{"resourceId":"ucsd-sdsc-grid-responsive-ai-power-trial-2026","versions":[{"version":"external-7acc9e618fd5440c779de7967b5eaff72b002ec9c38fd97e7965fbcb692e9f7f","resource":{"id":"ucsd-sdsc-grid-responsive-ai-power-trial-2026","title":"UC San Diego plans a grid-responsive AI data-center trial; savings remain projected","organization":"UC San Diego, San Diego Supercomputer Center","sector":"Campus data-center and energy operations","geography":"California, United States","publishedAt":"August 24, 2026","publicationDate":"2026-08-24","eventDate":null,"sourceName":"San Diego Supercomputer Center, Kimberly Mann Bruch","sourceLabel":"University and partner project announcement; not completed campus evaluation","sourceUrl":"https://www.sdsc.edu/news/2026/PR20260824-CEC-datacenter.html","evidenceClass":"vendor-claim","outcomeClass":"emerging","topics":["infrastructure","accessibility-workforce","operating-model"],"finding":"SDSC announces a planned demonstration linking power conversion and AI workload scheduling; it has not reported achieved campus energy savings.","sledRelevance":"Interpretation: campus data-center and utility teams may need coordinated workload and power planning. A research campus with a microgrid differs materially from a small institution buying cloud services.","evidence":"The operator describes an $8.48 million CEC-funded project, two solid-state transformers linking a 12 kV campus grid to 800 VDC and a target of 2 MW of AI compute. Equipment-footprint reduction above 50% and roughly 25% energy savings are projections. Hardware-in-the-loop testing is planned before live deployment.","architectureImplications":"Interpretation: assess power conversion, scheduler interfaces and workload deadlines together. The on-campus design is not evidence of an equivalent benefit in outsourced cloud infrastructure.","governanceImplications":"Interpretation: facility and workload owners must agree who may defer computation, which services are protected and what triggers safe fallback.","securityPrivacyImplications":"Interpretation: segregate utility/control-plane access from compute users, authenticate dispatch signals and minimize disclosure of workload details.","caveats":"Interested-party announcement placed in the claim category. No campus baseline, achieved savings, reliability series or workforce outcomes are reported. Other-site performance claims are not adopted here because underlying studies were not inspected.","streamIds":["campus-operations"],"roles":{"sales":"Interpretation: qualify whether campus facilities, research-computing operations, finance and the utility face a specific capacity or interconnection constraint. Ask which workloads tolerate delay, whether the site controls its power architecture and who bears reliability risk. A bounded feasibility engagement could compare scheduling flexibility, ordinary efficiency measures and infrastructure upgrades. The value hypothesis is better use of constrained electrical capacity if service obligations remain satisfied. This planned demonstration cannot justify a savings guarantee, immediate procurement recommendation or claims that its partners have proved the combined architecture on this campus. Cloud-only institutions have limited direct applicability.","engineering":"Interpretation: map electrical boundaries, compatible equipment, scheduler controls and workload deadlines before any live integration. Test the combined design in simulation and an authorized hardware environment, preserving independent protection and operator override. Validate authenticated control signals and least-privilege access between the utility, facility controller and compute scheduler. Proposed proof should record power at agreed boundaries, task completion and recovery after signal or controller failure. Separate short-term demand reduction from total energy savings and include any later workload rebound. Prerequisites include specialist power engineering, campus operating approval and explicit service commitments; the announcement supplies no completed campus validation.","delivery":"Interpretation: assign joint facilities and computing owners, with the utility and suppliers responsible for agreed interfaces and support. Dependencies include equipment readiness, approved commissioning procedures and a classification of workloads that may be delayed. Train operators on abnormal conditions and provide accessible escalation runbooks. Gate deployment on staged tests and review results before expanding authority. Proposed acceptance: demonstrate safe fallback, preserve agreed job deadlines, and report energy and demand changes against a documented baseline with rebound included. Track training participation separately from job placements. Risks include equipment incompatibility, unclear dispatch authority, delayed research workloads and projected benefits failing to materialize."},"retrievedAt":"2026-09-07T03:01:01Z","enrichedAt":"2026-09-07T03:08:55Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Interpretation: workforce pathways are planned, not demonstrated jobs. Provide accessible training and budget power-electronics, scheduling and operational-safety expertise.","procurementImplications":"Interpretation: request staged acceptance, interoperability and maintenance terms. Budget engineering and commissioning costs rather than applying projected savings directly to a business case.","operatingModelImplications":"Interpretation: utilities, facilities and computing teams need joint change control, incident escalation and deadline accountability.","updateExplanation":"New to the searched archive; included as evidence newly relevant to this first recorded campus-operations edition, not asserted to be newly published today.","sourceVerification":{"openedUrl":"https://www.sdsc.edu/news/2026/PR20260824-CEC-datacenter.html","referenceExcerpt":"the technology will undergo extensive testing including simulations","promptVersion":"sled-research-v3.1","model":null,"basis":"agent-reported inspection"}}}]}