Opanga Delivers Native Real-Time Customer Intelligence on NVIDIA BlueField-3 to Accelerate AI-Driven and Autonomous Networks
March 4, 2026 — Opanga today announced a reference architecture that enables its real-time customer experience data records (CXDRs) to run natively on the NVIDIA BlueField-3 DPU. By moving production-grade networking intelligence directly into the data path, the architecture allows service providers to accelerate advanced operational use cases today while establishing a scalable foundation for AI-driven and autonomous networks.
Embedding Intelligence Directly in the Data Path
For service providers, deploying advanced network intelligence has historically required host-based agents, application redesign, or additional infrastructure. Opanga’s reference architecture removes those barriers.
Opanga’s containerized runtime, which the company calls RAIN, now operates directly on the BlueField-3 DPU, delivering in-line, real-time customer experience intelligence without adding host overhead or requiring custom workflows. This native integration enables immediate deployment of production-ready intelligence at line rate.
With Opanga integrated at the hardware level, operators can:
Gain real-time visibility into application, workload, and network behavior without impacting host CPU or memory resources
Deploy in-line intelligence to support both operational and customer-facing use cases
Simplify large-scale deployment of advanced observability and control capabilities
The result is immediate time-to-value, with no need to redesign existing systems or expand compute infrastructure.
Why NVIDIA BlueField-3
NVIDIA BlueField-3 provides the processing capacity and hardware offload required to port Opanga’s real-time flow detection and customer experience scoring directly onto the DPU. Capabilities that previously required high-performance NICs combined with dedicated compute infrastructure now operate as a pluggable, native function on the DPU.
By executing intelligence directly on BlueField-3, Opanga reduces host CPU utilization and memory consumption while eliminating the need for additional analytics servers. Service providers can deploy advanced telemetry, scoring, and observability using infrastructure already present in their network and OSS/BSS environments.
An AI-Enablement Layer for Modern Networks
Beyond operational efficiency, the architecture establishes Opanga’s real-time CXDRs as a high-fidelity AI enablement layer.
Intelligence executed in-line on the DPU performs filtering, transformation, and customer experience scoring at the point of data generation. Instead of streaming raw telemetry to centralized systems, operators receive high-quality, session-level CX intelligence ready to power AI-driven analytics and automation.
This approach reduces data processing burdens on host compute and enables AI systems to operate on real-time, customer-validated intelligence rather than post-processed datasets. The result is more efficient AI execution, faster decision cycles, and improved automation accuracy.
From Accelerated Use Cases to Autonomous Networks
While the reference architecture accelerates current networking and operational use cases, it also supports a broader architectural evolution toward intent-based and autonomous networks.
By executing real-time intelligence on the DPU, Opanga delivers continuous, session-level insight into how services are actually experienced by customers. This customer experience ground truth can be used to validate infrastructure behavior against service intent, detect operational drift before customer impact occurs, and inform closed-loop automation systems.
“Autonomous networks require more than automation—they require real-time, customer-validated intelligence operating directly in the data path,” said Ben Hadorn, VP of Customer Engineering and Operations at Opanga. “By running natively on NVIDIA BlueField-3, we’re embedding customer experience ground truth into the infrastructure itself. That changes how operators deploy AI, validate intent, and progress toward true closed-loop operations.”
The reference architecture will be showcased at Mobile World Congress in Barcelona, demonstrating how hardware-native CX intelligence enables immediate operational value and a scalable path toward AI-driven network automation.