Sharon AI (SHAZ) Q2 2026 earnings review
A $1.9 Billion War Chest Meets an $8.8 Billion Pipeline
Sharon AI’s Q2 is a story of total structural transformation. Current revenue of $1.9M is essentially irrelevant. Backed by an oversubscribed $1.6B private placement and a $350M convertible note offering, the company has built a massive $1.86B cash reserve to fund an $8.8B Total Contract Value (TCV) pipeline. The headline GAAP net loss of $430M looks terrifying but is a red herring—driven almost entirely by a $400M non-cash accounting penalty on convertible notes because the stock price went up. Operationally, the company turned Adjusted EBITDA positive ($0.6M) and is aggressively deploying capital to bring 64,000+ GPUs online. The narrative has officially shifted from fundraising to extreme execution.
🐂 Bull Case
The company secured $8.8B in legally committed, take-or-pay contracts, anchored by a $4.9B 6-year NVIDIA collaboration and multi-billion-dollar deals with global AI labs. Demand is completely de-risked.
With $1.86B in cash on the balance sheet, Sharon AI has eliminated near-term capital constraints, allowing it to move immediately into massive infrastructure procurement for its 212MW capacity.
🐻 Bear Case
Transitioning from $1.9M in quarterly revenue to servicing an $8.8B pipeline requires flawless supply chain management, data center construction, and grid connection. Any delays will directly hit future revenues.
The capital to build this infrastructure came at a cost. Class A shares outstanding jumped from 11.8M to 35.6M in six months, alongside over $1B in convertible note liabilities that could trigger further dilution.
⚖️ Verdict: 🟢
Bullish. The market pays for infrastructure scale and de-risked AI demand. By securing the cash and signing the multi-billion-dollar take-or-pay contracts, Sharon AI has crossed the chasm from concept to platform. Execution is the only remaining hurdle.
Key Themes
Unprecedented Take-or-Pay Pipeline
The sheer scale of Q2 customer momentum redefines the company. Sharon AI locked in a $4.9B collaboration with NVIDIA (40,000 GB300 GPUs), a $950M contract with a global tech company, a $1.32B anchor contract for New Zealand, and a $373M platform deal. Because these are 'take-or-pay', revenue is highly predictable once the hardware is powered on.
Customer Deposits Funding Operations
Despite being in an aggressive build-out phase, operating cash flow for the first six months was actually positive $23.2M. This was driven completely by $143.9M in upfront customer deposits, proving that customers are willing to finance Sharon AI's capex in exchange for guaranteed future compute capacity.
The Optical Distortion of GAAP Earnings
GAAP reporting rules heavily penalized Sharon AI for its own stock price success. The $430M net loss includes a $400.4M fair value loss on convertible notes. As the stock price rises, the value of the conversion feature increases, forcing the company to record a non-cash expense. Investors must look entirely past net income to Adjusted EBITDA to understand core profitability.
Supply Chain and Deployment Logistics
The balance sheet shows $302.6M in equipment prepayments and $310.6M in advance payments for property and equipment in the cash flow statement. Procuring 64,000+ GPUs is only step one; securing the power, cooling infrastructure, and networking to deploy 212MW by mid-2027 carries immense macro supply chain risk.
Divestiture Completed Successfully
The company accelerated the receipt of $74M in proceeds from the divestment of Texas Critical Data Centers (TCDC). This effectively cleaned up the company's focus, making it a pure-play Asia-Pacific Neocloud provider rather than holding fractured global assets.
Other KPIs
Reversing. Flipped to profitability from a loss of $1.7M in Q2 2025. This metric completely strips out the $400.4M convertible note noise, $2.7M in share-based compensation, and $1.9M in depreciation, providing a much clearer look at baseline operations as the company prepares to scale.
Accelerating. This is the single most important number in the report. It represents legally binding, committed spend over the lifetime of the contracts (up to 6 years), establishing massive revenue visibility through 2032.
Guidance
Accelerating. Management explicitly guided that revenue will step up materially starting in Q3 2026 and continue ramping through 2027 as the pre-paid GPUs are installed and handed over to customers.
Accelerating. The company has upsized its capacity targets to 212MW (adding 80MW this quarter) and provided a hard target for hardware deployment. Achieving this will require historic execution speed over the next 12 months.
Key Questions
Gross Margin Profile at Scale
With $8.8B in TCV secured, what is the expected gross margin profile of these take-or-pay contracts once the initial hardware depreciation is accounted for?
Supply Chain Vulnerabilities
How insulated is the mid-2027 deployment timeline of 64,000 GPUs from potential NVIDIA supply chain bottlenecks, particularly for the newer B300 and GB300 architectures?
Power and Grid Certainty
Of the 212MW of total stated capacity, exactly how many megawatts are currently fully permitted, constructed, and grid-connected today?