Sharon AI Books $373M Deal and Discloses $8.8B in Contracts as AI Factory Hits 91%
Sharon AI Holdings (NASDAQ: SHAZ) said 91% of its 132 MW AI Factory was sold out as it disclosed $8.8B total contracted revenue and a platform scaled to 212 MW. It filed an 8-K for a five-year, $373M cloud computing deal starting Q1 2027 using 2,048 NVIDIA Blackwell Ultra B300 GPUs. Q2 revenue was $1.9M and net loss $430.4M.
How this was made

The 30-second read
Why it matters
The article combines an 8-K contract disclosure with Q2 earnings, giving a rare same-day view of contracted backlog ($8.8B), capacity sold-out (91% of 132 MW), and a rapid platform expansion to 212 MW, plus GPU upgrade plans to 64,000 units by mid-2027.
Market read
Traders get a concrete, time-stamped update on backlog, utilization, and revenue start dates, which can drive repricing of AI infrastructure risk and growth expectations.
What to watch
The customer is unnamed and the utilization math is sensitive to ramp execution, GPU supply constraints, and whether take-or-pay terms fully offset demand variability.
Background
Sharon AI is an Australian-incorporated GPU-only AI infrastructure provider (neocloud) operating in NEXTDC Tier IV data centers, positioning around data sovereignty and high-density inference compute.
Ticker impact
Sharon AI disclosed a five-year $373M cloud computing agreement and said its AI Factory is 91% sold out, with total contracted revenue of $8.8B.
Likely positive near-term bias as traders re-rate contracted backlog and capacity ramp, with volatility around utilization and cash burn.
The article provides fresh primary disclosures (8-K plus Q2 results) including $373M deal terms, contracted revenue $8.8B, capacity expansion to 212 MW, and revenue start timing in 2027. However, it also highlights large net losses and that most contracts have not yet begun generating revenue.
Market effects
Supports the narrative that sovereign, GPU-only neocloud capacity is converting into large multi-year enterprise contracts, potentially tightening supply expectations for inference GPUs.
Highlights Australia-based compute sovereignty demand and the role of Tier IV data centers (NEXTDC) in enabling higher-density Blackwell deployments.
Signals accelerating enterprise AI infrastructure contracting outside hyperscalers, which may influence how investors price GPU availability and inference capacity globally.
Counterpoint
Contracted revenue and capacity sold-out do not guarantee cash generation; with most revenue starting in 2027, near-term fundamentals may still deteriorate if costs rise faster than contract monetization.
Key entities
- companySharon AI Holdings
NASDAQ-listed neocloud provider disclosing a $373M five-year cloud computing agreement and $8.8B total contracted revenue alongside Q2 results.
- technologyNVIDIA Blackwell Ultra B300
Inference-optimized GPU referenced as the hardware for the initial 2,048-GPU deployment under the $373M deal.
- infrastructure_providerNEXTDC
Tier IV data center operator hosting Sharon AI’s infrastructure, enabling B300 cluster density and liquid cooling requirements.

