Sensor-dense autonomous systems don’t always need to max out AI compute.
There’s a tendency in Edge AI development to default to maximum performance: grab the most powerful module available, spec the system around it, and ship. That approach works. When you’re building products that need to hit a price point, run efficiently in the field, or scale across a fleet, there are options.
As part of the NVIDIA Jetson Thor series and powered by the NVIDIA Blackwell architecture, the Jetson T4000 brings up to 1200 FP4 TFLOPS of AI compute and 64 GB of memory in a power envelope configurable between 40 W and 70 W. It’s not a cut-down module; it’s a purpose-built one. And for a wide range of real-world Edge AI applications, it could be the right call.
The good news for Connect Tech customers: if you’re already building on a Jetson Orin or Jetson T5000-powered platform, the Jetson T4000 is an option for the Rogue lineup of products.
Rogue-T5 + Jetson T4000: Right-Sized for Sensor-Dense, Harsh-Environment Autonomy
Rogue-T5 was designed for deployments where things get hard fast: rugged field conditions, multi-sensor perception stacks, tight enclosures, and I/O-heavy autonomous systems. Its small form factor and broad connectivity, including GMSL3, GMSL2, SDI, HD-SDI, HDMI, dual 10GBase-T Ethernet, and four CAN channels, make it a strong fit for robotics, industrial automation, and aerospace and defense.
The Jetson T4000 fits that profile well. Projects don’t always need to max out AI compute. They need consistent, reliable inference across a wide range of inputs at a power budget that works in the field. With 1200 TFLOPS of FP4 performance and a 40 W-70 W power range, the Jetson T4000 keeps up with demanding multi-sensor workloads without running hot or breaking the power budget. For teams already deploying Rogue with Jetson AGX Orin, this is a well-supported step forward inside a familiar architecture.
Jetson T4000 at a Glance
| Technical Specification | NVIDIA Jetson T5000 | NVIDIA Jetson T4000 |
|---|---|---|
| AI Performance | 2070 TFLOPS (FP4 - sparse) | 1200 TFLOPS (FP4 - sparse) |
| GPU | 2560-core NVIDIA Blackwell architecture GPU with fifth-generation Tensor Cores | 1536-core NVIDIA Blackwell architecture GPU with fifth-generation Tensor Cores |
| CPU | 14-core Arm® Neoverse®-V3AE 64-bit CPU 64 KB I-Cache, 64 KB D-Cache 1 MB L2 Cache per core 16 MB Shared System L3 Cache | 12-core Arm® Neoverse®-V3AE 64-bit CPU 64 KB I-Cache, 64 KB D-Cache 1 MB L2 Cache per core 16 MB Shared System L3 Cache |
| Memory | 128 GB 256-bit LPDDR5X 273 GB/s | 64 GB 256-bit LPDDR5X 273 GB/s |
| Power | 40 W–130 W | 40 W-70 W |
| Networking | 4x 25GbE | 3x 25GbE |
| Storage | NVMe via PCIe, SSD via USB 3.2 | |
| USB | xHCI host controller with integrated PHY (Up to) 3x USB 3.2 4x USB 2.0 |
|
| * Specifications are subject to change. | ||
The Integration Advantage
Choosing the right module is only part of the equation. Getting it into a working system quickly is the other half. Connect Tech has been doing this for over a decade as an NVIDIA Elite Partner, the highest level of partnership, and it shows. Every Rogue-T5 for the Jetson T4000 ships with an integrated Connect Tech L4T BSP that includes hardware-specific drivers, Linux support aligned with NVIDIA JetPack releases, and tools built to reduce bring-up time. Another unique advantage of working with Connect Tech is CTai LABS, a department of Connect Tech, which works directly with customers on AI application development, optimization, and deployment for Physical AI systems. If the integration gets complex, there’s a team ready to move with you.
The Jetson ecosystem stays consistent across both modules, covering NVIDIA JetPack, NVIDIA CUDA, NVIDIA Isaac ROS, NVIDIA Metropolis, and NVIDIA’s accelerated libraries, which means switching between the Jetson T5000 and Jetson T4000 doesn’t mean starting over.
For media inquiries or interviews, please contact:
- Ceri Nelmes, Head of Marketing
- 519-836-1291
- [email protected]
###
About Connect Tech
Founded in 1985 and headquartered in Guelph, Canada, Connect Tech Inc. (CTI) is a global leader in rugged edge AI computing platforms engineered for robotics, autonomous systems, and mission-critical applications. For over 40 years, CTI has designed and manufactured in-house, delivering reliable edge AI architectures that unify compute, vision, networking, and thermal technologies to accelerate customers’ time to market in demanding real-world environments. Through its CTai LABS AI Engineering team, CTI transforms its EdgeAI Stack into deployable, production-ready physical AI solutions at scale. For more information, visit connecttech.com.