Compute for machines that move: NVIDIA Jetson modules and devkits, IGX industrial platforms, edge servers and simulation workstations for Isaac Sim — supplied from prototype quantities to production volumes.
| Platform | Class | Typical use | Availability |
|---|---|---|---|
| NVIDIA Jetson Orin Nano / NX | Entry / mid edge AI | Vision AI, drones, smart cameras, AMR prototypes | Devkits typically from stock |
| NVIDIA Jetson AGX Orin | High-end embedded | Autonomous mobile robots, industrial inspection | Frequently from stock |
| NVIDIA Jetson AGX Thor | Next-gen robotics (2026) | Humanoids, on-device VLA / transformer workloads (128 GB) | Devkit ~$3,499; modules on allocation |
| NVIDIA IGX Thor | Safety-critical edge | Functional safety (ISO 26262, IEC 61508), high compute | Project quote |
| NVIDIA IGX Orin | Industrial / medical edge | Functional-safety deployments, long lifecycle | Project quote |
| Pre-integrated edge platforms | Aetina, Advantech, Syslogic | Carrier-board systems on Jetson Thor / Orin | Quoted per model |
| Edge AI servers | Rack / ruggedized | Fleet coordination, on-prem inference near the line | Short lead times |
| Isaac Sim workstations | RTX 6000 Ada class | Simulation, synthetic data, digital twins | Components often from stock |
| Sensors & networking | Cameras, depth, 2D/3D LiDAR, radar; RDMA / 200GbE, 5G/Wi-Fi | Perception stacks and fleet connectivity | Quoted per BOM |
Want it built, not just supplied? See reference architectures (BOMs & pricing) → · Physical AI solutions →
Stock rotates daily — positions are "typically available" and confirmed per request, usually within one business day. Stock guides →
Orin Nano and Orin NX devkits are typically available from stock in Hong Kong; AGX Orin frequently. Production module volumes are quoted with realistic lead times against NVIDIA supply.
Yes. The AGX Thor developer kit is available from around $3,499, and production modules are supplied on allocation with honest lead times. Thor suits humanoids and on-device VLA workloads; AGX Orin remains a cost-effective choice for lighter perception and control — see our Orin vs Thor guide.
An RTX 6000 Ada-class workstation handles serious simulation and synthetic-data generation; we quote complete builds from stocked components, typically from ~$12k.
Edge AI compute is generally less restricted than data-center GPUs, but screening still applies. We check destination and end use on every order — compliant orders ship with full documentation.
Yes — DDP delivery with customs handled, including split shipments: devkits to your lab, production volumes to the contract manufacturer.
Running AI on this infrastructure? Haink also builds the LLM & ML software that runs on it — model, pipeline and GPUs under one contract.
Part numbers or workload description. Pricing, availability and delivered lead time within one business day.