From billions of smart devices to self-learning radio systems, digital twins, and accelerated computing are redefining how 6G networks are designed, tested, and deployed.
San Jose, California, 10 December 2025 – The next era of wireless connectivity is taking shape as AI-native 6G networks prepare to support billions of intelligent devices, autonomous agents, and machines. Unlike previous generations, 6G will operate across new and more sensitive radio spectrums, including FR3 (7–24 GHz), where radio behavior becomes increasingly complex. This shift demands a new approach, one that views the network not as fixed infrastructure, but as a living, dynamic, AI-driven system.
Traditional “build and test” methods simply cannot keep pace. The cost, time, and complexity of testing every AI algorithm in real physical environments make old processes unworkable. To unlock the full potential of 6G, the industry must embrace a continuous integration/continuous development (CI/CD) model powered by high-accuracy digital twins.
NVIDIA is helping lead this transformation with a three-computer AI development model that aligns with the needs of software-defined, AI-optimized Radio Access Networks (AI-RAN).
The Three-Computer Framework Driving 6G Innovation
Computer 1: Design and Training
The development journey begins with advanced computing platforms such as NVIDIA DGX and DGX Spark, which accelerate training and simulation for next-generation radio systems. This phase relies on:
- NVIDIA Aerial CUDA-Accelerated RAN is a real-time, high-performance framework for designing and deploying GPU-powered RAN systems.
- NVIDIA Sionna is a GPU-optimized library for modeling and training physical-layer communication systems.
Together, these tools reduce development time and provide a powerful sandbox for testing algorithms before they ever reach a simulation environment.
Computer 2: The Simulation Bridge
Before real-world deployment, next-generation RAN designs must be validated in authentic radio environments. The NVIDIA Aerial Omniverse Digital Twin (AODT) serves as a hyper-realistic simulation platform enabling developers to move from simplified models to full-scale, real-time digital twins.
AODT provides two essential capabilities:
- Physics-accurate RF simulation: It recreates real-world radio behavior using deterministic ray tracing and accurate modeling of antenna arrays.
- Real-time connectivity: A low-latency data fabric links RAN software with the digital twin, enabling closed-loop testing and refinement.
This ensures that what works in simulation will work in reality.
Computer 3: Field Deployment
Once validated, designs move seamlessly into the field using NVIDIA Aerial RAN Computer (ARC), a GPU-accelerated platform built for executing AI-driven RAN functions. The transition is
simplified through the Aerial framework, which automates deployment and speeds up integration.
This creates a unified pipeline from design to testing to rollout.
Breaking Through the Three Barriers Blocking Fully Digital Deployment
The core mission of AODT is to enable a future where cellular networks can be continuously tested, trained, and improved in a safe digital environment. To do this, it addresses three long-standing barriers:
- Accuracy
For 6G, the simulation must match real-world physics.
Traditional models treat antennas as single points—an approach unsuitable for Extremely Large Antenna Arrays (ELAA) and near-field propagation, both crucial to 6G.
AODT uses advanced, element-level ray tracing to deliver results that reliably predict physical performance.
- Integration
Most developers lack the resources to build their own high-fidelity physics engines.
With a gRPC-based architecture, AODT integrates easily with existing C++, Python, or MATLAB simulators, acting as the physics engine for the entire ecosystem.
- Operation
Network operators are understandably cautious about deploying aggressive AI algorithms on live networks.
AODT solves this by running a parallel digital twin, allowing operators to test every update, optimization, and AI decision without risking outages or customer impact.
A New Era for 6G Development
As the world moves toward intelligent, software-defined networks, AI-native 6G will depend on the synergy between accelerated computing, digital twins, and automated deployment pipelines. NVIDIA’s ecosystem of tools is making this future possible faster, safer, and more reliably than ever before.
