June 1, 2026

GTC Taipei is underway, and while there’s no shortage of big AI infrastructure news to absorb, one of the most practically significant updates for embedded and Edge AI teams is worth slowing down for: NVIDIA JetPack 7.2 now brings full JetPack 7 support to NVIDIA Jetson Orin AGX and NVIDIA Jetson Orin NX.

If your team is deployed on either of those platforms, or heading in that direction, this one matters. Here’s why and where partners like Connect Tech (CTI) and CTai LABS fit into the picture.

 

Why JetPack 7.2 Is a Bigger Deal Than It Sounds

When JetPack 7 launched, it was built around NVIDIA’s newest Jetson system-on-module (SoM), Jetson Thor. The Jetson Orin AGX and Jetson Orin NX modules stayed on JetPack 6, which is a solid platform but built on Ubuntu 22.04, an OS base that’s approaching end-of-life. For teams planning multi-year deployments, timelines are important.

JetPack 7.2 closes that gap by bringing Jetson Orin AGX and Orin NX into the same software generation as Jetson Thor, complete with Ubuntu 24.04 and a modern Linux kernel. In practical terms that means:

  • A longer software runway. No more choosing between staying on an aging OS or swapping hardware to keep up.
  • A unified toolchain across the Orin family. Your Board Support Packages (BSPs), workflows, and tooling can align across Jetson Orin NX, Jetson AGX Orin, and Jetson Thor rather than being managed as separate tracks.
  • A real upgrade path from JetPack 5 and 6. It’s important for teams evaluating their upgrade path from JetPack 5 to look at JetPack 7.2 on Jetson Orin as a compelling destination.

Jetson Orin AGX and Jetson Orin NX are at the center of a huge and growing installed base across robotics, industrial automation, intelligent transportation, and precision agriculture. This is not a minor release. It’s investment protection for programs already shipping or nearly there.

 

Yocto Project Gets Its Moment

Alongside JetPack 7.2, NVIDIA has given Yocto Project a formal seat at the table as a supported OS strategy for Jetson deployments. For anyone who’s been watching the embedded Linux space, that’s a notable signal.

The trade-offs between JetPack on Ubuntu and Yocto Project have always been real. Ubuntu is the fast on-ramp: NVIDIA CUDA, NVIDIA TensorRT, NVIDIA DeepStream, and NVIDIA Isaac come pre-integrated and ready to go. But Ubuntu-based images are mutable. Packages drift. When you’re managing a fleet of 50 or 500 devices in the field, that variability becomes a genuine operational risk over time.

Yocto Project solves that with deterministic, image-based builds. Every device in your fleet runs the exact same validated image. The challenge has traditionally been the investment required to get there: custom BSP work, meta-layer maintenance, and a team with embedded Linux knowledge to hold it together. For most organizations, that’s a high bar.

NVIDIA formalizing Yocto Project support in the JetPack 7 era is an acknowledgment that production Edge AI deployments need a more controlled software foundation than a general-purpose Linux distro can provide.

The Hardware Layer: Where Connect Tech and CTai LABS Fit

NVIDIA Jetson modules handle demanding compute at the Edge. Connect Tech carrier boards take that further by meeting the physical demands of the environments where these systems actually get deployed: operating temperatures from -40°C to +85°C, MIL-STD-810H shock and vibration ratings, IP67+ sealed enclosures, multi-camera support via FPD-Link III, GMSL2/3, HDMI, CSI/MIPI, HD-SDI, and expansion via M.2 PCIe and NVMe.

But hardware alone is not enough. Connect Tech’s BSPs make JetPack 7.2 actually run on carrier boards, and that includes full integration of the camera sensors and imaging pipelines customers depend on. BSPs close the gap between a supported module and a system that works correctly in a real deployment. With JetPack 7.2 extending to Jetson AGX Orin and Jetson Orin NX/Nano, Connect Tech has done the integration work, so customers don’t have to. For teams that need additional support getting AI workloads running on Edge platforms, CTai LABS, a department of Connect Tech, offers hands-on integration services to accelerate deployment.

Same hardware investment, updated software stack, validated BSP, all lead to a clear path forward.

 

The Bigger Picture

JetPack 7.2, NVIDIA’s Yocto Project endorsement, and tools, like Peridio’s Avocado OS, are all pointing in the same direction: the Edge AI deployment landscape is growing up. The Edge AI deployment landscape has matured well beyond prototype configurations.

Connect Tech’s priority in this ecosystem is to ensure the hardware layer is never the thing holding a program back. CTI’s carrier boards for Jetson Orin AGX and Jetson Orin NX/Nano are built to hold up in the environments customers actually deploy to as a solid foundation for the software around them to continue to evolve.