Megaport lands A$458.9 million AI deals, raises funds for inference cloud

Megaport has landed four AI infrastructure contracts worth A$458.9 million and is raising A$827.3 million, or about $594 million, to build out an inference cloud aimed at enterprise AI demand.

The Brisbane-based network and cloud provider said the agreements are with U.S.-based technology companies running AI applications and are due to begin in the first half of 2027. Megaport said the contracts will require nearly A$369.5 million in capital expenditure, mostly for high-performance Nvidia GPUs, network systems and storage infrastructure.

The numbers matter because they offer Megaport a clearer path into one of the fastest-growing corners of AI infrastructure. The company said the new platform will be supported by an on-demand GPU pool backed by A$350 million in investment, with services sold to enterprise customers through both contracted and consumption-based pricing. That structure suggests Megaport is trying to build recurring infrastructure revenue around demand for inference, the process of running trained models in real time for live applications, rather than chasing the more cyclical training market.

Chief executive Michael Reid said AI inference is “one of the biggest infrastructure opportunities of the next decade.” The company’s pitch is that the next phase of the AI arms race will be decided less by model builders than by the network, storage and compute layers that keep applications responsive at scale.

The raise also indicates how capital-intensive that race has become. Megaport’s planned A$827.3 million fundraising is larger than the A$369.5 million of capital expenditure tied directly to the four contracts, giving the company room to finance a broader build-out rather than a single customer project. The contracts themselves represent a significant anchor for the strategy, covering more than half of the fresh capital being sought, even before any additional demand flows through the new platform.

Megaport has been laying the groundwork for this move for months. In November 2025, the company agreed to acquire Latitude.sh, a Compute-as-a-Service provider with high-performance CPU and GPU infrastructure in 20 key markets across 10 countries. That deal gave Megaport a wider geographic footprint and a deeper compute stack just as enterprise buyers began looking for distributed inference capacity.

Megaport’s website now describes its AI offering as bare-metal GPUs for AI training and inference with private connectivity to a global ecosystem of GPU providers and AI models. Taken together, the contracts, the capital raise and the Latitude.sh deal show a company moving beyond connectivity into the plumbing of AI deployment, where access to chips, networks and storage is becoming as strategic as the models themselves.