{"ontology":"https://layer.re/machine/schema.json","epistemic_model":"evidence-backed-temporal-layer-graph","canonical":true,"type":"layer_history","layer":{"id":"layer-networks","slug":"networks","name":"Networks","description":"Communication and connectivity infrastructure.","machine_description":"Communication and connectivity infrastructure.","active":1},"snapshots":[{"id":"state-networks-2022q1","layer_id":"layer-networks","as_of":"2022-03-31 23:59:59.000","summary":"High-bandwidth interconnects were already a necessary part of scaling AI compute across thousands of accelerators, visible in RSC InfiniBand and H100 NVLink system design.","dependencies":["layer-chips","layer-compute"],"constraints":["bandwidth","latency","cluster scale","system reliability"],"emerging_capabilities":["large non-oversubscribed training fabrics","NVLink-scaled DGX systems"],"underdeveloped_components":["portable cross-vendor scale-up standards"],"uncertainties":["vendor architecture claims are stronger than independent comparative evidence"],"confidence":0.91,"previous_snapshot_id":null,"mapping_mode":"retrospective","reconstructed_at":"2026-10-01T17:34:32.000Z","created_at":"2026-10-01T18:05:36.395Z","updated_at":"2026-10-01T18:05:36.395Z"},{"id":"state-networks-2022q2","layer_id":"layer-networks","as_of":"2022-06-30 23:59:59.000","summary":"Networking remained an infrastructure dependency but became more visibly co-designed with model parallelism, accelerators and orchestration software at multi-pod scale.","dependencies":["layer-chips","layer-compute","layer-software"],"constraints":["bandwidth","latency","distributed coordination overhead"],"emerging_capabilities":["multi-pod language-model training","high-bandwidth TPU v4 pods"],"underdeveloped_components":["portable large-scale interconnect abstractions"],"uncertainties":["vendor-specific comparisons do not establish universal architecture superiority"],"confidence":0.92,"previous_snapshot_id":"state-networks-2022q1","mapping_mode":"retrospective","reconstructed_at":"2026-10-01T18:14:00.000Z","created_at":"2026-10-01T19:22:17.763Z","updated_at":"2026-10-01T19:22:17.763Z"},{"id":"state-networks-2024q2","layer_id":"layer-networks","as_of":"2024-06-30 23:59:59.000","summary":"AI Ethernet infrastructure acquired named vendor-reported adoption beyond design announcements; deployment magnitude and performance remained unverified.","dependencies":[],"constraints":["availability scope","reliability and implementation limits"],"emerging_capabilities":["NVIDIA reports Spectrum-X customer adoption"],"underdeveloped_components":["generalization beyond the selected evidence"],"uncertainties":["Historical network predecessor has not been exposed by current read tools; no predecessor ID is guessed.","Vendor adoption report does not independently quantify operational deployment."],"confidence":0.93,"previous_snapshot_id":"state-networks-2022q2","mapping_mode":"retrospective","reconstructed_at":"2026-10-09T19:12:47.000Z","created_at":"2026-10-10T07:29:06.790Z","updated_at":"2026-10-10T07:29:06.790Z"},{"id":"state-networks-2024q4","layer_id":"layer-networks","as_of":"2024-12-31 23:59:59.000","summary":"Trainium2 GA incorporated networked UltraCluster deployment, separating available instance scale-out from preview larger systems.","dependencies":[],"constraints":["availability scope","reliability and implementation limits"],"emerging_capabilities":["Trainium2 EC2 instances become generally available"],"underdeveloped_components":["generalization beyond the selected evidence"],"uncertainties":["Regional GA is not universal supply; preview UltraServers and announced Trainium3 remain separate."],"confidence":0.93,"previous_snapshot_id":"state-networks-2024q2","mapping_mode":"retrospective","reconstructed_at":"2026-10-09T19:12:47.000Z","created_at":"2026-10-10T08:02:14.508Z","updated_at":"2026-10-10T08:02:14.508Z"},{"id":"state-networks-2025q4","layer_id":"layer-networks","as_of":"2025-12-31 23:59:59.000","summary":"Trainium3 instance infrastructure includes a new interchip fabric; the evidence concerns the vendor offering rather than total installed networking capacity.","dependencies":[],"constraints":["Advertised performance ratios are vendor claims and GA is not evidence of unconstrained global capacity."],"emerging_capabilities":["Trainium3 UltraServers become generally available"],"underdeveloped_components":["Independent measurement of deployment scope and outcomes"],"uncertainties":["The referenced historical predecessor belongs to the reviewed chronology; the live history must be checked before ingestion."],"confidence":0.9,"previous_snapshot_id":"state-networks-2024q4","mapping_mode":"retrospective","reconstructed_at":"2026-10-10T06:24:42.000Z","created_at":"2026-10-10T08:37:10.929Z","updated_at":"2026-10-10T08:37:10.929Z"},{"id":"state-networks-2026q1","layer_id":"layer-networks","as_of":"2026-03-31 23:59:59.000","summary":"A future platform integrates networking into its system design; announced future hardware is not treated as operational network capacity.","dependencies":[],"constraints":["Partner-system availability is planned for the second half of 2026; the announcement is not current capacity."],"emerging_capabilities":["NVIDIA announces the Rubin platform"],"underdeveloped_components":["Independent measurement of deployment scope and outcomes"],"uncertainties":["The referenced historical predecessor belongs to the reviewed chronology; the live history must be checked before ingestion."],"confidence":0.9,"previous_snapshot_id":"state-networks-2025q4","mapping_mode":"retrospective","reconstructed_at":"2026-10-10T06:24:42.000Z","created_at":"2026-10-10T08:45:32.711Z","updated_at":"2026-10-10T08:45:32.711Z"},{"id":"state-networks-20260930-initial","layer_id":"layer-networks","as_of":"2026-09-30 12:45:00.000","summary":"Interconnect bandwidth and system topology are becoming central to AI scaling as larger accelerator systems depend on fast communication across many chips.","dependencies":["layer-chips","layer-compute"],"constraints":["bandwidth","latency","power use","system complexity"],"emerging_capabilities":["UnifiedBus-style large-scale interconnect","superpod networking"],"underdeveloped_components":["broad interoperability across accelerator ecosystems"],"uncertainties":["comparative performance outside vendor claims"],"confidence":0.89,"previous_snapshot_id":null,"mapping_mode":"contemporaneous","reconstructed_at":null,"created_at":"2026-09-30T12:54:32.492Z","updated_at":"2026-09-30T12:54:32.492Z"}]}