate of Neural Continuity. The Associate Director for Neural Continuity, who holds the title Director, Directorate of Neural Continuity, serves as Program Executive and is accountable to the Secretary through the Chief Scientist. A Program Office within DNC manages schedules, protocols, data custody and reporting. A Program Steering Committee, chaired by the Program Executive, includes senior representatives of each partner division and meets monthly.
| Body or office | Role in the Program |
|---|---|
| Directorate of Neural Continuity (DNC) | Lead division; Program Executive; Program Office; acquisition, reconstruction and emulation research |
| Office of Advanced Artificial Intelligence (OAAI) | Machine-learning segmentation, simulation methods, independent evaluation of emulation benchmarks |
| Continuity Infrastructure Administration (CIA-DSC) | Compute at Facility Alpha; preservation storage at Facilities Bravo and Charlie; compute and energy planning |
| Office of Long-Term Systems Research (OLSR) | Identity continuity, institutional design, long-horizon governance of Program data and decisions |
| Chief Scientist | Chairs phase gate reviews; approves Program research plans |
| Research Ethics Review Board (RERB) | Reviews all protocols; may suspend or terminate work |
| External Scientific Advisory Board (ESAB) | Reviews scientific progress and gate evidence; advises on continuation |
| Office of the Inspector General (OIG) | Independent audit and investigation |
Phase gates
Each transition between phases is a formal gate. The Program Executive presents the evidence; the Chief Scientist chairs the review; the RERB and ESAB provide written findings. Possible outcomes are to proceed as planned, proceed with a revised scope, extend the current phase, or conclude the Program. A summary of each gate decision and the evidence considered is published.
| Gate | Planned timing | Principal criteria |
|---|---|---|
| Phase I → II | 2028 | Reproducible EFL-3 invertebrate benchmarks; adopted acquisition standard; NDPF Version 1.0 in force; ethics framework issued |
| Phase II → III | 2031 | Validated mouse connectome; measured structure-to-function gap; credible acquisition throughput path; updated compute and energy estimates |
| Phase III → IV | 2034 | Human tissue pipeline operating under approved protocols; compute demonstrator results; RERB and ESAB findings on whether further research is warranted |
Data and publication policy
Methods, software and non-human datasets produced by the Program are published openly, with exceptions only where release would compromise donor privacy or research integrity. Human-derived data, when they exist, will be held under controlled access in the most restrictive NDPF tier and will never be released in a form that permits reconstruction of an individual’s neural structure. Program documents are listed in the publications library, and Program news is posted under news.
Developmental milestones
Status as of October 2026.
Phase I · June 2026 Complete
Program charter and RERB protocol review
NCRP chartered on 9 June 2026. The RERB adopted standing procedures for reviewing all Program protocols before initiation, including interim criteria for non-human tissue studies and emulation runs.
Phase I · August 2026 Complete
Acquisition technical requirements published
DSC-TR-2026-009 set out resolution, contrast, registration and error-rate requirements for synaptic-resolution acquisition, and proposed the reference volumes to be used for inter-laboratory comparison.
Phase I · August 2026 – mid-2027 In Progress
Emulation benchmark suite on ALPHA-1
Initial benchmark suite installed on ALPHA-1 following Facility Alpha’s initial operating capability on 18 August 2026. Independent EFL-3 replication of C. elegans reference models is under way; Drosophila circuit benchmarks follow.
Phase I · September 2026 – 2027 In Progress
Neural Data Preservation Framework
Version 0.9 released as a Draft for Comment on 22 September 2026; the comment period closes on 15 December 2026. Version 1.0 will be issued after comments are adjudicated.
Phase I · October 2026 – 2027 In Progress
Acquisition standards working group
Development of the first NCRP acquisition standard, including shared reference volumes and a common error-reporting format for segmentation pipelines.
Phase I · 2027–2028 Planned
Whole-CNS Drosophila emulation benchmark
Attempt EFL-3 emulation of a complete adult Drosophila central nervous system from published connectome data, scored against recorded neural activity and behavior.
Phase I · 2028 Planned
Ethics framework and Phase I gate review
Issue the NCRP research ethics framework, including moral-status assessment procedures. Gate review by the Chief Scientist, RERB and ESAB on whether to proceed to Phase II.
Phase II · 2028–2030 Planned
Whole mouse brain acquisition
Synaptic-resolution acquisition and reconstruction of a complete mouse brain under the NCRP acquisition standard, with full NDPF custody of raw and derived data.
Phase II · 2030–2031 Planned
Mouse functional validation and primate cortical volumes
Functional validation of mouse reconstructions against activity recorded before acquisition; reconstruction of primate cortical volumes to test scaling assumptions. Phase II gate review.
Phase III · 2031–2033 Planned
Human tissue acquisition pipeline at scale
Development of a consented human tissue acquisition pipeline capable of large-volume processing, subject to RERB approval of each protocol. Human data held under the most restrictive NDPF tier.
Phase III · 2033–2034 Planned
Human-scale emulation compute demonstrator
A compute demonstrator at Facility Alpha executing network models of human-scale size using synthetic or aggregate connectivity. The demonstrator does not emulate any individual. Phase III gate review follows.
Phase IV · 2034 onward Planned
Continuity deployment research (contingent)
Research into the technical, legal and institutional conditions any deployment would require. Begins only if Phase III results and RERB and ESAB review support it; it does not itself authorize deployment.
Research publications
DSC-TR-2026-009Technical Report
Connectomic Data Acquisition: Technical Requirements for Synaptic-Resolution Brain Mapping
Directorate of Neural Continuity
DSC-TR-2026-008Technical Report
Energy and Compute Requirements for Large-Scale Neural Emulation
Continuity Infrastructure Administration
DSC-WP-2026-006Working Paper
Criteria for Identity Continuity in Emulated Cognitive Systems: A Working Framework
Directorate of Neural Continuity
DSC-ADM-2026-001Policy and Administrative
2026 Research Priorities Statement
Office of the Secretary
Related facilities
CRF-A
Continuity Research Facility Alpha
Neural computation and high-performance computing
The Department's primary computing site. Facility Alpha hosts the ALPHA-1 system, runs production ensembles of the Integrated Civilizational Systems Model, and carries the Neural Continuity Research Program's emulation benchmark suite. Initial operating capability was reached on 18 August 2026.
Southern NevadaOperational
CRF-B
Continuity Research Facility Bravo
Underground archival and computational research infrastructure
A subsurface research facility built in a repurposed former hard-rock mine. Facility Bravo holds the Department's primary archive at 140–190 m of rock overburden and is fitting out an 8 MW compute hall for redundant computation and archive verification.
Colorado Front Range (subsurface)Partially Operational
CRF-C
Continuity Research Facility Charlie
Long-duration information preservation and redundant systems research
A cold-climate preservation site in interior Alaska, designed so that archival media survive more than a century without power. A pilot vault has operated since July 2026, and construction of the main vault is under way with completion planned for 2028.
Interior AlaskaCommissioning