# NVIDIA Releases JetPack 7.2 with NemoClaw Support > Source: > Published: 2026-06-20 07:07:40.133711+00:00 # NVIDIA Releases JetPack 7.2 with NemoClaw Support NVIDIA announced JetPack 7.2, an update to its Jetson edge AI platform, at GTC Taipei on June 1, 2026. According to NVIDIA's developer and corporate blogs, the release adds one-command deployment of NemoClaw - an open-source enterprise AI agent blueprint - along with a new library of Jetson agent skills for robotics and automation workflows. JetPack 7.2 also includes CUDA 13 on Orin, official Yocto Project support for production Linux builds, Multi-Instance GPU (MIG) support on Jetson Thor, and a "Super Mode" for Jetson AGX Orin 32 GB that lifts AI throughput from 200 to 241 TOPS at 60W (up from 40W at 200 TOPS), per embedded partner Syslogic. Deepu Talla, NVIDIA's vice president of robotics and edge computing, is quoted in the NVIDIA corporate blog: "Agentic AI is here, and Jetson's programmability and high performance enable developers to instantly deploy physical AI agents in production at the edge." The Edge AI and Vision Alliance framed the release as a production-readiness milestone for physical AI deployments. ### What happened NVIDIA announced JetPack 7.2 on June 1, 2026, per the NVIDIA corporate and developer blogs. According to NVIDIA's technical post, the release includes one-command deployment of NemoClaw - NVIDIA's open-source enterprise AI agent blueprint - along with Jetson agent skills, CUDA 13 on Orin, official Yocto Project support, Multi-Instance GPU (MIG) support on Jetson Thor, and a "Super Mode" performance option for Jetson AGX Orin 32 GB. Deepu Talla, NVIDIA vice president of robotics and edge computing, is quoted in the NVIDIA corporate blog: "Agentic AI is here, and Jetson's programmability and high performance enable developers to instantly deploy physical AI agents in production at the edge." ### Technical context NemoClaw is described by NVIDIA's newsroom (May 31, 2026) as an open-source blueprint for deploying autonomous agents with policy-based security and privacy governance on local or cloud hardware, complementing the broader NVIDIA Agent Toolkit for physical AI. One-command deployment via JetPack 7.2 consolidates runtime dependencies, container images, and device-specific drivers into a single setup step. The Super Mode for Jetson AGX Orin 32 GB raises AI throughput from 200 TOPS to 241 TOPS - an increase of more than 20% - by lifting the power envelope from 40W to 60W, according to embedded systems partner Syslogic (June 2, 2026). These gains bring the 32 GB module close to the performance of the 64 GB variant for generative AI and robotics workloads. ### Developer implications Official Yocto Project support, highlighted by NVIDIA and covered in detail by the Edge AI and Vision Alliance, shifts Jetson from an Ubuntu/L4T-centric reference platform toward a workflow compatible with production embedded Linux practices. Yocto builds allow deterministic image creation, smaller footprints, and easier supply-chain integration - standard requirements for OEM product builds. MIG support on Jetson Thor and real-time kernel capabilities enable isolated, deterministic execution for multiworkload robotics and safety-critical industrial applications. ### Context and significance The Edge AI and Vision Alliance frames JetPack 7.2 as a step moving agentic AI from research demos toward deployable physical agents. For practitioners, the release reduces integration risk when porting agentic frameworks to constrained devices, consolidating deployment tooling, runtime isolation, and production Linux support into a single SDK update. This matters to OEMs, systems integrators, and robotics teams requiring reproducible, maintainable edge stacks. The 241 TOPS Super Mode figure is reported by NVIDIA and its OEM partner Syslogic; no independent third-party benchmarks have been published as of this writing. ## Scoring Rationale JetPack 7.2 consolidates NemoClaw deployment, Yocto production Linux support, MIG isolation, and a measurable 20%+ TOPS gain into a single SDK release, meaningfully lowering the bar for production-grade agentic edge AI. The update is significant for embedded AI practitioners and OEMs but is Jetson-platform-specific and driven primarily by NVIDIA and partner reporting, placing it in the useful new tool / notable range rather than industry-shaking. Practice interview problems based on real data 1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with. [Try 250 free problems](/problems)