Fortifai Ltd Nol8 Joins Megaport AI Exchange for Data Plane POCs

By Josua Ferreira -

FortifAI’s Nol8 joins the Megaport AI Exchange, unlocking the data plane layer for AI workloads

FortifAI (ASX: FTI), through its subsidiary Nol8, has entered a collaboration with Megaport (ASX: MP1), the global Network as a Service (NaaS) provider, to bring the Nol8 AI Data Plane to the Megaport AI Exchange.

Proof-of-concept deployments are already underway on live customer workloads, with general availability planned for later this year. The collaboration positions Nol8 within Megaport’s private software-defined network spanning more than 1,100 data centres worldwide.

For investors, the significance is clear. A blue-chip ASX-listed distribution partner gives Nol8 immediate access to a global customer base ahead of its commercial launch, even though commercialisation has not yet occurred.

Completing Megaport’s AI infrastructure stack

Over the past year, Megaport has assembled an AI infrastructure stack spanning Compute (Latitude.sh bare-metal and GPU services), Network (Megaport’s global private connectivity) and Storage, all available through the Megaport AI Exchange.

Through this collaboration, Nol8 adds the missing fourth layer, the AI Data Plane, now available through the Megaport AI Exchange for proof-of-concept deployments ahead of general availability. Under the arrangement, Nol8 “will pair its AI Data Plane with Megaport’s Compute, Network, and Storage layers.”

The table below maps how each layer fits together within the stack.

Layer Provider Function
Compute Latitude.sh Bare-metal and GPU services
Network Megaport Global private connectivity
Storage Unspecified Storage across the AI Exchange
AI Data Plane Nol8 (new addition) Wire-speed, governed data movement to GPUs

Michael Reid, CEO, Megaport

“AI is reshaping what our customers need from infrastructure. We’ve built the Compute, Network, and Storage layers they run on, and collaborating with Nol8 adds an important new layer, the AI Data Plane, that helps ensure GPUs are supplied with clean, governed data. We’re glad to support Nol8 as they bring this capability to our customers through proof-of-concept deployments on the Megaport AI Exchange.”

What is the AI Data Plane, and why does it matter?

Modern AI workloads place heavy demands on infrastructure, and one layer in particular has struggled to keep pace. The layer sitting between storage and the GPU currently relies on CPU arrays and middleware built for what the company describes as “human-pace data movement.”

That older approach cannot feed a GPU at the rate modern AI workloads demand. The result is idle, unbilled GPU capacity, which represents wasted and expensive compute resources.

Nol8 FPGA benchmark data published in April 2026 quantified that cost gap concretely: a single appliance costs under A$50,000 per year to operate, compared to A$6-7.5 million for the CPU array it replaces at equivalent throughput under P99 AI-grade load.

Nol8’s AI Data Plane is designed to replace that improvised layer with a single accelerated data plane. It feeds GPUs at wire speed, applies governance inline, and enforces policy before data ever reaches the model.

The investment relevance is direct. Recovering idle GPU capacity improves GPU economics, which is the core value Nol8 is selling to NeoClouds and the customers they serve.

The collaboration targets four initial use cases for joint proof-of-concept engagements:

  1. Agent-to-agent in agentic commerce: Mediating and governing high-frequency data exchange between autonomous buyers and sellers at machine speed, with policy enforced inline.

  2. Pre-inference: Retrieval, data preparation, classification and filtering of inbound data before it reaches the model, so the GPU receives clean, governed, model-ready data.

  3. Post-inference: Governance, redaction and routing of model outputs as they leave the GPU, enforcing compliance on results in motion.

  4. Inline data governance: Applying classification and policy to sensitive data as it moves, allowing data previously held back for compliance to be safely used.

Proof-of-concepts with select customers, and what is being measured

Nol8 is now listed on the AI Exchange service in the Megaport Marketplace. Here, Megaport’s global customer base can discover the AI Data Plane and begin proof-of-concept deployments alongside the compute, connectivity and storage they already consume.

The company is entering POC deployments with select customers to measure the real-world impact of the AI Data Plane on GPU performance and economics, before and after deployment.

The specific metrics being measured include:

  • Effective GPU utilisation

  • Throughput

  • Time-to-first-token

  • Revenue-per-GPU-hour

A key accuracy point for investors: per the annexure, the AI Data Plane is not commercially available at this stage. It is available only for proof-of-concept engagements, with general availability intended for later this year. No revenue or commercial sales should be inferred from the current deployments.

The investment case and what comes next

According to the annexure, the proof-of-concept will “provide clear evidence of the value of the Nol8 AI Data Plane, a key step toward commercial adoption.”

The collaboration builds on the company’s existing product roadmap, described as an important step “further to the announcement dated 16 February 2026.” It represents a progression of that roadmap rather than a standalone event.

The near-term path includes:

  • POC deployments now underway with select customers

  • Measurement of GPU performance and economic impact across the defined metrics

  • Planned general availability later this year

Kelly Herrell, CEO, Nol8

“Megaport connects the world’s infrastructure. Nol8 is building the data plane that will feed every GPU on it at wire speed, the layer the industry has been missing as AI workloads outpace the infrastructure built to serve them. The Megaport AI Exchange gives us the fastest path to bring that to customers, ahead of general availability later this year.”

Securing a global distribution channel through an ASX-listed NaaS leader gives FortifAI’s Nol8 a route to a worldwide customer base ahead of its commercial launch. With proof-of-concept deployments now live and general availability planned for later this year, the next phase will centre on the measured evidence of GPU performance gains, the metrics Nol8 has identified as central to its commercial proposition.

For investors tracking execution milestones, our full explainer on the Nol8 commercialisation roadmap covers the three-phase plan through end CY2026, including the performance benchmarks and capital position underpinning the pathway to first commercial contracts.

Don’t Miss the Next ASX Tech Breakout

Big News Blast delivers FREE breaking ASX tech news straight to your inbox within minutes of release, complete with in-depth analysis already done for you. Join 20,000+ subscribers who stay ahead of the market the moment announcements drop. Click the “Free Alerts” button at StockWire X to start receiving alerts today.


Frequently Asked Questions

What is the Nol8 AI Data Plane and what does it do?

The Nol8 AI Data Plane is an accelerated infrastructure layer that sits between storage and GPUs, feeding data at wire speed while applying governance and policy inline — replacing CPU arrays that cannot keep pace with modern AI workloads and leaving GPUs idle.

What is the Megaport AI Exchange and why does it matter for FortifAI?

The Megaport AI Exchange is a marketplace within Megaport's private software-defined network spanning more than 1,100 data centres globally, where customers can access Compute, Network, Storage, and now Nol8's AI Data Plane — giving FortifAI immediate distribution reach to a global enterprise customer base ahead of its commercial launch.

Is Nol8's AI Data Plane commercially available yet?

No — as of this announcement, the AI Data Plane is available only for proof-of-concept engagements with select customers on the Megaport AI Exchange, with general availability planned for later in 2026.

What metrics is Nol8 measuring during its Megaport proof-of-concept deployments?

Nol8 is measuring effective GPU utilisation, throughput, time-to-first-token, and revenue-per-GPU-hour to quantify the real-world impact of the AI Data Plane on GPU performance and economics before and after deployment.

How does Nol8's cost compare to the CPU infrastructure it replaces?

According to FPGA benchmark data published in April 2026, a single Nol8 appliance costs under A$50,000 per year to operate, compared to A$6–7.5 million for the CPU array it replaces at equivalent throughput under P99 AI-grade load.

Josua Ferreira
By Josua Ferreira
Partnership Director
Josua Ferreira holds a Bachelor of Commerce in Marketing and Advertising and brings a background in publication, business development, and ASX market storytelling. He has worked with listed companies across the resource sector and broader market, combining sharp commercial instincts with a genuine commitment to keeping investors informed.
Learn More
Companies Mentioned in Article

Breaking ASX Alerts Direct to Your Inbox

Join +20,000 subscribers receiving alerts.

Join thousands of investors who rely on StockWire X for timely, accurate market intelligence.

About the Publisher