SOURCES
Official references, reporting dates and the conditions that affect interpretation.
All sources checked 8 November 2026. Current pages have an access date, not an invented publication date. This ledger is a reference list, not an affiliation list.
Current product page; checked 8 November 2026
NVIDIA presents GPUs, CPUs, networking and software as a complete accelerated computing platform across AI, HPC and rendering.
Use: Infrastructure scope and workload taxonomy. Confidence: High; company description.
Open official sourceCurrent product page; checked 8 November 2026
DGX Cloud Lepton connects development, training and inference across compute environments and provider choices.
Use: Multi-provider discovery context; existing alternative a buyer already owns. Confidence: High; company product description.
Open official source21 October 2026
NVIDIA announced partner-built modular Vera Rubin AI factories, with ecosystem deliveries planned for early 2027. Site preparation remains separate from factory integration.
Use: Deployment dependencies and forward-looking status. Confidence: High for announcement; delivery remains forward looking.
Open official source31 May 2026
NVIDIA announced a production ramp for Vera Rubin systems through its manufacturing ecosystem.
Use: Hardware generation context; not a statement of available capacity on our site. Confidence: High; company announcement.
Open official sourceCurrent page; checked 8 November 2026
CoreWeave presents integrated infrastructure and Forge as complementary layers for production AI.
Use: Buyer voice and platform context. Confidence: High; company description.
Open official source30 September 2026
CoreWeave announced Forge, connecting model and agent development, evaluation and iteration across existing tools and clouds.
Use: Iteration evidence and useful output, beyond GPU count. Confidence: High; announcement.
Open official sourceCurrent product page; checked 8 November 2026
SUNK combines familiar Slurm research workflows with Kubernetes operations, infrastructure health visibility and topology-aware scheduling.
Use: Operations, training and productive capacity. Confidence: High; company description; no performance guarantee adopted.
Open official source26 January 2026
The companies announced an expanded relationship, a US$2 billion NVIDIA equity investment, and an ambition to accelerate more than 5 GW of CoreWeave AI factories by 2030.
Use: Public expansion direction, with explicit forward-looking qualification. Confidence: High for announcement, not completed capacity.
Open official source30 April 2026
CoreWeave announced guided self-service and cross-environment SUNK capabilities, with operational consistency as a focus.
Use: Deployment comparison and orchestration research. Confidence: High; announcement.
Open official sourceCurrent page; checked 8 November 2026
Lambda markets GPU instances, clusters and integrated AI campuses for training and inference. Its current primary website is lambda.com.
Use: Buyer language and infrastructure breadth. Confidence: High; company description.
Open official sourceCurrent product page; checked 8 November 2026
Lambda describes dedicated single-tenant clusters, managed orchestration, observability and specialist support.
Use: Enterprise deployment evidence questions. Confidence: High; product description.
Open official source4 November 2026
Lambda introduced role-based permissions, groups and service identities, with workspace membership and roles controlling access together.
Use: Enterprise readiness and access-control research. Confidence: High; announcement.
Open official source2026 report; checked 8 November 2026
IEA examines data-center electricity growth, grid constraints and power supply options. Global data-center electricity demand grew 17% in 2025; the report distinguishes AI-related growth and local infrastructure bottlenecks.
Use: Energy context; global totals do not size a specific project. Confidence: High; research estimate.
Open official sourceCurrent benchmark page; checked 8 November 2026
MLPerf publishes scenario- and configuration-specific inference benchmark results and methodology.
Use: Benchmark context and comparison discipline; no rankings reproduced. Confidence: High; primary benchmark publisher.
Open official sourceTechnical blog, updated 8 January 2026
The technical overview describes a common interface for interactive, training and inference work across compute environments.
Use: Independent explanation of workflow portability. Confidence: High; technical publisher.
Open official source