Megaport secures four AI deals to build inference cloud

Megaport has secured four AI infrastructure contracts worth A$458.9m and plans to raise A$827.3m, or $594m, through an entitlement offer. The company plans to use the capital to build an inference cloud supported by Nvidia GPUs, network and storage infrastructure. The signal is that interconnection platforms are being pulled closer to compute as enterprise AI demand shifts towards latency-sensitive inference.

• US technology contracts total A$458.9m and start in early 2027
• The move shifts Megaport from connectivity into GPU-backed infrastructure

The fact

Australia's Megaport has secured four AI infrastructure contracts worth A$458.9m with US-based technology providers and plans to raise A$827.3m, or $594m, through an entitlement offer. The contracts are expected to start in the first half of 2027 and require about A$369.5m in capital expenditure, mainly for Nvidia GPUs, network and storage infrastructure. Megaport also plans an A$350m on-demand GPU pool for enterprise inference cloud services.

The Assessment

This is a clear step beyond Megaport's traditional network-as-a-service role. The company is trying to turn its data centre connectivity footprint into a platform for GPU access, distributed inference and consumption-based AI infrastructure. The market signal is specific: as enterprise AI demand moves towards latency-sensitive inference, interconnection platforms are being pulled closer to compute, not just carrying traffic between clouds.

What to Watch

Watch whether Megaport can deploy the GPU pool on schedule, start the 2027 contracts without capex overrun, and convert inference demand into recurring revenue rather than balance-sheet-heavy expansion.

Also read: ChatGPT reaches 1 billion monthly app users

Also read: Verne updates brand with AI infrastructure focus