SharonAI Holdings' Earnings Calls Reveal Capacity Source and Deployment Timeline Contradictions

Date of Call: Aug 6, 2026

Guidance:

• Demand continues to materially outweigh supply, with 212 MW of secured AI factory capacity for deployment by end of 2027 and contracting visibility through 2031.
• Well-funded for near-term build-out following the $1.6 billion financing.
• Expect first material revenue to commence in Q4 2026 as large-scale B300 and GB300 deployments come online.
• Targeting more than 64,000 GPUs deployed by mid-2027 across Australia and New Zealand.

Business Commentary:

Capacity Expansion and Customer Demand:

• Sharon AI reported an increase in total secured AI factory capacity to 212 megawatts, an upgrade from the previous guidance of 132 megawatts, with 120 megawatts contracted through multi-year take-or-pay agreements.
• The company expects to deploy more than 64,000 NVIDIA GPUs by mid-2027 and has executed roughly $8.8 billion of total contract value year-to-date.
• The growth in secured capacity and contracted demand is driven by strong customer interest and the ability to expand supply, reflecting the strength of demand in the AI infrastructure market.

Strategic Partnerships and Financing:

• Sharon AI signed a six-year strategic compute collaboration with NVIDIA worth $4.9 billion in total contract value and completed a $1.6 billion oversubscribed financing round.
• The company raised approximately $2.2 billion of capital since December 2025.
• The strategic partnerships, such as with NVIDIA and Vast Data, provide access to advanced AI infrastructure and bolster market credibility, while successful financing rounds support the capital-intensive deployment model.

Customer Contracts and Pricing:

• Sharon AI signed significant contracts, including a five-year take-or-pay agreement with a global AI lab worth $1.32 billion and a five-year agreement with a global technology company worth $950 million.
• The company observed record pricing for GPU capacity, with a recent deal featuring a price of over $4 per GPU hour.
• The strong demand and constrained supply market allow Sharon AI to secure higher pricing, supported by strategic partnerships and long-term customer commitments.

Geographic Expansion and Sovereign Infrastructure:

• Sharon AI extended its footprint to New Zealand with its first facility and serves a global customer base despite being headquartered in Australia.
• The company's Australian and New Zealand-hosted sovereign infrastructure caters to regulated and sensitive customers, aligning with increasing data residency and sovereignty requirements.
• Geographic expansion and sovereign infrastructure positioning strengthen Sharon AI's market presence and appeal to a broad customer base, supporting its growth strategy.

Sentiment Analysis:

Overall Tone: Positive

• Management expresses strong validation of their scarcity framework, highlighting 'the contracted book has grown by roughly four times and our secured capacity has more than doubled since' and 'we are very confident in our ability to deliver.' They note 'oversubscribed' financing, record GPU prices, and a pipeline that leaves them 'exactly the strong position we wanted.'

Q&A:

• Question from Darren Evtahi (Lucid Capital Markets): The additional capacity, the 80 megawatts you guys added this morning in the release, is that source coming from a same partner you're working with or is it a new partner? Second question, on the NVIDIA partnership, the 72 megawatts, any updates on releasing that? And with that question on the release, are conversations with customers, I assume, in the ballpark of where your latest contract was north of $4 GPUR?
  Response: The 80 MW additional capacity is from a new partner. The NVIDIA capacity (72 MW) is expected to be unlocked with some early megawatts potentially as early as late 2026/early 2027, and delivery through 2027. Customer pricing discussions for NVIDIA capacity have reached record prices per GPU hour.

• Question from Brett Knobloch (Cancer Fitzgerald): What is your priority to resell the potential, the backstop capacity from NVIDIA or to sell the remaining capacity or the remaining 92 megawatts that you have? Is there a preference for what would come first or what would NVIDIA want first?
  Response: Priority is to resell the NVIDIA capacity first due to its global market recognition and the strong demand from AI natives, which also provides valuable customer insight for planning and selling the remaining 92 MW.

• Question from Brett Knobloch (Cancer Fitzgerald): On contract duration, if I look at all the contracts you've signed, maybe absent some of the really small ones, it's been five years except for NVIDIA at six. Is there a target duration you're looking for when you do ultimately get into the reselling the NVIDIA capacity?
  Response: Contract durations are largely driven by customer preference, with demand in the 3 to 5 year range. Management aims for a balanced book to manage the 40,000 GPU resource and expand relationships with existing customers.

• Question from Brett Knobloch (Cancer Fitzgerald): If I kind of do some back-of-the-map math here on the storage with your partnership with Vast, it's about 100,000 GPUs, which is about similar to how much megawatts you've now secured following the 80. At what point would you look to expand that just ahead of additional capacity ramp in the future?
  Response: The company is in ongoing discussions with VAST to expand storage capacity, with the recent 10 petabytes per 1,000 GPUs contract indicating a shift in customer demand for more storage, which influences facility design and capacity planning.

• Question from Ian Generous (Compass Point): Your partnerships now include NVIDIA, Dell, VAST, and a number of data center operators. Can you talk about how those relationships support the growth strategy from here, whether that's validating next-gen GPUs and what kind of line of sight they give you into future demand?
  Response: Partnerships lower execution risk, provide customer referrals, and offer insights into market trends (e.g., increased storage per GPU demand). They validate Sharon AI's position and support the partner-led ecosystem model.

• Question from Ian Generous (Compass Point): As those conversations extend into the next generation, how are you observing pricing dynamics on bare Rubens? Are customers engaging on Ruben commitments today for late 27, 28 deliveries? And how do you see pricing trending relative to GB300s?
  Response: There is extraordinary demand for Vera Rubin GPUs, with customers actively seeking commitments for late 2027/early 2028 deliveries. Pricing discussions have not yet begun, but demand is very strong.

• Question from Jonathan Higgins (Unified Capital Partners): How do you sort of strategically think about that capacity? You know, you've raised it today to obviously 200 or above. How should we think about that, you know, probably in the 2028 and what you're seeing on the demand side of things?
  Response: Capacity growth will be customer-driven and measured, with a focus on securing contracts and financing before announcing new megawatts. The company expects to increase forward period capacity announcements as customer contracts are secured and deployed.

• Question from Jonathan Higgins (Unified Capital Partners): Can you talk about that, you know, give us an idea of what the demand is or the shortages are out of, you know, ANZ and Asia versus say like what you're seeing in the US?
  Response: Demand is equally strong from US and Asia customers. The entire market is constrained, and Sharon AI is prioritizing customers with strong growth profiles to expand multi-year relationships, balancing the customer book.

• Question from Jonathan Higgins (Unified Capital Partners): I mean, the financing, as you say, you know, in your release, you're talking about you've got a lot more dry powder than you had at the start of the year. And within the video deal and the movements that we've seen in sort of financing and the like, can you talk about how you're sort of seeing the IRRs in the business?
  Response: The company is very thankful for strong shareholder support in raising capital. Debt facilities are advanced, with strong pricing and IRRs reflective of strong customer demand, supporting a favorable lending environment.

• Question from Fedora Chatelain (B Reilly Securities): On the NVIDIA partnership and the GB300 capacity under the management, what kind of customers are you targeting to fill that capacity? And can you frame how much of it you expect to be contracted, like take a paver? versus sold on demand?
  Response: Target customers are AI natives (model builders, inference providers). The vast majority of NVIDIA capacity is expected to be contracted on a take-or-pay basis for 3-5 year terms, with a small portion on spot. Deployment is planned over the first half of 2027.

• Question from Fedora Chatelain (B Reilly Securities): What's the biggest swing factor that could push that ramp into 4Q? For example, SDS, if I recall correctly, service start date is September 16th.
  Response: The primary risk factors are hardware delivery and data center readiness. Management believes revenue will materially ramp in Q4 2026, with the first B300 deployment already handed over to a customer this quarter.

• Question from Fedora Chatelain (B Reilly Securities): Is that a port revision being driven by SYNCON? contracts pulling capacity forward or by anticipated demand ahead of signed paper? And kind of what conditions make you confident to post a contracted megawatt target if it will happen?
  Response: The increased GPU target is driven by strong customer demand and confidence in data center delivery timelines. The company is comfortable with the contracted demand and financing in place, aiming to expand capacity in a measured way.

Contradiction Point 1

Source of the 80 MW Capacity Addition

Contradiction on whether capacity comes from new or existing partners, affecting transparency and trust in company operations.

Darren Evtahi (Lucid Capital Markets) - Darren Evtahi (Lucid Capital Markets)

2026Q2: The 80 megawatts of additional capacity comes from new partners. - [James Manning](CEO)

Is the 80 MW added recently from a new partner or existing? - Darren Aftahi (Lucid Capital Markets)

2026Q2: The new 80 MW comes from a new partner in Australia. - [James Manning](CEO)

Contradiction Point 2

Preferred Customer Strategy for Reselling NVIDIA Capacity

Shift from targeting large single customers to focusing on AI-native companies for NVIDIA capacity, indicating a strategic pivot in market approach.

Brett Knobloch (Cancer Fitzgerald) - Brett Knobloch (Cancer Fitzgerald)

2026Q2: The company's immediate priority is the resale of the NVIDIA (AICP) capacity, as it is already designed and online... driving strong customer demand. - [James Manning](CEO)

What is the priority between reselling NVIDIA's backstop capacity and selling the remaining 92 megawatts, and is there a preference for the order or NVIDIA's desired approach? - Alex Frohman (Lucid Capital Markets)

2026Q1: The 40 MW M2 block is likely for a large single customer (hyperscaler or AI-native enterprise), not broken into small chunks. - [James Manning](CEO)

Contradiction Point 3

Target Contract Duration with AI-Native Customers

Shift from longer-term contracts (5-year+) to a more flexible 3-5 year range for AICP capacity, reflecting a change in business model strategy.

Brett Knobloch (Cancer Fitzgerald) - Brett Knobloch (Cancer Fitzgerald)

2026Q2: Contract durations are customer-driven, typically in the 3-5 year range. - [James Manning](CEO)

Is there a target contract duration for reselling NVIDIA capacity, considering most contracts are five years except NVIDIA's six-year term? - Alex Frohman (Lucid Capital Markets)

2026Q1: Targets 5-year-plus contracts with options (e.g., ESDS has 5 years + 2 one-year options). - [James Manning](CEO)

Contradiction Point 4

Deployment Schedule for NVIDIA GB300 Capacity

Contradiction on the deployment timeline for the 40,000 GB300 GPUs, impacting revenue recognition and financial outlook clarity.

Fedora Chatelain (B. Reilly Securities) - Fedora Chatelain (B. Reilly Securities)

2026Q2: Deployment of this capacity is planned for the first half of 2027, with revenue recognition commencing in Q3 2026 for early B300 deployments and ramping materially in Q4 2026. - [James Manning](CEO)

What is the deployment schedule for the 40,000 GPUs? - Fedora Shabalin (B. Riley Securities)

2026Q2: A solid Q4 2026 is expected, with material revenue ramping in that quarter as hardware comes online. - [James Manning](CEO)

Contradiction Point 5

Deployment Timeline for NVIDIA AICP Program Capacity

Specific deployment window shifted from being aligned with revenue recognition to a more general first-half-2027 target, affecting financial forecasting precision.

Fedora Chatelain (B. Reilly Securities) - Fedora Chatelain (B. Reilly Securities)

2026Q2: Deployment of this capacity is planned for the first half of 2027, with revenue recognition commencing in Q3 2026 for early B300 deployments and ramping materially in Q4 2026. - [James Manning](CEO)

What is the deployment schedule for the 40,000 GPUs? - Brett Knoblauch (Cantor Fitzgerald)

2026Q1: Very confident in signing the 100 MW of additional capacity announced for early 2027. - [James Manning](CEO)