Elacity Joins NVIDIA Inception

We've been accepted into NVIDIA's program for startups building with accelerated computing. Here's what that means, what it brings, and where it points.

Elacity has been accepted into NVIDIA Inception, NVIDIA's global program for startups building with accelerated computing. It's free to join, but it isn't automatic: acceptance follows an application and a review of what you're building. Ours came out of conversations at the AI Agent World's Fair in San Francisco, where we met members of NVIDIA's developer community, talked through what we're building, and were encouraged to apply. This month, we were approved.

What Membership Actually Brings

Real, practical things:

Hardware access — member programs and preferred pricing on NVIDIA hardware, including NVIDIA DGX Spark, the desk-sized AI machine we've named as our next target.
Cloud credits — compute credits from NVIDIA's cloud partners, so early experiments don't need early capital.
Engineering depth — free technical training through NVIDIA's Deep Learning Institute, sessions with NVIDIA experts, and guided access to the tools we build on, from CUDA (the toolkit that makes software run fast on NVIDIA chips) to inference tooling (the software that runs AI models and serves up their answers).
Software — SDK access (developer toolkits) and early-access programs across NVIDIA's stack.
Investor reach — exposure to NVIDIA's network of venture investors through Inception Capital Connect.
Go-to-market support — Inception can put member products in front of NVIDIA's customers and partners, with co-marketing opportunities and a seat at events like GTC, NVIDIA's flagship conference.

It's a membership, not a partnership, but for our team building AI at the edge (on devices people own, not in distant data centers), it's exactly the right room to be in.

What We're Building

ElastOS turns a small device in your home into your private cloud: your files, your websites, your wallet, and your own AI assistant — reachable from any browser at your own ela.city domain. It runs today on NVIDIA Jetson Orin (a compact AI computer about the size of a paperback), and the AI runs on the device itself, accelerated by its GPU (the part of the chip built for AI work). Your conversations with your own AI never leave your house.

Beneath it, we're building a new core in Rust (a modern programming language built for safety and speed) — a small, trusted runtime (the base layer that runs and controls all other software) where every app and every AI agent runs sealed in its own sandbox (an isolated box that can't reach anything outside it) and acts only on explicit, revocable permission. This matters now because agents act on your behalf and can be tricked. The only durable defense is limiting what a tricked agent can touch — and that protection has to live in the operating system, not in the agent.

On top sits Elacity's marketplace, where creators seal their work into encrypted Capsules (locked digital packages that only a paying customer's key can open) and sell access directly — keeping their rights, owning their distribution, paid the moment a sale happens. Own the hardware. Own the AI. Own your work. That's the whole idea — and NVIDIA builds the hardware that makes it practical.

Where This Points

Sovereign computing needs two halves: software that answers only to its owner, and hardware powerful enough to carry a private AI. We build the first. NVIDIA's line — from the Jetson in a home appliance to the NVIDIA DGX Spark on a desk — is the second half arriving on schedule, and Inception puts us closer to the people building it.

The near-term path is concrete: Jetson support ships today; NVIDIA DGX Spark — with 128GB of unified memory (one big shared pool of memory, enough to run far larger AI models at home) — is our named next target; and we're evaluating TensorRT-LLM (NVIDIA's software for making AI models answer faster) and NVIDIA NIM (ready-made packages for deploying AI models) for the AI layer. Those are targets, not features: we publish what works and what's still in progress, always.

And this is a door, not a finish line. The same room holds the engineers building the next generation of edge hardware, the training to master it, and the customers, partners, and investors who want to see what runs on it. Every one of those is a path we intend to walk. Our mission and this program point the same way: more intelligence on devices people own, less dependence on clouds they don't — and we've never been better positioned to build it.

If you have a Jetson Orin developer kit, ElastOS is a weekend project — the guide is at docs.ela.city. The new runtime is being built in the open at github.com/Elacity/elastos-runtime. Come see what a computer you actually own looks like.