Megaport Ltd. plans to raise A$827.3 million, or US$594 million, to build an artificial-intelligence inference cloud and fund new infrastructure [1].

This investment marks a significant shift for the Australian data center and interconnection provider as it pivots to capture the growing demand for AI-specific computing power. By establishing an inference cloud, Megaport aims to provide the specialized environment necessary for running AI models at scale.

The funding initiative comes as the company secures four new AI infrastructure contracts valued at A$458.9 million [2]. These agreements underscore the accelerating need for localized AI capacity within the Australian market. To execute these specific contracts, Megaport requires a capital expenditure of A$369.5 million [3].

Megaport expects the work associated with these contracts to begin in the first half of 2027 [3]. The company intends to use the raised capital to bridge the gap between current capabilities and the requirements of these high-value deals, ensuring the infrastructure can support the computational load of modern AI applications.

The company's strategy focuses on the "inference" side of AI, which is the process of using a trained model to make predictions or generate content. This differs from "training," which requires different hardware and power profiles. By focusing on inference, Megaport is positioning itself as a critical layer in the AI delivery chain for businesses in the region [1].

The company said the announcement was made on Wednesday, June 3, 2026 [1].

Megaport plans to raise A$827.3 million to build an artificial-intelligence inference cloud.

Megaport's move signals a broader trend of data center operators transitioning from general-purpose connectivity to AI-optimized infrastructure. By focusing specifically on an inference cloud, the company is targeting the operational phase of AI deployment where businesses apply existing models to real-world data, suggesting a bet on the commercialization and scaling of AI services rather than just the initial development phase.