Volume 342, Issue 5 · 13 articles from 2 issues
CoreWeave and Weights & Biases describe the 'AI Loop,' a framework linking training, inference, observability, and evaluation into one cycle. Read more
CEO Michael Intrator recaps CoreWeave's first Fully Connected conference and its new Forge platform announcements. Read more
CoreWeave makes ARIA generally available: an agent that analyzes experiments, proposes next steps and launches them with user approval. Read more
CoreWeave's Brian Venturo argues AI can connect knowledge across disciplines, and the infrastructure stack decides who gets to build. Read more
CoreWeave Notebooks, built on marimo, bring reactive Python and SQL notebooks connected to experiments and evaluation data into Forge. Read more
CoreWeave launches Forge, one environment to serve, observe, curate, build and evaluate AI models and agents in a continuous improvement loop. Read more
CoreWeave made its Mission Control MCP generally available and extended the Mission Control Agent into its console for AI workload operations. Read more
CoreWeave Registry offers versioning, aliases, lineage, permissions and discovery for models, agents, datasets and other AI assets. Read more
CoreWeave introduced IAM and Remote Key Encryption to let enterprises federate existing identity providers and key management into AI workloads. Read more
CoreWeave Forge combines Weights & Biases, OpenPipe and marimo with CoreWeave services in one AI development layer. Read more
CoreWeave puts NVIDIA Vera Rubin NVL72 into production, with Cognition reporting up to 4.8x higher token throughput per GPU. Read more
theCUBE analysts say inference growth makes tokens per watt the key metric, positioning CoreWeave beyond GPU compute. Read more
CoreWeave Forge unifies W&B, OpenPipe and marimo into one environment for the run-observe-improve AI loop. Read more