Jetson Thor
- Perf
- up to 2,070 TOPS
- Power
- 40 to 130 W
- Size
- 100 mm x 87 mm
Robotics, autonomous machines, full-stack agentic edge
AirfyOS supported
Airfy Compute
The Operating System for AI
in your Home and Business.
AirfyOS runs across the entire Nvidia Jetson family, from the 0.5-TFLOPS Nano to the 2,070-TOPS Thor, plus Blackwell-class servers. One operating system. Same APIs at home, same APIs in the rack. Manage every board by voice from the Airfy app on your phone.
AirfyOS on all 8 Jetsons
Same OS on Blackwell 2U
Voice + app control
AirfyOS runs here. And on 7 more.
One OS across the entire Nvidia line.
The Nvidia Jetson family ranges from a 5-watt Nano to a 130-watt Thor. AirfyOS abstracts the silicon so your apps, voice agents, and operator console look the same on every board.
AirfyOS on every Jetson.
The same operating system on all 8 Nvidia Jetson modules, from the 0.5-TFLOPS Nano to the 2,070-TOPS Thor. Same APIs, same management surface, same upgrade path.
Controlled from the app.
Every Jetson board you provision appears in the Airfy app on your phone. Reboot, redeploy, look at sensors, ship a new container. No SSH-into-the-box rituals.
Made-in-USA firmware.
AirfyOS is written in Texas. FCC-aligned, TAA-aligned. Every board signs a device token at boot. The board is Nvidia, the operating system is yours, the supply chain is auditable.
Supported modules
The Jetson family. AirfyOS on all 8.
Provision, manage, and ship updates to every Nvidia Jetson module from the same Airfy console. Same image, same APIs, same fleet view.
Jetson AGX Orin
- Perf
- up to 275 TOPS
- Power
- 15 to 60 W
- Size
- 100 mm x 87 mm
Industrial AI, multi-camera vision, edge inference servers
AirfyOS supported
Jetson Orin NX
- Perf
- up to 100 TOPS
- Power
- 10 to 25 W
- Size
- 70 mm x 45 mm
Compact AI, retail kiosks, drones, smart cameras
AirfyOS supported
Jetson Orin Nano
- Perf
- up to 40 TOPS
- Power
- 7 to 15 W
- Size
- 70 mm x 45 mm
Entry edge AI, low-power inference, developer kits
AirfyOS supported
Jetson AGX Xavier
- Perf
- up to 32 TOPS
- Power
- 10 to 30 W, 20 to 40 W
- Size
- 100 mm x 87 mm
Long-cycle industrial deployments, certified fleets
AirfyOS supported
Jetson Xavier NX
- Perf
- 21 TOPS
- Power
- 10 to 20 W
- Size
- 70 mm x 45 mm
Compact industrial AI, established Xavier ecosystem
AirfyOS supported
Jetson TX2
- Perf
- 1.3 TFLOPS
- Power
- 7.5 to 15 W, 10 to 20 W
- Size
- 70 mm x 45 mm
Long-life embedded SKUs already in field service
AirfyOS supported
Jetson Nano
- Perf
- 0.5 TFLOPS
- Power
- 5 to 10 W
- Size
- 70 mm x 45 mm
Tinkerers, classrooms, lightweight edge sensors
AirfyOS supported
Thor up to 2,070 TOPS
AGX Orin up to 275 TOPS
Orin NX up to 100 TOPS
Orin Nano up to 40 TOPS
AGX Xavier up to 32 TOPS
Xavier NX 21 TOPS
TX2 1.3 TFLOPS
Nano 0.5 TFLOPS
Blackwell 2U server
RTX PRO 6000 GPU
AirfyOS Made in USA
Thor up to 2,070 TOPS
AGX Orin up to 275 TOPS
Orin NX up to 100 TOPS
Orin Nano up to 40 TOPS
AGX Xavier up to 32 TOPS
Xavier NX 21 TOPS
TX2 1.3 TFLOPS
Nano 0.5 TFLOPS
Blackwell 2U server
RTX PRO 6000 GPU
AirfyOS Made in USA
Thor up to 2,070 TOPS
AGX Orin up to 275 TOPS
Orin NX up to 100 TOPS
Orin Nano up to 40 TOPS
AGX Xavier up to 32 TOPS
Xavier NX 21 TOPS
TX2 1.3 TFLOPS
Nano 0.5 TFLOPS
Blackwell 2U server
RTX PRO 6000 GPU
AirfyOS Made in USA
Thor up to 2,070 TOPS
AGX Orin up to 275 TOPS
Orin NX up to 100 TOPS
Orin Nano up to 40 TOPS
AGX Xavier up to 32 TOPS
Xavier NX 21 TOPS
TX2 1.3 TFLOPS
Nano 0.5 TFLOPS
Blackwell 2U server
RTX PRO 6000 GPU
AirfyOS Made in USA
Plus Blackwell-class 2U servers and RTX PRO 6000 GPUs for the AI controller layer. One platform, one operating system, top to bottom.
One app, eight boards
Controlled from the app.
The Airfy mobile app lists every AirfyOS device in your fleet, from the Nano on the desk to the Thor in the rack. Add a board, ship a container, hand control to a teammate, or pull a stack trace. Voice works too, ask the system what changed in the last hour.
- Provision a new Jetson from the app in under 60 seconds
- Hot-swap containers and configs without SSH
- Voice commands for fleet-wide health, restart, rollback
- Same UI for a Nano in a classroom or a Thor on a factory line
Edge reasoning loop
Your factory never leaves your factory.
The Jetson is not a thin client. It is where the reasoning happens. Cameras, sensors, microphones, PLCs feed the board. The local agent reads the data, decides what matters, and ships only the meaningful signal to the cloud. Your bandwidth bill drops. Your decision latency drops. Your raw data never leaves the building.
- 01Local sensors feed the Jetson.
Cameras, mics, PLCs, OPC-UA streams, Modbus, CAN bus. Whatever protocol your line speaks, the Jetson terminates it on-prem.
- 02The local agent reasons.
AirfyOS runs the inference loop on the Jetson itself. Vision, voice, anomaly detection, quality grading. All on the board, in milliseconds.
- 03Only the meaningful signal ships up.
Not 4K video streams. Not 200,000 sensor pings per second. An event, a verdict, a thumbnail. Bandwidth drops by orders of magnitude.
- 04The cloud aggregates the fleet.
Airfy console shows every Jetson, every site, every event. Audit trail, rollback, retraining, governance. Cloud is the operator surface, not the brain.
Sensors in
Reason here
Signal up
Network and uplink become radio modules. The reasoning is the product.
Factory floor.
Vision QA on every part. Anomaly detection on every motor. No 4K video ever leaves the line. Only the verdict and a thumbnail.
Retail and edge.
Footfall, shrink, voice queries at the counter. The Jetson runs the model locally so the answer comes back before the customer asks again.
Remote and offline.
Starlink and WLAN are radio modules. The Jetson keeps reasoning even when the link drops. Buffer the events, sync when uplink returns.
Own your cloud
Own your cloud. Don't rent it.
Every modern AI platform wants your fleet in their cloud. We put it in yours: your Jetsons, your AirfyOS image, your control plane, hosted by you or your local AI partner. Self-hosted. Partner-owned. Not rented from someone else's stack.
- Your image, your commit. AirfyOS ships as an auditable image. Pin a version, fork a branch, ship to a fleet without a vendor portal in between.
- Your control plane. Each Jetson signs a device token at boot. AirfyOS identity is attestable from the cloud you run, not the cloud we run.
- Your exit path. Open source where it matters. Walk away with your data, your image, your fleet. No re-platforming clause.
The AI five-layer stack: Applications, Models, Infrastructure, Chips, Energy. Airfy plays at App and Infra. Your Jetson is the Infrastructure tier you own, not rent.
Their cloud
Your cloud
Same Jetson. Same workload. The control plane is the difference.
Why AirfyOS on a Jetson, not just JetPack.
Nvidia ships JetPack as the base SDK. AirfyOS sits on top, so the operator surface, identity layer, voice control, and AI agent stack work the same on every module without per-board glue.
Built and maintained by Steffen Siewert and Mona Siewert, with the Airfy team in Austin.
One image, eight boards.
AirfyOS builds against the Jetson Linux base and exposes the same API surface on the Nano, the Orin Nano, the Orin NX, the AGX Orin, the Thor, and the long-life Xavier and TX2 SKUs.
Operator-owned firmware.
AirfyOS is Made in USA and open source where it matters. You can build it, audit it, pin a commit, and ship to a fleet without a vendor portal between you and your hardware.
App-first management.
Every AirfyOS device shows up in the Airfy app on iOS and Android. Provisioning, voice, telemetry, and rollback are one tap, not a JetPack rebuild.
The integration story
Talk to your company. Chat with your system.
Voice your question. Chat your idea. Query your knowledge. Every answer runs on the GPU you own, not on a vendor's cloud. The five Airfy AI products turn your operator network into a nervous system you can speak to.
Airfy Talk
Voice your question. The network listens.
Airfy Chat
Team chat on your GPU. Slack-shaped, operator-owned.
Airfy Metis
Wisdom layer built from Talk + Chat.
All on
Your GPU
Your hardware, your boundary. Nothing leaves your rack.
Stop being training fodder for someone else's model. When you use Slack, your conversations train somebody else. When you use Airfy Chat on your own GPU, they do not.
See the AI PlatformBuy it the way that fits the room you are in.
Airfy products are sold through partners who handle the install, the trust, and the local ownership. Operator-scale buyers can still talk to us directly.
Through your ISP partner
Bundle Airfy with your broadband line.
Subscribers get the Home App and the operator runs Compass on the AI-Network Controller. Same product, your provider's name on the bill.
Through your AI consultancy or MSP
Let an Airfy partner deliver and run it.
MSPs and AI consultancies install the network, set up the AI products on your GPU, and run the AI App Store with you. Local trust, Airfy stack.
Operator-scale direct
Buy directly when the deal is operator-scale.
For ISPs, sovereign customers, or enterprise rollouts above 100 sites: talk to us directly. We connect you with the right partner for the install if you prefer.
Bring your Jetson. We bring the operating system.
Pilot AirfyOS on a single Jetson Orin Nano in two weeks. Scale to a full Thor cluster on the same image when you are ready. Same operating system, same APIs, same peace of mind.