{"resourceId":"nvidia-ai-infra-summit-vera-rubin-dsx-20260915","versions":[{"version":"external-05001af4557e8c4147b1b14c234207d47281aad3264d5285ac85b00b80923caf","resource":{"id":"nvidia-ai-infra-summit-vera-rubin-dsx-20260915","title":"AI Infra Summit: Vera Rubin, DSX MaxLPS, and partner tokens-per-watt results","organization":"NVIDIA","sector":"AI infrastructure / cloud hyperscale partners","geography":"United States (Santa Clara AI Infra Summit coverage)","publishedAt":"September 15, 2026","publicationDate":"2026-09-15","eventDate":"2026-09-15","sourceName":"NVIDIA Blog","sourceLabel":"Vendor event recap / partner claims","sourceUrl":"https://blogs.nvidia.com/blog/ai-infra-summit-vera-rubin-dsx-energy-efficiencies-tokens-per-watt-ai-factories/","evidenceClass":"vendor-claim","outcomeClass":"emerging","topics":["infrastructure","operating-model","developers-agents"],"finding":"On September 15, 2026 NVIDIA published an AI Infra Summit recap claiming Vera Rubin NVL72 MLPerf Inference v6.1 preview throughput up to 3.7x vs GB300, Lambda DSX MaxLPS validation (+24% token throughput / +23% perf-per-watt within a fixed power budget), Emerald AI flexible-load demos with Silicon Valley Power, and SemiAnalysis AgentX claims of up to 30x throughput per megawatt vs GB300 for agentic coding sessions.","sledRelevance":"Relevant as ecosystem direction on power-constrained AI factories and agentic inference economics. Most claims are hyperscale/cloud-partner measured; SLED campuses rarely match 250 MW-class DSX sites. Treat Lambda/Emerald results as transferability signals for power budgeting—not turnkey campus outcomes.","evidence":"NVIDIA Writers post dated September 15, 2026 reports Ian Buck AI Infra Summit remarks; lists Amazon Annapurna NVHBM and d-Matrix NVLink Fusion collaborations; states Lambda ran 19 nodes in a 16-node power budget (+24% tokens/s, +23% perf/watt) and cites Vera Rubin MLPerf preview up to 3.7x GB300 and AgentX up to 30x tokens/MW vs GB300.","architectureImplications":"Codesign power management (DSX MaxLPS/Flex), scale-up fabric (NVLink), and Dynamo/TensorRT-LLM serving as one stack when evaluating next-gen racks; do not treat GPU SKU alone as the capacity plan.","governanceImplications":"Grid flexible-load programs (Emerald AI / municipal utility) imply operational policies for which jobs may shed—relevant if a public institution joins demand-response, but not evidenced for typical SLED estates.","securityPrivacyImplications":"Post emphasizes BlueField DPUs for infrastructure security at factory scale; no attestation measurements or sovereign-control evaluation.","caveats":"Vendor-authored summit recap aggregating partner demos and MLPerf preview numbers without independent replication in this source. Hyperscale power envelopes and agentic coding AgentX trajectories may not map to campus/agency workloads.","streamIds":["nvidia"],"roles":{"sales":"Frame tokens-per-megawatt and MaxLPS as power-envelope productivity claims from cloud partners, not campus guarantees. Qualify whether the prospect's constraint is utility interconnect, room kW, or software serving stack before citing 3.7x or 30x headlines.","engineering":"Map Vera Rubin / DSX Flex / Dynamo claims to the estate's actual rack TDP, cooling loop, and K8s serving path. Require workload cards that match MLPerf or AgentX shapes before accepting throughput multipliers.","delivery":"Sequence facility power audits, partner reference checks (Lambda/Emerald-class), and pilot SLOs before promising density gains inside an existing megawatt budget."},"roleDetails":{"sales":{"customerProblem":"Leadership wants more AI capacity without new utility interconnects or megawatt upgrades.","stakeholders":["CIO/CTO","facilities/energy manager","AI platform owner","finance"],"discoveryQuestions":["What is the hard site or room power ceiling today?","Are workloads mixed training/inference with variable draw?","Which partner-measured result (MaxLPS, MLPerf, AgentX) maps to your jobs?"],"valueHypothesis":"Dynamic power allocation and next-gen racks can raise tokens per megawatt when the estate resembles partner AI-factory conditions.","potentialEngagement":"Power-envelope discovery workshop comparing current rack draw to MaxLPS-style reclaim assumptions and a scoped pilot SLO.","unsupportedClaims":["Guaranteed 3.7x or 30x campus gains","Immediate utility interconnect relief","Turnkey SLED DSX factory"]},"engineering":{"fit":"Fit for large AI factories or cloud estates evaluating Vera Rubin/DSX; limited direct fit for single-room campus DGX without partner-class power telemetry.","architectureIntegration":"Integrate DSX MaxLPS/Flex-style power control with rack PDUs, serving stack (Dynamo), and job priority policies.","prerequisites":["Per-GPU/rack power telemetry","Workload priority taxonomy","Validated cooling headroom"],"constraints":["Hyperscale reference envelopes","MLPerf/AgentX workload specificity","Partner-reported—not third-party audited here"],"security":"BlueField and factory security mentioned at marketing depth only; design attestation and tenancy separately.","proofOfValue":["Measured tokens/s within fixed kW budget","Documented shed/restore of low-priority jobs","Side-by-side vs static power caps"]},"delivery":{"work":"Facility telemetry enablement, power-policy definition, and controlled density pilot.","dependencies":["Facilities engineering","utility coordination if Flex","serving-platform team"],"ownership":"Joint facilities + AI platform operations.","skillsAdoption":"Operators need power-aware scheduling literacy beyond GPU ops.","governanceCheckpoints":["Utility/program eligibility review","Workload priority approval","Pilot go/no-go on measured tokens/MW"],"acceptanceCriteria":["Agreed baseline tokens/s at capped kW","Documented reclaim without priority-job breach","Written transferability limits to campus rooms"],"risks":["Over-promising hyperscale multipliers","Cooling/power trip during reclaim","Partner demo non-reproducibility"]}},"retrievedAt":"2026-09-17T03:14:00Z","enrichedAt":"2026-09-17T03:20:00Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Summit narrative is infrastructure/economics focused; no workforce accessibility evidence for public-sector staff.","procurementImplications":"Useful for RFP language on tokens-per-megawatt and power reclaim, but require bidder-measured pilots—not summit headlines alone.","operatingModelImplications":"Implies factory-like ops with dynamic power and job hierarchy; most SLED IT orgs will need new facilities–platform operating agreements.","sourceVerification":{"openedUrl":"https://blogs.nvidia.com/blog/ai-infra-summit-vera-rubin-dsx-energy-efficiencies-tokens-per-watt-ai-factories/","referenceExcerpt":"Lambda ran 19 nodes within the same power budget typically allocated to 16 full-power nodes","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}