Justin Boitano

Palantir has launched a new intelligent engine using NVIDIA Nemotron open models to serve U.S. government agencies in air-gapped, secure environments. The system allows federal agencies to build custom AI models on their own infrastructure, train on proprietary data, and retain full ownership of model weights. This deployment combines Nemotron's open models with Palantir's Sovereign AI Operating System to enable trusted, auditable AI in sensitive government operations.

NVIDIA has released Agent Toolkit, an open-source foundation for building specialized AI agents that can reason, use tools, and take action within enterprise workflows. The toolkit combines customizable models, tools that connect to existing systems, and a secure runtime environment. Companies like CrowdStrike, Cadence, and Synopsys are already deploying specialized agents for security, chip design, and other domain-specific tasks.

SAP and NVIDIA announced an expanded collaboration to embed NVIDIA's OpenShell runtime into SAP Business AI Platform, providing security and governance controls for autonomous agents operating in enterprise systems. OpenShell offers isolated execution environments, policy enforcement, and infrastructure-level containment to guard against failures when agents access systems of record across finance, procurement, supply chain and manufacturing. SAP engineers are codesigning OpenShell alongside NVIDIA while contributing back to the open source project, and NVIDIA's NemoClaw reference blueprint will be available directly in Joule Studio to accelerate agent development.

OpenClaw, a self-hosted persistent AI agent framework created by Peter Steinberger, reached 250,000 GitHub stars in 60 days by offering local deployment without cloud dependencies. Unlike traditional agents that respond to prompts and stop, OpenClaw agents run continuously in the background, checking task lists at regular intervals and acting autonomously. The rapid adoption has surfaced security concerns around data management and local deployment risks, prompting NVIDIA to collaborate with the project on hardening defenses and introducing NemoClaw, a reference implementation with security defaults.

OpenAI's GPT-5.5 model, running on NVIDIA's GB200 NVL72 infrastructure, now powers Codex, an agentic coding application that NVIDIA has deployed across its workforce. Over 10,000 NVIDIA employees across engineering, product, legal, marketing, finance, sales, HR, operations, and developer programs are using the system, reporting significant productivity gains including debugging cycles compressed from days to hours and feature shipping accelerated from weeks to overnight. The deployment reflects a decade-long partnership between the two companies and demonstrates enterprise-scale viability of frontier model inference through improved economics: 35x lower cost per million tokens and 50x higher token output per second per megawatt compared with prior systems.

NVIDIA is donating its Dynamic Resource Allocation (DRA) Driver for GPUs to the Cloud Native Computing Foundation, moving the software from vendor control to community ownership under the Kubernetes project. The driver simplifies GPU resource management in Kubernetes environments by enabling smarter sharing, multi-node scaling, dynamic reconfiguration, and fine-tuned resource requests. NVIDIA is also collaborating with the CNCF to add GPU support for Kata Containers, extending hardware acceleration into lightweight virtual machines for enhanced security in confidential computing scenarios. The donation, announced at KubeCon Europe, reflects a broader industry effort involving AWS, Google Cloud, Microsoft, Red Hat, and others to standardize high-performance AI infrastructure.