Sharon AI, the Nasdaq-listed Australian neocloud provider, has signed a landmark five-year $1.25 billion infrastructure agreement with India’s ESDS Software Solution, committing to deploy a massive 8,000 Nvidia B300 GPU cluster within Australian data centres to power enterprise AI workloads amid acute global compute shortages. The deal, which includes an optional two-year extension, catapults Sharon AI’s capacity ambitions and validates its sovereign cloud strategy just months after a high-profile $125 million IPO.
Revenue recognition begins in Q3 2026 as the B300 deployment—successor to Nvidia’s blockbuster H100 and B200 architectures—comes online across Sharon AI’s leased facilities at Equinix SYD3/SYD5 in Sydney and NextDC M3 in Melbourne. James Manning, co-founder and CEO, positioned the agreement as a cornerstone achievement: “This contract represents one of many we’ve nurtured through our pipeline. With confirmed additional data centre capacity, we’re primed to accelerate delivery across enterprise, hyperscale, research, government and AI-native sectors.”
Anatomy of a Sovereign AI Powerhouse
Neoclouds like Sharon AI occupy a critical niche between traditional hyperscalers (AWS, Azure, Google Cloud) and specialised GPU brokers. Rather than owning physical infrastructure outright, Sharon AI leases premium colocation capacity and overlays proprietary orchestration, networking and security fabrics optimised for AI workloads. This asset-light model enables rapid scaling while maintaining sovereign data residency—a regulatory must for government, defence and financial services.
The ESDS deal underscores why sovereign AI matters in APAC. Australia’s National AI Plan commits $1 billion+ to local compute, but Nvidia’s Blackwell ramp remains constrained. Indian enterprises like ESDS—serving PaaS, SaaS, IaaS and managed services globally—require low-latency Australian capacity for compliance with data localisation mandates while avoiding US hyperscaler egress costs.
From Startup to Supercluster Scale
Sharon AI’s trajectory reflects AI infrastructure’s warp-speed economics. September 2025: 432 GPUs across footprint. January 2026: 1,000 B200 supercluster announced at NextDC Melbourne. February: $125M Nasdaq debut under ticker SHAZ. March: Lenovo TruScale’s largest-ever IaaS engagement; Cisco Secure AI Factory with 1,024 Blackwell Ultras.
B300s represent next-gen Blackwell: 208 billion transistors, 141GB HBM3e memory, FP4 precision for trillion-parameter models. Eight thousand units deliver raw compute rivaling mid-sized hyperscalers, targeting training, inference, RAG pipelines and agentic systems where latency kills ROI.
ESDS Partnership Bridges India-Australia Corridor
ESDS Software Solution—India’s cloud infrastructure stalwart—gains strategic APAC beachhead. Serving blue-chip multinationals, ESDS now provisions sovereign Australian AI capacity without building from scratch. The partnership exemplifies deepening Indo-Australian tech ties post-Quad frameworks, with data flowing Mumbai-Sydney-Melbourne.
Manning highlighted regional momentum: “Australia emerges as global AI contender through sovereignty-first innovation. We’re bridging enterprise demand with available capacity.” Post-announcement, SHAZ shares surged 15%, valuing firm at ~$1.2B despite revenue ramp commencing Q3.
Regional AI Infrastructure Context
Australia’s data centre capacity reached 847MW in 2025, with Sharon AI’s 8,000-GPU deployment adding significant compute power. India launched its National AI Mission with 10,000 GPUs across three clusters, while Singapore’s sovereign funds invested S$5.6 billion in AI infrastructure.
Power consumption concerns persist—data centres used 2.6% of national electricity in 2025. The Australian government identified a shortage of 3,000 AI specialists. Data localisation regulations continue evolving across APAC markets.
