Executive Summary
Sharon AI, an Australian neocloud provider, has signed a $950 million, five−year cloud computing infrastructure agreement with an unnamed global technology company. The capacity will be deployed across multiple Next DC data centers in Australia, utilizing Vast Data′s AI Operating System.
The Players
Sharon AI is a small but rapidly growing Australian neocloud. As of September 2025, it had only 432 GPUs and 195 CPUs online. By January 2026, it announced a “supercluster” of 1,000 Nvidia B200s at a NextDC Melbourne facility. The company’s most recent quarterly filing (ending March 31, 2026) showed GPU infrastructure revenue of just $294,012, with zero crypto mining income—a complete pivot from its previous hybrid model.
The customer is described only as a “global technology company with major Asia-Pacific presence.” This could be a hyperscaler (AWS, Google, Microsoft, Alibaba), a large enterprise (e.g., Samsung, Toyota), or an AI-native firm. The anonymity suggests competitive sensitivity or a yet-to-be-announced strategic partnership.
NextDC is the data center partner, with 18 facilities across Asia-Pacific including major Australian cities (Sydney, Melbourne, Brisbane, Perth) and secondary markets (Port Hedland, Sunshine Coast, Newman). NextDC’s CCO explicitly welcomed the partnership, indicating this is a strategic relationship, not just a colocation rental.
So What?
This deal validates three critical trends, but from the perspective of a tiny neocloud punching far above its weight class:
First, Australia is becoming a tier-one AI infrastructure market. Most AI power and data center coverage focuses on the US (Ohio, Texas, Wyoming, Oregon) and Europe. But Sharon AI’s back-to-back billion-dollar deals suggest Asia-Pacific demand is equally intense, driven by US hyperscalers expanding regionally, Chinese firms seeking neutral ground, and Australian enterprises modernizing.
Second, capacity is now the product, not the GPU. Sharon AI owns contracts with NextDC and relationships with customers. This is a capacity aggregator model—lease data center space, fill it with GPUs, sell the compute as a service. The barrier to entry is not technology; it is speed to secure data center power and space.
Third, the neocloud model is defying the “scale or die” narrative. Conventional wisdom says AI cloud requires billions in capital, thousands of GPUs, and proprietary infrastructure. Sharon AI had 432 GPUs in September 2025 and is now signing $2.2B in contracts. The customer does not care about asset ownership—they care about getting B200 capacity online in Melbourne by Q3 2026. Sharon AI is essentially a financed reseller of NextDC power and floor space.
What Should You Do With This?
If you are a data center operator (like NextDC): your idle capacity is an asset class. Sharon AI is monetizing your power and space at a markup of 2-5x. Consider launching your own neocloud division to capture that margin directly, rather than leasing wholesale.
If you are a hyperscaler or large enterprise: this deal is a signal that secondary providers can deliver. You do not need to build your own Australian region or wait for a Google/Microsoft/AWS launch. Firms like Sharon AI, ESDS, and others can deploy capacity in existing colocation facilities faster than your internal real estate team can sign a lease.
If you are a competitor (other neoclouds): Sharon AI just proved that contracts come before infrastructure. They signed $2.2B in deals with virtually no operating history. Your path to scale is not building first—it is selling first, then financing the GPUs and colocation with customer prepays or IPO proceeds.
Source: Data Center Dynamics
