Ethernovia's New Platform Bridges Real-Time AI and Digital Twins

SAN JOSE, CA – March 16, 2026 – As the race to deploy intelligent, autonomous machines accelerates, the immense challenge of processing a relentless flood of real-world data in real time remains a primary obstacle. Addressing this critical bottleneck, Ethernovia, Inc., a leader in high-performance networking silicon, today announced a new networking platform designed to integrate seamlessly with NVIDIA's Holoscan ecosystem. The solution promises to deliver a high-speed, deterministic data pipeline from sensors directly to AI processors, while simultaneously forging a powerful link to virtual simulation environments built on NVIDIA Omniverse.

The company is showcasing the new platform this week at the NVIDIA GTC 2026 Conference, demonstrating a technology that could significantly streamline the development of next-generation robotics, autonomous vehicles, industrial automation, and advanced medical devices.

The Data Bottleneck in Physical AI

Modern autonomous systems are complex sensory organisms. A self-driving car, a factory robot, or a surgical system relies on a suite of high-resolution cameras, LiDAR, radar, and other sensors to perceive and interact with its environment. Each sensor generates a massive stream of data that must be aggregated, synchronized, and delivered to a central AI brain for processing with minimal delay. Any latency in this pipeline can be the difference between a smooth, safe operation and a catastrophic failure.

To manage this data deluge, NVIDIA developed Holoscan, a streaming AI platform, and the Holoscan Sensor Bridge (HSB), a specialized technology for streaming sensor data over Ethernet directly to GPU memory. HSB was designed to replace complex and often proprietary sensor interfaces with a standardized, low-latency, and scalable method. By using Ethernet, HSB allows for longer cable runs and simplified system architectures, but it requires a networking backbone capable of handling the stringent demands of real-time AI.

This is the precise challenge Ethernovia's new platform is built to solve. It provides the physical and logical link that ensures the vast amounts of data captured by sensors don't get stuck in traffic on their way to the AI compute engine, ensuring the system can perceive and react to its environment in lockstep with reality.

An Ethernet Backbone for Intelligent Machines

Ethernovia's new HSB-compatible platform functions as a software-defined Ethernet backbone for physical AI systems. Building on the company's expertise in packet-processing silicon, the board is engineered to transport high-bandwidth, multi-sensor data streams with deterministically low latency. This means the data's arrival time is predictable and consistent, a non-negotiable requirement for synchronized tasks like fusing data from a camera and a LiDAR sensor to identify an object.

By leveraging the HSB standard, the platform sends data directly into the GPU memory of NVIDIA's AI processors, such as the Orin or upcoming Thor systems. This direct-to-GPU path bypasses the system's main CPU and memory, dramatically reducing latency and freeing up the CPU for other tasks. The result is a more efficient and responsive AI system capable of making faster decisions.

“Physical AI systems demand a networking foundation that is deterministic, scalable, and software-defined,” said Ramin Shirani, CEO and co-founder of Ethernovia, in the company's announcement. “Our HSB-enabled platform delivers real-time Ethernet sensor connectivity optimized for NVIDIA Holoscan while extending seamlessly into digital twins built on Omniverse libraries. This creates a unified pipeline from live sensor capture to simulation and AI refinement — dramatically accelerating the development of autonomous and intelligent machines.”

Bridging Reality and Simulation with Omniverse

The most forward-looking aspect of Ethernovia's announcement is the platform's ability to serve as a two-way bridge between the physical and virtual worlds. The same high-fidelity Ethernet data stream captured from live sensors can be seamlessly streamed into applications built on NVIDIA Omniverse, a real-time 3D simulation and collaboration platform often used to create 'digital twins' of physical systems and environments.

This integration creates a powerful closed-loop development cycle. Engineers can:

• Validate AI Models: Stream real-world sensor traffic directly into a simulated environment to test how an AI perception stack performs under live conditions.
• Replay Scenarios: Record deterministic Ethernet data from a field test and replay it in the digital twin to debug complex issues or edge cases.
• Accelerate Training: Use the combination of real and synthetic data to train AI models more rapidly and robustly.
• Reduce Integration Cycles: Test and verify software updates on a digital twin before deploying them to physical machines, significantly reducing risk and development time.

This tight coupling between live sensor data, AI compute, and simulation promises to fundamentally change how complex autonomous systems are built, tested, and maintained. It allows developers to iterate faster and build more reliable systems by blurring the lines between the lab, the simulation, and the field.

A Strategic Piece in the NVIDIA Ecosystem

Ethernovia's launch is more than just a new piece of hardware; it represents a strategic move that reinforces NVIDIA's end-to-end platform strategy for AI. While NVIDIA provides the core compute engines (GPUs) and software platforms (Holoscan, Omniverse), it relies on a robust ecosystem of partners to provide critical enabling technologies. Companies like NXP and Altera (Intel) have also developed solutions for the HSB ecosystem, but Ethernovia's focus on a unified Ethernet fabric that also deeply integrates with the Omniverse digital twin pipeline carves out a crucial niche.

By providing a high-performance networking solution that is purpose-built for the NVIDIA stack, Ethernovia is positioning itself as a key enabler for the thousands of developers building on these platforms. It makes the prospect of developing on Holoscan and Omniverse more attractive by solving a complex and foundational networking problem. As industries from automotive to healthcare increasingly adopt AI, the demand for this kind of integrated, high-performance infrastructure is set to explode. Ethernovia’s platform appears ready to serve as a vital connective tissue powering the next generation of intelligent systems.