From the NVIDIA edition of September 16, 2026
Military.com independent report: NPS DGX GB300 online with ~135 kW and data-path bottlenecks
Military.com · Defense media / independent reporting · United States (NPS Monterey coverage)
- Publisher
- Military.com
- Original publication
- September 6, 2026
- Source retrieved
- 2026-09-17
What happened
Military.com's Haley Fuller reports NPS brought the first U.S. military DGX GB300 online for education/research, stressing storage/network bottlenecks (DDN interview), partner stack (NVIDIA donation, Vertiv power/cooling, VAST, DDN), ~135 kW peak with liquid cooling, and the gap between campus research systems and ship/expeditionary power-space limits before fleet tooling.
Why it matters
Newly relevant independent narrative beyond archived NPS press: reinforces facility power figures and adds data-integrity/poisoning and fleet-transfer constraints useful for education and government buyers evaluating Blackwell on-prem.
Evidence and measured results
Article published Sep 6, 2026 quotes DDN's Kevin Delane on data-path bottlenecks; states 72 Blackwell Ultra + 36 Grace; lists Vertiv/DDN/VAST/NVIDIA roles; explicitly notes ~135 kW peak and liquid cooling; discusses training-data integrity and DoD continuous monitoring expectations.
Limitations and uncertainty
Relies partly on vendor executive interview (DDN); not a controlled benchmark. NPS commissioning already archived from primary sources—this adds independent framing and fleet-transfer caution.
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
- Defense education buyers need on-prem AI without confusing research kits for operational products.
- Stakeholders
- CIO
- research computing
- fleet transition sponsors
- Discovery questions
- Is the goal education, research, or fleet tooling?
- What storage throughput matches GPU feed rates?
- Who funds facility liquid cooling?
- Value hypothesis
- A GB300 research platform can teach AI limits and accelerate models if data and facility gates are funded.
- Potential engagement
- Research-platform discovery covering power, storage, and integrity controls.
- Unsupported claims
- Immediate fleet deployment of campus models
- GPU-only bottleneck view
- Donation implies zero facility cost
Pre-sales engineering
Role takeaway
- Fit
- Fit for graduate defense education HPC; not shipboard without redesign.
- Architecture and integration
- DGX GB300 + high-throughput storage/network + liquid cooling plant.
- Prerequisites
- ~135 kW-class power
- liquid cooling
- HPC storage
- Constraints
- Expeditionary power/space
- Partner-donated stack specifics
- Security
- Data provenance and poisoning defenses for training corpora.
- Proof of value
- Sustained GPU utilization without storage stall
- Integrity alerts exercised
- Documented edge-porting plan
Delivery
Role takeaway
- Work
- Commissioning support, storage integration, curriculum/research onboarding, integrity ops.
- Dependencies
- facilities
- storage vendor
- research security
- Ownership
- Research computing CIO office.
- Skills and adoption
- Faculty/students need data-pipeline literacy.
- Governance checkpoints
- CRADA/gift review
- Research-security approval
- Fleet-transition gate
- Acceptance criteria
- System online for approved users
- Power/cooling within published envelope
- Written non-goals for shipboard use
- Risks
- Storage under-provisioning
- Overclaiming operational readiness
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?
Design storage and east-west data movement as first-class with GPUs; plan distillation to smaller edge form factors for operational use.
Governance
Who approves, reviews and stays accountable for outcomes?
Education/research framing with partner donations differs from fleet acquisition; map CRADA/gift rules carefully.
Security and privacy
What data, permissions and controls need testing?
Highlights training-data poisoning risk and need for integrity monitoring beyond storage ACLs.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Framed as educating officers who return to Joint Force roles—workforce development is the primary outcome metric suggested.
Procurement
What should contracts, pricing and exit terms secure?
Donation/partner model is not a competitive buy; still budget storage, cooling, and integrity tooling.
Operating model
Which teams own the service once it runs?
Separate research platform ops from fleet product engineering ownership.
Publication history
- 2026-09-16NVIDIA · Issue 119 resources
Stable resource ID: military-com-nps-dgx-gb300-storage-power-20260906