Fortifai (ASX: FTI) has launched the formal roadmap for Nol8’s Agentic AI Data Plane, following completion of its acquisition of the technology platform.
The company said the AI Data Plane is designed to eliminate the “scalability ceiling” that constrains enterprise AI deployment and enable the next generation of autonomous AI agents.
Management reported that Nol8’s demonstration engine has already reduced infrastructure overhead from 5,000 central processing units (CPUs) to a single field-programmable gate array (FPGA) while improving latency from 500 milliseconds (ms) to 3ms, representing a 160-fold improvement.
Nol8’s current technology has achieved an increase in processing throughput from 5,000 to 2,000,000 events per second (EPS), equating to a 400-fold improvement in data handling capability.
The company said these gains were achieved while materially reducing the space and power required to run mission-critical AI workflows in real time.
Addressing The AI Data Bottleneck
By shifting data processing from traditional software running on CPU clusters to neural-network-based algorithms optimised for FPGA architecture, Nol8 targets millisecond-grade processing of high-volume data streams.
The AI Data Plane functions as the high-speed bridge between large language model inference and execution, operating in real time at the speed of the data stream in what the company describes as data-in-motion.
Fortifai said AI adoption is driving a structural shift in global data infrastructure as enterprises move beyond prompt-response models toward persistent agentic systems capable of continuous inference and action—next-generation systems that require continuous high-throughput data ingestion and deterministic low-latency processing measured in milliseconds rather than seconds.
More than 90 per cent of new AI-generated data is unstructured—including text, documents, images, audio, and video—making it materially more compute-intensive than structured database workloads.
Legacy data pipelines built for batch processing and structured datasets are increasingly unable to support the scale, latency, and cost profile required for enterprise-grade agentic AI deployment.
Customer Benchmarking Engine
The roadmap outlines three stages of development—beginning with the proven demonstration engine, followed by a customer benchmarking engine scheduled for July 2026.
Nol8 is in active conversations with enterprise partners across AI verticals to design and implement data benchmark testing in high-performance environments ahead of a revenue-ready commercial platform launch by the end of calendar year 2026.
Following benchmarking, the company plans to release a production-grade commercial engine available for first commercial contracts and designed to support scalable real-time intelligence for global organisations.
Management said the agentic AI market represents a USD four trillion dollar opportunity by 2030, with global data flows projected to expand from 334 zettabytes (ZB) in 2025 to 19,267ZB by 2030, driven predominantly by autonomous AI-generated unstructured data.
‘Widening Structural Gap’
“There is a widening structural gap between having an intelligent AI model and having an actionable outcome,” Nol8 chief technology officer Alon Rashelbach said.
“The AI Data Plane is the high-speed bridge between large language model inference and execution—an architecture we built to operate at the speed of the data stream itself.”
By enabling enterprises to validate high-throughput, millisecond-grade data-in-motion processing against real-world workloads, Fortifai sees Nol8 establishing the execution layer required for scalable Agentic AI.
“This milestone shifts the focus from model capability to infrastructure readiness—ensuring AI systems are architected not just for intelligence, but for sustained, future-proofed autonomous action,” Mr Rashelbach added.
