Megaport's A$978.6M AI Infrastructure Deals Signal the Inference Boom
Megaport signed three AI infrastructure contracts worth A$978.6M and raised its 2027 forecast, as demand shifts from AI training to large-scale inference.
The numbers
Australian technology company Megaport announced on September 29, 2026 that it has signed three new AI infrastructure contracts with a combined value of approximately A$978.6 million (about US$686.8 million). The contracts are expected to contribute roughly A$232.4 million in annual recurring revenue once the associated infrastructure is fully deployed. The customers were not publicly identified.
The deals cover GPU and CPU computing capacity plus networking and storage infrastructure designed specifically for AI applications and inference workloads -- the capacity companies need to run trained models at scale, not just train them.
Megaport also raised its fiscal 2027 outlook: revenue guidance is now A$720-810 million (up from A$620-730 million), the expected EBITDA margin range rose to 42-44% (from 38-40%), and planned capital expenditure increased to A$1.78-1.88 billion to fund the buildout.
A pattern, not a one-off
These are not Megaport's first AI contracts. The company secured four similar deals worth about A$458.9 million in early June, bringing the value of strategic AI infrastructure contracts announced since April to approximately A$2.3 billion.
The shift reflects Megaport's expansion into compute following its acquisition of Latitude.sh, a high-performance computing infrastructure company. The traditional software-defined networking provider is becoming a full-stack AI infrastructure supplier -- connectivity, compute, and storage in one platform.
The bigger signal is where the demand is coming from. AI infrastructure spending is moving beyond model developers and hyperscalers into the underlying ecosystem: networking, storage, and regional providers in markets like Australia are now signing near-billion-dollar contracts.
What this means for your business
Inference is the new gold rush. The training boom got the headlines; the inference boom -- running AI for real users, every day -- is where the recurring revenue lives. Megaport's contracts are explicitly for inference workloads.
Action steps:
1. Price inference into your AI products now. If you sell anything AI-powered, your unit economics depend on inference cost. Model it per-request, not per-month. 2. Diversify beyond the hyperscalers. Regional and specialized infrastructure providers are winning serious contracts. Get quotes from more than the big three clouds -- competition is your leverage. 3. Watch customer prepayments. Megaport noted that customer prepayments help finance its buildout. If you are a heavy compute buyer, prepayment deals can lock in capacity and pricing. 4. For service businesses: resell the stack. Companies bundling connectivity + compute + storage for AI workloads are winning. If you serve mid-market clients, packaging managed AI infrastructure is a real revenue line.