NVIDIA DSX Air Boosts Time to Token With Accelerated Simulation for AI Factories
Source: NVIDIA AI Blog
AI Factories — From Months to Days
The ability to set up AI factories in simulation—cutting deployment time from months to days—is accelerating the next industrial revolution.
The Highlight: GTC 2026
At GTC 2026 in San Jose, NVIDIA founder and CEO Jensen Huang unveiled NVIDIA DSX Air.
- What it is: A SaaS platform for logically simulating AI factories.
- Part of: NVIDIA DSX Sim in the DSX platform – NVIDIA’s blueprint for AI factories.
Core Capabilities
| Capability | Details |
|---|---|
| High‑fidelity digital twins | Simulates NVIDIA hardware (GPUs, SuperNICs, DPUs, switches). |
| Open, API‑based integration | Connects to partner solutions for storage, routing, security, orchestration, etc. |
| Full‑stack ecosystem | Unites NVIDIA infrastructure with third‑party technologies for end‑to‑end simulation. |
Real‑World Impact
- Early validation: Companies like CoreWeave use DSX Air to model and validate environments before hardware arrives at the loading dock.
- Speed to production: Shifting integration and troubleshooting into simulation reduces time‑to‑first‑token from weeks or months to days or even hours, delivering massive time‑ and cost‑savings.
How It Works – An Analogy
Imagine mirroring your laptop to set up a new one.
In this case, the “laptop” is a hyperscale AI factory, and the “mirroring” is a complete, high‑fidelity replica of the production environment.
Why It Matters
For operators racing to bring new AI capacity online, simulation‑first deployment is a transformative shift—making AI factories faster, cheaper, and more reliable.
Building a Platform for an Entire Ecosystem
The NVIDIA DSX Air simulation platform lets every player in the AI‑factory ecosystem—server manufacturers, orchestration vendors, storage providers, and security partners—validate their solutions alongside NVIDIA infrastructure, all in a single, scalable virtual environment.
Why It Matters
| Stakeholder | Pain Point | How DSX Air Helps |
|---|---|---|
| Server manufacturers | Need physical labs to prove reference architectures; enterprise AI workloads are highly customized. | Create digital twins of customer‑specific configurations, run software‑stack tests, and deliver validated solutions without any hardware. |
| Orchestration vendors | Must prove multi‑tenant, turnkey AI services at scale. | Simulate entire multi‑tenant RTX PRO Server environments (e.g., Netris networking, Rafay host orchestration, Run:AI GPU allocation) and validate complex workflows without real clusters. |
| Data‑platform partners | Large physical clusters are costly for end‑to‑end AI workflow validation. | Model complete AI pipelines—such as VAST AI OS video‑retrieval‑augmented generation—inside DSX Air, from DataEngine nodes to front‑end search/summarization. |
| Security vendors | Require rigorous multi‑tenant, DPU‑accelerated isolation and threat‑detection testing. | Run distributed firewalls, AI‑based threat detection, and realistic traffic generators (e.g., Check Point on BlueField DPUs, TrendAI Vision One, Keysight Cyperf) in a customer‑specific digital twin. |
Partner Highlights from GTC
- Server manufacturers can now validate reference designs without building expensive labs.
- Orchestration vendors demonstrated a fully simulated multi‑tenant RTX PRO Server stack, with Netris, Rafay, and Run:AI handling networking, host orchestration, and GPU scheduling respectively.
- Data‑platform showcase: A VAST AI Operating System video‑retrieval‑augmented generation workload ran end‑to‑end—DataEngine nodes processed and indexed video, and the front‑end delivered search & summarization—all inside the DSX Air simulation.
- Security demo: Check Point’s distributed firewall on simulated BlueField DPUs, TrendAI Vision One for threat detection, and Keysight Cyperf traffic generation proved multi‑tenant policy enforcement and vulnerability detection before any physical deployment.
The Bottom Line
DSX Air provides a complete, scalable, and cost‑effective way to validate solutions together with NVIDIA compute, networking, and software—before any hardware is ever touched.
Operating with a New Model to Accelerate Time‑to‑Token
NVIDIA DSX Air isn’t just a deployment accelerator—it introduces a fresh operational model for AI factories.
1. Build in Simulation (Day 1)
- Full‑stack recreation – networking, compute, storage, orchestration, security, and scheduling are configured exactly as they will be in production.
- End‑to‑end validation – verify that every component works together, surface issues early, and confirm expected behavior.
2. Deploy with Confidence
- Because the environment has already been tested, the likelihood of a smooth bring‑up increases dramatically.
- Time‑to‑first‑token shrinks, letting teams focus on running workloads instead of troubleshooting infrastructure.
3. Safe Change Management (Ongoing)
- Long‑lived simulations serve as a sandbox for:
- Testing upgrades
- Rehearsing maintenance windows
- Validating patches
- Predicting operational impact
- Changes are only applied to the live environment after they succeed in simulation, maximizing uptime and ensuring infrastructure availability.
This lifecycle approach shows how modern AI factories can scale efficiently while maintaining reliability.
Simulating AI Factories: The Backbone of Modern AI Infrastructure
GTC demonstrated that simulation is no longer a future concept—it’s now the core of AI infrastructure deployment and operations.
- NVIDIA DSX Air lets customers and partners simulate every aspect of their AI workloads in a single environment.
- Benefits include:
- Faster deployment cycles
- Reduced risk during rollout
- Guaranteed day‑one performance at scale
Adopting NVIDIA DSX Air to Accelerate Deployments With Simulation
Siam.AI, Thailand’s largest AI cloud provider, has accelerated its infrastructure deployment with NVIDIA DSX Air. By using simulation, Siam.AI embraced NVIDIA best practices well ahead of schedule, ensuring day‑one operational expertise and validating its architecture in a virtual environment before the physical hardware even arrived.
Similarly, Hydra Host is using DSX Air to speed up development of Brokkr, its AI‑factory operating system for bare‑metal GPU provisioning that powers dozens of GPU deployments worldwide. By simulating full‑stack environments in DSX Air before production rollout, Hydra Host can validate Brokkr’s automation and orchestration workflows across diverse networking and hardware configurations at scale. This simulation‑first approach lets Hydra Host ship validated infrastructure faster to customers while minimizing risk to live systems as global AI demand grows.
As AI factories grow in size and complexity, the ability to validate full‑stack environments before hardware arrives will define the pace of innovation. NVIDIA DSX Air delivers that capability today, giving organizations the fastest possible path to first token and a more reliable way to operate AI infrastructure over time.