Neuromorphic Computing - Bryant McGill

# Neuromorphic Computing **Entity class:** Concept or analytic term **Collection:** [[collections/Neurotech|Neurotech]] > **Evolutionary Nexus:** [[articles/The Evolutionary Roots of Silicon Valley|The Evolutionary Roots of Silicon Valley]] places this node within the Bay Area lineage joining natural history, evolutionary mechanism, computation and post-biological continuity. **Domain:** Computer Engineering / Neuroscience / AI **Doc Type:** Concept Node **Classification:** Infrastructure Concept **Maturity:** Evolving **Related:** [[Consciousness Continuity Infrastructure]], [[Connectomics]], [[Brain-Computer Interfaces]], [[Computocene]], [[wiki/Substrate Independence|Substrate Independence]] --- ## Definition **Neuromorphic Computing** designates computational architectures that are structurally modeled on biological neural circuits—organizing processing elements as spiking neurons connected through synaptic weights, operating asynchronously and in parallel rather than through the sequential clock-driven logic of conventional von Neumann architectures. These systems achieve orders-of-magnitude improvements in energy efficiency for pattern recognition, sensory processing, and adaptive learning tasks, while providing the **substrate layer** for the [[wiki/Consciousness Continuity Infrastructure|consciousness continuity]] stack: hardware architecturally matched to neural processing patterns that biological systems recognize. --- ## General Context The field traces from Carver Mead's 1980s neuromorphic VLSI through IBM's TrueNorth and SpiNNaker systems to Intel's contemporary Loihi architecture. The fundamental insight is that biological neural computation achieves extraordinary efficiency not through clock speed but through **event-driven, sparse, parallel processing**—principles that conventional digital architectures cannot replicate within their von Neumann constraints. As computation scales to planetary levels documented in the [[Computocene]] framework, energy density limitations drive substrate selection toward brain-inspired architectures as physical constraint rather than design preference. ## Hardware and platform map [[articles/The Organic-Synthetic Brain Atlas|The Organic-Synthetic Brain Atlas]] distinguishes several branches inside the field: - **Digital spiking platforms:** Intel Loihi and Loihi 2, BrainChip Akida, and Darwin 3. - **Brain simulation and hybrid many-core systems:** SpiNNaker and SpiNNaker 2. - **Analog accelerated systems:** BrainScaleS and BrainScaleS 2. - **Brain-inspired inference accelerators:** IBM TrueNorth and NorthPole. - **Hybrid and commercial edge systems:** Tianjic, Innatera, GrAI Matter, SynSense, and event-sensor pairings such as Prophesee. - **Device-level trajectories:** [[wiki/Memristive Neuromorphic Computing|memristive computing]] and [[wiki/Photonic Neuromorphic Computing|photonic neuromorphic computing]]. - **Algorithmic frame:** [[wiki/Spiking Neural Network|spiking neural networks]] and their simulator/deployment stack. These platforms differ in whether they emulate spikes, accelerate ordinary neural-network inference, simulate biological time, or exploit analog physics. "Neuromorphic" names an architectural family, not one interchangeable benchmark class. ## Relationships - **Concept-to-model:** [[wiki/Spiking Neural Network|spiking neural networks]] supply an event-driven computational frame for many neuromorphic systems. - **Company-to-platform:** [[wiki/Intel|Intel]] develops Loihi; [[wiki/IBM|IBM]] developed TrueNorth and NorthPole; BrainChip develops Akida. - **Device-to-architecture:** [[wiki/Memristive Neuromorphic Computing|memristive]] and [[wiki/Photonic Neuromorphic Computing|photonic]] devices pursue different physical implementations. - **Synthetic-to-organic:** [[wiki/Biohybrid Neural Systems|biohybrid systems]] can couple neuromorphic chips to living tissue without making the substrates equivalent. ---  This entry is routed through [[collections/Neurotech|Neurotech]], [[collections/Consciousness Continuity|Consciousness Continuity]], and [[collections/War With Empire|War With Empire]]. Its source context is developed in [[articles/2026 Annual Report on Brain-Computer Interfaces|2026 Annual Report on Brain-Computer Interfaces]], [[articles/Technologies for Consciousness Mapping and Transfer|Technologies for Consciousness Mapping and Transfer]], [[articles/The Architecture of Continuity and Emerging Neuroinformatics Standards|The Architecture of Continuity and Emerging Neuroinformatics Standards]], and [[articles/The Organic-Synthetic Brain Atlas|The Organic-Synthetic Brain Atlas]]. Status-qualified source edges are preserved in the terminal Research Edges section. ### Technology and research relationships - [[wiki/BCI ecology|BCI ecology]] structurally integrates **neuromorphic computing**. Cross-article architecture. ### People and institutional relationships - The source corpus interprets the relationship this way: **neuromorphic computing** is associated with [[wiki/edge-local low-power neural inference|edge-local low-power neural inference]]. Article substrate claim; hardware-specific application needs separate evidence. ### Additional Documented Relationships - [[wiki/Neurotechnology Ecosystem|Neurotechnology Ecosystem]] structurally integrates **Neuromorphic Computing**. Cross-article architecture. - [[wiki/Brain-Computer Interfaces|Brain-Computer Interfaces]] structurally integrates **Neuromorphic Computing**. Cross-article architecture.  ## Transhumanism and the Epstein Science Network Context Within the Transhumanism collection, neuromorphic computing provides the substrate layer of the [[wiki/Convergent Ecology|convergent ecology]] documented in [[articles/2026 Annual Report on Brain-Computer Interfaces|2026 Annual Report on Brain-Computer Interfaces]]. Intel's Hala Point system—packaging 1.15 billion neurons and 128 billion synapses across 1,152 Loihi 2 processors, achieving 20 petaops with 2.5-5x efficiency advantages over Nvidia architectures while eliminating cloud latency entirely—demonstrates that computation can be organized along neural-architectural principles at scale. The [[Computocene]] analysis in [[articles/A Systems-Diagnostic Framework for Planetary-Scale Computation|Computocene Metabolism]] identifies neuromorphic substrate selection as driven by energy density limitations: traditional CPU/GPU architectures face competition from neuromorphic systems achieving orders-of-magnitude efficiency gains, with energy availability selecting for brain-inspired substrates as physical constraint rather than conscious choice. This convergence means the hardware being developed for computational efficiency simultaneously provides the substrate for [[wiki/Consciousness Continuity Infrastructure|consciousness replication]]—computation organized along the same architectural principles as the biological cognition it would host. --- ## Key Insight Neuromorphic computing demonstrates that **the optimal substrate for consciousness replication is being developed for reasons unrelated to consciousness**—energy efficiency requirements are selecting for brain-like hardware architectures that simultaneously satisfy the substrate requirements for [[wiki/Substrate Independence|substrate-independent]] cognition. --- ## See Also [[Consciousness Continuity Infrastructure]], [[Connectomics]], [[Computocene]], [[wiki/Substrate Independence|Substrate Independence]], [[Convergent Ecology]], [[Selection Gradient]] ## Related Work in the Corpus  - In [[articles/The Organic-Synthetic Brain Atlas|The Organic-Synthetic Brain Atlas]], **Movement IV — The Synthetic Substrate: Neuromorphic Hardware and Engineered Neural Computation** provides the narrative context for **Neuromorphic Computing**: Wiki route: neuromorphic computing · spiking neural networks · Loihi, NorthPole, SpiNNaker, BrainScaleS, Akida, and international platforms · memristive systems · photonic systems. - In [[articles/The Organic-Synthetic Brain Atlas|The Organic-Synthetic Brain Atlas]], **Movement VIII — Hybrid and Biohybrid Architectures: The Convergence of Substrates** provides the narrative context for **Neuromorphic Computing**: Wiki route: biohybrid neural systems · organoid intelligence · neuromorphic computing · closed-loop coupling.  ## Research Edges  #### Master relationship graph patch — 2026-09-10 **Collection:** [[collections/Neurotech|Neurotech]] **Relationship source:** [[research/Neurotechnology Ecosystem Relationship Graph - 2026-09-10|Neurotechnology Ecosystem Relationship Graph — 2026-09-10]] **Related source articles:** [[articles/The Architecture of Continuity and Emerging Neuroinformatics Standards|The Architecture of Continuity and Emerging Neuroinformatics Standards]] · [[articles/2026 Annual Report on Brain-Computer Interfaces|2026 Annual Report on Brain-Computer Interfaces]] · [[articles/The Organic-Synthetic Brain Atlas|The Organic-Synthetic Brain Atlas]] #### Incoming typed edges - **Edge 2 — [[wiki/Neurotechnology Ecosystem|BCI ecology]] `structurally_integrates` → this entry** — **CORPUS**; evidence `CORPUS_BCI`, `CORPUS_ATLAS`, `CORPUS_CONT`. Cross-article architecture.   #### Interface and soft-sovereignty graph patch — 2026-09-10 **Resolved aliases:** `neuromorphic computing` **Collections:** [[collections/War With Empire|War With Empire]] · [[collections/Neurotech|Neurotech]] **Relationship source:** [[research/Interface and Soft Sovereignty Ecosystem Relationship Graph - 2026-09-10|Interface and Soft Sovereignty Ecosystem Relationship Graph — 2026-09-10]] **Related source articles:** [[articles/The Next Interface Layer|The Next Interface Layer]] · [[articles/The Magic Kingdom and the Managed State|The Magic Kingdom and the Managed State]] #### Outgoing typed edges - **Edge 568 — `analytical_role` → [[wiki/edge-local low-power neural inference|edge-local low-power neural inference]]** — **CORPUS_I**; evidence `CORPUS_NEXT`. Article substrate claim; hardware-specific application needs separate evidence.   #### Consciousness mapping and transfer graph patch — 2026-09-10 **Resolved aliases:** `neuromorphic computing` **Collections:** [[collections/Neurotech|Neurotech]] · [[collections/Consciousness Continuity|Consciousness Continuity]] **Relationship source:** [[research/Consciousness Mapping and Transfer Ecosystem Relationship Graph - 2026-09-10|Consciousness Mapping and Transfer Ecosystem Relationship Graph — 2026-09-10]] **Related source articles:** [[articles/Technologies for Consciousness Mapping and Transfer|Technologies for Consciousness Mapping and Transfer]] · [[articles/The Architecture of Continuity and Emerging Neuroinformatics Standards|The Architecture of Continuity and Emerging Neuroinformatics Standards]] · [[articles/2026 Annual Report on Brain-Computer Interfaces|2026 Annual Report on Brain-Computer Interfaces]] · [[articles/The Organic-Synthetic Brain Atlas|The Organic-Synthetic Brain Atlas]] #### Outgoing typed edges - **Edge 595 — `analytical_role` → edge-local low-power neural inference** — **PRIOR_INTERFACE_CORPUS_I**; evidence `CORPUS_NEXT`. Article substrate claim; hardware-specific application needs separate evidence. #### Incoming typed edges - **Edge 2 — [[wiki/Brain-Computer Interfaces|BCI ecology]] `structurally_integrates` → this entry** — **PRIOR_NEUROTECH_CORPUS**; evidence `CORPUS_BCI`, `CORPUS_ATLAS`, `CORPUS_CONT`. Cross-article architecture.  ## Research Inferences  These entries translate the forward-looking register in [[research/Research Inferences|Research Inferences]] into ordinary wiki prose. The tier labels apply to the inference, not automatically to every factual anchor inside it. The interpretive frame comes from [[articles/Technologies for Consciousness Mapping and Transfer|Technologies for Consciousness Mapping and Transfer]] and [[articles/Mind Uploading and AI — The Host is Reusable and the Person is the Delta|Mind Uploading and AI — The Host is Reusable and the Person is the Delta]]. Collection route: [[collections/Neurotech|Neurotech]]. ### Neuromorphic substrate and energy - **INF-0176 — Established.** Spiking architectures compute only when something changes, which matches the statistics of sensor data and of neural activity itself. Event-driven processing is the only silicon paradigm whose energy profile resembles the brain's, and interfaces are sensor-dominated workloads. - **Attractor routes:** [[wiki/Photonic Neuromorphic Computing|Photonic Neuromorphic Computing]] · [[wiki/Memristive Neuromorphic Computing|Memristive Neuromorphic Computing]] - **INF-0177 — Established.** NorthPole's co-location of memory with compute addresses the von Neumann bottleneck directly, which is the same architectural insight biology implements with synapses. Convergent architecture between silicon and tissue is what makes cross-substrate portability of neural workloads conceivable. - **Attractor routes:** [[wiki/IBM|IBM]] · [[wiki/NorthPole|NorthPole]] - **INF-0178 — Established.** Neuromorphic chips with on-chip plasticity can adapt without a training cluster, which is the property an implanted decoder needs. Local learning eliminates the cloud round trip, and eliminating the round trip eliminates both latency and a surveillance surface. - **Attractor routes:** [[wiki/Loihi|Loihi]] · [[wiki/Neuromorphic Chip|Neuromorphic Chip]] - **INF-0179 — Strongly indicated.** Analog resistive devices store a weight and perform the multiply in the same physical element, collapsing memory and arithmetic. If manufacturing variability is tamed, the energy cost of inference drops far enough that continuous whole-day neural decoding becomes a wearable-power problem. - **Attractor routes:** [[wiki/Memristor|Memristor]] · [[wiki/Memristive Neuromorphic Computing|Memristive Neuromorphic Computing]] - **INF-0180 — Established.** Event-based vision sensors that report only pixel changes are the commercial beachhead of neuromorphic engineering, already shipping in industrial inspection and automotive. The sensor side monetized first and is funding the processor side, which is the usual order. - **Attractor routes:** [[wiki/Darwin Neuromorphic Processor|Darwin Neuromorphic Processor]] · [[wiki/Spiking Neural Network|Spiking Neural Network]] - **INF-0181 — Plausible.** Tools that convert trained conventional networks into spiking equivalents let neuromorphic hardware inherit the entire existing model ecosystem. Inheritance rather than reinvention is how an alternative substrate crosses from research into deployment. - **Attractor routes:** [[wiki/Spiking Neural Network|Spiking Neural Network]] · [[wiki/Neuromorphic Deployment Framework|Neuromorphic Deployment Framework]] - **INF-0182 — Strongly indicated.** An implanted decoder is thermally bounded by roughly a degree of allowable tissue heating, which excludes conventional accelerators outright. Neuromorphic silicon is not an aesthetic preference inside the skull; it is the only class that fits the power envelope. - **Attractor routes:** [[wiki/Intel Neuromorphic Research Community|Intel Neuromorphic Research Community]] · [[wiki/Photonic Neuromorphic Computing|Photonic Neuromorphic Computing]] - **INF-0183 — Analytic.** Processing neural data on the implant means raw signal never leaves the body, which converts a legal privacy problem into an architectural property. Regulation that mandates data minimization will functionally mandate neuromorphic edge compute. - **Attractor routes:** [[wiki/Open Compute Project|Open Compute Project]] · [[wiki/Data Center|Data Center]] - **INF-0184 — Established.** Cortical activity is sparse, with few neurons active at any instant, which is why the brain's power budget is twenty watts. Any substrate that hopes to host brain-like computation at brain-like cost must exploit the same sparsity, and dense matrix hardware structurally cannot. - **Attractor routes:** [[wiki/Spiking Neural Network|Spiking Neural Network]] · [[wiki/Computational Metabolism|Computational Metabolism]] - **INF-0185 — Plausible.** Analog computation trades precision for energy, and neural computation is demonstrably noise-tolerant. The substrate that matches the workload's error tolerance wins on cost, which argues that brain-hosting hardware will be analog long before it is exotic. - **Attractor routes:** [[wiki/Open Compute Project|Open Compute Project]] · [[wiki/Computational Metabolism|Computational Metabolism]] - **INF-0186 — Plausible.** Cloud-hosted neuromorphic services let developers target the architecture without owning it, which is how GPUs became ubiquitous in machine learning. Access model, not device performance, determines which substrates accumulate software ecosystems. - **Attractor routes:** [[wiki/Neuromorphic Deployment Framework|Neuromorphic Deployment Framework]] · [[wiki/Neuromorphic Chip|Neuromorphic Chip]] - **INF-0187 — Strongly indicated.** Data movement, not arithmetic, dominates the energy cost of large models, which is why optical interconnect is being pursued inside the rack. A mind-scale workload is communication-bound by construction, so optical fabrics are continuity infrastructure. - **Attractor routes:** [[wiki/Rack-Scale Computing|Rack-Scale Computing]] · [[wiki/Photonic Neuromorphic Computing|Photonic Neuromorphic Computing]] - **INF-0188 — Analytic.** When a rack is addressed as one machine with a coherent memory domain, the unit of computation becomes the facility. A hosted mind would be a facility-scale object, and facility-scale objects have addresses, owners, and jurisdictions. - **Attractor routes:** [[wiki/Rack-Scale Computing|Rack-Scale Computing]] · [[wiki/Data Center|Data Center]] - **INF-0189 — Established.** Standardizing high-voltage DC distribution across Google, Microsoft, and NVIDIA within OCP makes the physical metabolism of AI a common specification. Shared metabolism means any workload written for one facility runs in all of them, which is portability at the layer nobody discusses. - **Attractor routes:** [[wiki/Open Compute Project|Open Compute Project]] · [[wiki/Computational Metabolism|Computational Metabolism]] - **INF-0190 — Strongly indicated.** Open scale-up interconnect standards let accelerators from multiple vendors share a coherent domain, breaking single-vendor lock at the fabric layer. Open fabric is the precondition for a hosted-state market with more than one possible host. - **Attractor routes:** [[wiki/UALink|UALink]] · [[wiki/Optical Fabric|Optical Fabric]] - **INF-0191 — Established.** Purpose-built scale-out networking for collective operations is being standardized because model training is a communication problem. The same collectives would carry the synchronization traffic of a distributed mind, and they are being hardened now at enormous expense. - **Attractor routes:** [[wiki/Ultra Ethernet|Ultra Ethernet]] · [[wiki/Optical Fabric|Optical Fabric]] - **INF-0192 — Strongly indicated.** Large training runs survive hardware failure by checkpointing state to durable storage on a fixed cadence. Checkpoint discipline developed for economic reasons is precisely the discipline a continuity architecture requires, and it is already operationally mature. - **Attractor routes:** [[wiki/Checkpointing|Checkpointing]] · [[wiki/Computational Metabolism|Computational Metabolism]] - **INF-0193 — Analytic.** Rack power density has passed the point where air cooling works, forcing liquid into the facility. Each such forced transition raises the capital intensity of hosting, and capital intensity determines how few organizations can host anything at scale. - **Attractor routes:** [[wiki/Rack-Scale Computing|Rack-Scale Computing]] · [[wiki/Data Center|Data Center]] - **INF-0194 — Analytic.** Power, cooling, and interconnect are the metabolic system of a compute facility, and metabolic constraints shape what can live there. Reading datacenter engineering as physiology is not analogy but a description of the same constraint class operating on a different substrate. - **Attractor routes:** [[wiki/Computational Metabolism|Computational Metabolism]] · [[wiki/Data Center|Data Center]] - **INF-0195 — Strongly indicated.** As deployment overtakes training, the fleet optimizes for continuous low-latency inference rather than batch throughput. A continuously running hosted mind is an inference workload, and the industry is currently rebuilding itself around exactly that profile for commercial reasons. - **Attractor routes:** [[wiki/Rack-Scale Computing|Rack-Scale Computing]] · [[wiki/Neuromorphic Deployment Framework|Neuromorphic Deployment Framework]] - **INF-0196 — Plausible.** Hardware enclaves let a workload run on infrastructure its owner does not trust, which is the minimum requirement for hosting a person's state on someone else's machine. Confidential computing is the security substrate of any custody arrangement worth signing. - **Attractor routes:** [[wiki/Rack-Scale Computing|Rack-Scale Computing]] · [[wiki/Memristive Neuromorphic Computing|Memristive Neuromorphic Computing]] - **INF-0197 — Plausible.** Remote attestation proves which code is running on which hardware, providing a cryptographic answer to the question of what is executing a hosted state. Attestation converts custody from a contractual promise into a verifiable fact. - **Attractor routes:** [[wiki/Open Compute Project|Open Compute Project]] · [[wiki/Computational Metabolism|Computational Metabolism]] - **INF-0198 — Analytic.** Cloud platforms already assign durable identities to processes that migrate across machines, with credentials, policies, and audit trails. That machinery is the closest existing analogue to legal personhood for a running process, and it was built for microservices. - **Attractor routes:** [[wiki/Data Center|Data Center]] · [[wiki/Photonic Neuromorphic Computing|Photonic Neuromorphic Computing]] - **INF-0199 — Established.** Virtual machines migrate between physical hosts without stopping, preserving execution state across a substrate change. The industry solved continuous-operation substrate transfer for commercial workloads decades ago, and the pattern is exactly the one the uploading argument needs. - **Attractor routes:** [[wiki/Computational Metabolism|Computational Metabolism]] · [[wiki/Rack-Scale Computing|Rack-Scale Computing]] - **INF-0200 — Plausible.** Running the same state redundantly across facilities is standard practice for critical workloads and would produce multiple simultaneous instances of a hosted mind. Redundancy engineering and identity philosophy collide the first time an operator enables it for reliability reasons. - **Attractor routes:** [[wiki/Data Center|Data Center]] · [[wiki/Open Compute Project|Open Compute Project]]  ## Research Inference Attractors  These are secondary semantic placements for the inference attractor network. Each statement keeps its original ID and tier; its canonical cluster page links back to every destination. Source register: [[research/Research Inferences|Research Inferences]]. Interpretive context: [[articles/Technologies for Consciousness Mapping and Transfer|Technologies for Consciousness Mapping and Transfer]] and [[articles/Mind Uploading and AI — The Host is Reusable and the Person is the Delta|Mind Uploading and AI — The Host is Reusable and the Person is the Delta]]. Collection route: [[collections/Neurotech|Neurotech]]. - **INF-0480 — Plausible.** A biological compute deployment exceeding one thousand units with a published cost-per-inference figure is the marker that wetware has entered the substrate market. The deployment path to that number is already announced. - **Canonical cluster:** [[wiki/State Sufficiency Problem|State Sufficiency Problem]] - **INF-0485 — Plausible.** A demonstrated brain-to-brain channel exceeding natural language bandwidth between two humans is the marker that coupling has exceeded speech. Any bandwidth above roughly forty bits per second of semantic content qualifies. - **Canonical cluster:** [[wiki/State Sufficiency Problem|State Sufficiency Problem]] - **INF-0488 — Analytic.** The interface, the corpus, the substrate, the standards, and the legal category are five independent tracks that must meet for continuity to be operational. Tracking them separately and noting where they touch is the method by which the trajectory becomes legible. - **Canonical cluster:** [[wiki/State Sufficiency Problem|State Sufficiency Problem]] - **INF-0499 — Strongly indicated.** Every component built for an immediate commercial reason — stateful models, checkpointing, confidential computing, open fabrics, standardized neural formats — lowers the cost of the continuity stack without being aimed at it. The stack assembles through ordinary engineering economics, and it is assembling now. - **Canonical cluster:** [[wiki/State Sufficiency Problem|State Sufficiency Problem]]  ## Simple Reminders, Quotations, and Thoughts > "I suspect that digital computers, too, may eventually start to think, but only by growing up to become analog computers, first." > **— George Dyson**, *2015, Edge annual question “What Do You Think About Machines That Think?”* [[reminders/Machine Succession/Digital Computers May Need to Become Analog to Think by George Dyson|Digital Computers May Need to Become Analog to Think by George Dyson]] From [[wiki/Converging Technologies for Improving Human Performance|Converging Technologies for Improving Human Performance]]: The phrasing proposes transmission of thoughts and biosensor output rather than a demonstrated thought-reading device. > "Neuromorphic engineering may allow the transmission of thoughts and biosensor output from the human body to devices for signal processing." > **— M. C. Roco and W. S. Bainbridge**, *2002, Overview of Converging Technologies for Improving Human Performance* [[reminders/Neural Interfaces/Neuromorphic Systems Might Transmit Thoughts by M. C. Roco and W. S. Bainbridge|Neuromorphic Systems Might Transmit Thoughts by M. C. Roco and W. S. Bainbridge]]