Oriole Networks Announces PRISM Ultra: The One-Hop Photonic Network Fabric with 50 Exabit per Second Throughput

- Built on PRISM’s proven photonic foundation, PRISM Ultra unlocks unprecedented bandwidth per xPU and a flat one-hop fabric that scales to a million xPUs, far beyond the reach of electronic switching.

PALO ALTO, Calif.--(BUSINESS WIRE)--Oriole today announced PRISM Ultra, the next major evolution of its pure photonic AI networking platform. PRISM (Photonic Routing for Scalable Models) has solved long-standing challenges in nanosecond speed optical circuit switching, transceiver synchronization and control. PRISM Ultra, building on PRISM foundations, introduces a step change in performance, scalability, energy-efficiency and system architecture for the next generation of AI data centers.

PRISM Ultra is the next evolution and a new architecture for frontier-scale AI

PRISM Ultra merges scale-up bandwidth with scale-out connectivity, delivering unmatched bandwidth, including:

- 51.2 Tb/s per xPU across 1,000 zero-hop nodes
- 180ns xPU to xPU latency
- One-hop contention-free connectivity extending to one million nodes

Electronic switching, even with 4x higher performance than current UEC switches and future improved co-packaged optics, would require up to 5 hops and consume the vast majority of the system power to achieve this. PRISM Ultra delivers such a massively connected network with less than 25% of system power, unleashing previously unimaginable AI models and inference efficiency.

PRISM Ultra further enables memory disaggregation and KV-cache pooling, decoupling inference scaling from on-device memory.

“By combining photonic bandwidth and a flat one-hop topology, PRISM Ultra delivers a contention-free, ultra-fast, optically switched network with ultra-low, deterministic latency. This enables synchronously deterministic collective data movement and remote memory access that break today’s model-scaling and memory limits.” - Professor George Zervas, Co-founder & CTO

This results in tightly coupled, distributed computation, faster and more predictable collective operations, and seamless remote memory access, allowing AI workloads to scale in size, speed, and complexity without architectural bottlenecks. PRISM Ultra enables larger models, faster training cycles, and more efficient inference at scales previously infeasible.

To find out more about PRISM Ultra, click here.

Oriole will also be at OFC 2026 in Los Angeles from March 17 – 19. If you’d like to connect to learn more about what we are doing, please reach out to Nina Groothuijzen to set up a meeting or come and find us at booth #5433 in the Corporate Village in the West Hall.

About Oriole:

AI Networking, Reimagined. Oriole Networks is a photonic networking company, developing disruptive technologies for AI/ML and HPC networking that will revolutionize data centers. These technologies address AI’s biggest challenges – speed, latency, and sustainability. Our holistic approach replaces energy-hungry electrical switching with photonic switching. By using only light to move data in the network, our solution will increase the efficiency of LLM training and inference to unprecedented levels while dramatically reducing the energy consumption of data centers, currently putting a huge strain on energy grids. We can offer faster, more efficient, and more sustainable AI without sacrificing the planet.

oriolenetworks.com