FortifAI (ASX: FTI) has reported benchmark results indicating its Nol8 technology may materially outperform existing software-based data processing systems.
Testing against Google’s RE2 engine has showed a significant performance gap under AI-grade workloads.
The company said the results highlight a potential shift in how data is handled in large-scale AI systems, particularly as processing demands increase with the rise of autonomous and agentic applications.
The findings come ahead of a planned June 2026 release of an enterprise-ready benchmarking engine, positioning the technology for commercial evaluation.
Extreme Workload Limitations
Benchmark testing compared Nol8 against RE2, a widely adopted software engine used for high-speed data pattern matching across enterprise systems.
While RE2 performed adequately under low-complexity conditions, results showed its throughput deteriorated rapidly as rule counts and system load increased.
At the highest complexity tier and under extreme load (P99), RE2 throughput fell to 0.007MB/s, effectively stalling under conditions that reflect real-world AI data classification demands.
In contrast, Nol8 maintained a constant throughput of 1,500MB/s across all complexity tiers and load conditions tested—a difference that reflects an architectural shift, with Nol8 using FPGA hardware to process data in parallel rather than relying on CPU-based sequential execution.
This approach removes the need to scale large arrays of processors to handle increasing workloads, which the company said is a key constraint in current AI infrastructure.
Driving an Infrastructure Rethink
The benchmark results are framed against a backdrop of rapidly expanding data volumes, with global data generation expected to increase sharply over the next decade.
FortifAI said most of this growth will come from unstructured data generated by AI systems, which requires real-time filtering, classification, and routing before reaching models.
This processing layer—referred to by the company as the “AI Data Plane”—is emerging as a critical component of next-generation AI infrastructure, and the company is continuing benchmark and optimisation work to translate performance gains into measurable reductions in infrastructure cost and computational load.
Additional results are expected as FortifAI works toward validating the technology’s performance under broader real-world deployment scenarios, with the upcoming enterprise release expected to provide a clearer pathway to commercial adoption.
“The scalability ceiling that has constrained AI data infrastructure is not a software problem, it is an architectural one —Nol8 solves it at the hardware level," Nol8 co-founder Alon Rashelbach said.
