That's Lambda in the NVIDIA GTC Berlin keynote. We're building with @NVIDIA on secure, efficient AI infrastructure, including an NVIDIA DSX MaxLPS proof of concept, plus support for Confidential Bare Metal and the NVIDIA OpenShell Reference Architecture. #NVIDIAGTC

Power is the hardest constraint in AI infrastructure right now. Shawn Smith, Lambda's VP of Energy Infrastructure and Grid Development, joins the panel "Power plays - Surviving the new energy constraints" at @dcdnews Connect Virginia, Nov 3–4.

Heading to @stacresearch Summit NYC on Oct 21? Lambda's a platinum sponsor this year. Catch Jessica Nicholson and Devina Jain at 10:20 a.m. for "FSI at Lambda: research at the edge, performance in production." Then Quinn Kennedy gives the five-minute version of how Lambda got here at 2:25 p.m. We'll have our audited STAC-AI LANG6 results with us. Find us between sessions. Book time with the team: lambda.ai/stac-ny-book-a-mee…

Three weeks of Build-a-Claw hackathons across Europe, and today the winners took the stage in Berlin. Congrats to Sebastian Kwiedor and HOAIclaw, an AI advisor for German SMEs that keeps company data local and completed 3,200+ scheduled jobs with zero failures in production. And to Pasha Abdulaev and Kautionskralle, which helps German renters recover their deposits by tracking deadlines and drafting objection letters. We’re glad to support a community building with agents, open models, and physical AI.

Moving from prototype to production? Lambda is at @NVIDIA GTC Berlin 2026, booth 7022. Our team is there all week to talk about scaling compute infrastructure for ML workloads. If you're working through capacity, deployment, or production-readiness decisions, book a time with us. lambda.ai/gtc-berlin-book-a-…

We also know that users care about multiple hardwares and the precisions at which they're being served at, so if applicable the cards will also showcase various quantizations such as NVFP4 vs BF16, and serving on H100 vs B200:

Today we are excited to announce a fresh update on the Lambda Model Cards! Understanding how serving models requires more than knowing the raw tokens/s, and serving real users means caring about more than just concurrency at 1. As of today, our model cards now showcase benchmarks off of @NVIDIAAI's AIPerf benchmarking package on minimal Lambda hardware. With this, you can now properly get an understanding of how your model deployments will behave at certain user thresholds, what its interactivity will look like, and know when you need to scale up to more replicas. lambda.ai/inference-models

Lambda has closed its second institutional debt facility and its first $1 billion-plus GPU debt financing. The facility was oversubscribed and rated A (low) by Morningstar DBRS and Baa1 by Moody's. Proceeds will fund the purchase and development of GPU infrastructure supporting three committed customer deployments with two investment-grade offtakers across multiple data centers. Read the announcement: lambda.ai/blog/lambda-closes…

A realistic video is not enough for a robot to act reliably. That was the central question at CVPR 2026’s WMAS workshop. RoboWM-Bench, the Best Paper, found that a video world model can produce convincing footage while still failing when its predicted behavior is turned into robot actions. The failures came from problems such as spatial reasoning and unstable contact. GEM-4D improved real-world manipulation success from 61% to 81% by adding geometric consistency across generated frames. SAW-Bench found a 37.66-point gap between humans and the best multimodal model on situated-awareness tasks. lambda.ai/blog/world-models-…