From the Research Universities edition of September 16, 2026
NSF Expanse2 award funds 2027 SDSC capacity for AI/HPC research; outcomes still prospective
San Diego Supercomputer Center at UC San Diego (NSF award) · University research computing / national cyberinfrastructure · United States (California; national ACCESS/NAIRR user community)
- Publisher
- San Diego Supercomputer Center
- Original publication
- September 15, 2026, SDSC news (PR20260916)
- Source retrieved
- 2026-09-17
- Event date
- 2026-09-15
What happened
NSF awarded $10M for Expanse2 to expand AI/HPC capacity and NAIRR-facing access, but production is planned for 2027 and no Expanse2 scientific-outcome metrics are published yet.
Why it matters
Directly relevant to U.S. university research computing and NAIRR-linked infrastructure planning; announcement defines capacity and education goals, not measured discovery gains.
Evidence and measured results
Published September 15, 2026: $10M NSF award (2614012) to SDSC/UC San Diego to deploy NSF Expanse2 replacing Expanse (since late 2020); AMD processors/accelerators; all-NVMe AI/ML storage plus capacity tier; Jupyter and AI/ML frameworks; EOT for universities, teaching colleges, high schools; industry discretionary allocations; production in 2027; intended NAIRR contribution. No utilization, time-to-science, or reproducibility metrics for Expanse2 reported.
Limitations and uncertainty
Prospective system entering production in 2027. Institutional announcement, not an independent evaluation. Architecture details are high-level. Prior Expanse user counts do not prove Expanse2 outcomes.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-09-17; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Campuses need foreseeable national AI/HPC capacity for long-tail science without overstating when benefits will arrive.
- Stakeholders
- Research computing leadership
- Vice president for research
- NAIRR/ACCESS liaisons
- College research facilitators
- Discovery questions
- Which workloads are blocked on current Expanse or commercial GPU limits?
- What metrics define successful Expanse2 onboarding?
- Who prioritizes AI training vs traditional HPC queues?
- Which data classes can use federated ecosystems?
- Value hypothesis
- A timed national system can reduce ad-hoc commercial GPU sprawl if allocation and storage policies are explicit before production.
- Potential engagement
- Bounded 2027 readiness assessment for one research center: portability checklist, data classes, and allocation strategy.
- Unsupported claims
- Measured discovery acceleration from Expanse2 today
- Guaranteed NAIRR wait-time cuts
- Automatic compliance with every sponsor rule
Pre-sales engineering
Role takeaway
- Fit
- National NSF HPC resource for mixed AI/ML and traditional simulation via batch and gateways.
- Architecture and integration
- Map campus workflows to Expanse2 queues, Jupyter, NVMe-hot storage, and capacity tier; retain local scratch/egress controls.
- Prerequisites
- Working allocation pathway
- Container/module portability tests
- Data classification
- User training plan
- Constraints
- 2027 production date
- AMD/accelerator software ecosystem
- National queue competition
- Security
- Enforce data-class rules on federated datasets; audit gateway tokens and shared project spaces.
- Proof of value
- Pilot job success/error rates
- Queue wait vs baseline
- I/O throughput on representative AI workflows
Delivery
Role takeaway
- Work
- Prepare pilot ports, publish eligible-use guidance, and train first research teams before production claims.
- Dependencies
- SDSC/NSF schedule
- Campus identity and storage teams
- PI time for workload selection
- Ownership
- Research computing operates access; PIs own scientific validation.
- Skills and adoption
- Short workshops on AI/ML frameworks, Jupyter vs batch, and storage tier choice.
- Governance checkpoints
- Pre-production allocation policy sign-off
- Shadow-run review
- 30-day pilot incident review
- Acceptance criteria
- Pilot users run authenticated jobs
- Scheduling rules published
- Telemetry retained for pilot cohort
- Risks
- Premature benefit claims
- Queue contention
- Data misclassification on federated paths
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Plan heterogeneous CPU/GPU scheduling, NVMe-hot paths for AI workflows, and gateway/Jupyter access alongside batch—while keeping allocation policies distinct from scientific acceptance criteria.
Governance
Who approves, reviews and stays accountable for outcomes?
Treat the award as a capacity roadmap; require allocation transparency and preregistered service metrics before claiming AI-enabled discovery acceleration.
Security and privacy
What data, permissions and controls need testing?
Federated data ecosystem claims still need campus data-classification rules and sponsor-specific residency controls for NAIRR-linked workloads.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
EOT across teaching colleges and high schools can widen access, but still teach allocation literacy and data classification.
Procurement
What should contracts, pricing and exit terms secure?
Compare national allocation opportunity cost with commercial GPU contracts using local workload telemetry before multi-year cloud commitments.
Operating model
Which teams own the service once it runs?
Central research computing brokers ACCESS/NAIRR; colleges sponsor pilots; PIs remain accountable for scientific claims.
Publication history
- 2026-09-16Research Universities · Issue 114 resources
Stable resource ID: sdsc-expanse2-nsf-award-20260915