AIToday
AI Stocks & MarketsAI Business & IndustryYahoo Finance AIPublished: Aug 7, 2026, 22:03 JST5 min read

SharonAI secures 212 MW GPU capacity, $8.8B contract backlog

SharonAI secures 212 MW GPU capacity, $8.8B contract backlog

Key takeaway

  • SharonAI Holdings has tripled its secured GPU capacity to 212 megawatts and grown its contracted revenue backlog to $8.8 billion as it deploys NVIDIA graphics processors across Australia and New Zealand for AI workloads.

  • The company signed major agreements including a $4.9 billion six-year deal with NVIDIA and expects to deploy more than 64,000 GPUs by mid-2027, with first material revenue beginning in the fourth quarter of 2026.

3 Key Points

  1. What happened

    SharonAI Holdings expanded its secured AI factory capacity to 212 megawatts (up 80 MW from prior guidance of 132 MW), with 120 MW contracted under multiyear take-or-pay agreements and 92 MW available to sell. The company's contracted revenue book reached approximately $8.8 billion year to date, up from $2.2 billion three months earlier. Key deals include a $4.9 billion six-year strategic compute collaboration with NVIDIA involving 40,000 GB300 GPUs, plus agreements with a global AI lab ($1.32 billion), a global technology company ($950 million), and a global AI platform ($373 million).

  2. Why it matters

    SharonAI is building a major GPU infrastructure business across Australia and New Zealand to serve AI model builders and inference providers. The fourfold increase in contracted revenue and more-than-doubled capacity in three months suggests strong demand for GPU services as customers need more computing power for AI workloads. The company expects to deploy more than 64,000 NVIDIA GPUs by mid-2027, with first material revenue beginning in the fourth quarter of 2026 as large-scale deployments come online.

  3. What to watch

    SharonAI expects the vast majority of the 40,000 GPUs in the NVIDIA program to be deployed during the first half of 2027 under three- to five-year take-or-pay contracts. The company's revenue ramp will depend on hardware delivery from suppliers and data-center readiness, which management is actively tracking. SharonAI's contracting visibility extends through 2031, and the company has raised approximately $2.2 billion since December 2025.

In Depth

Read the full story

SharonAI Holdings announced in its Q2 earnings call that it has secured 212 megawatts of total AI factory capacity, a substantial increase from its prior guidance of 132 MW. Of this total, 120 MW is contracted under multiyear take-or-pay agreements, while 92 MW remains available to sell. The company is planning to deploy more than 64,000 NVIDIA GPUs by mid-2027 across Australia and New Zealand as part of its sovereign infrastructure strategy.

Chief Executive Officer and co-founder James Manning highlighted that the company has executed approximately $8.8 billion in total contract value year to date. Three months earlier, in late Q1, SharonAI had 100 MW of secured capacity and $2.2 billion in contract value, meaning secured capacity has more than doubled while the contracted book has increased roughly fourfold. The company announced several major customer agreements: a six-year strategic compute collaboration with NVIDIA valued at $4.9 billion minimum revenue involving an initial 72 MW deployment and 40,000 GB300 GPUs; a five-year take-or-pay agreement with a global AI lab valued at $1.32 billion; a five-year take-or-pay agreement with a global technology company valued at $950 million; and a five-year take-or-pay agreement with a global AI platform valued at $373 million. Manning noted that the most recent global AI platform agreement involves a B300 deployment priced at more than $4 per GPU hour, which he characterized as a record rate for the company. He also emphasized that demand for GPU capacity remains constrained, allowing SharonAI to pursue higher pricing.

The additional 80 MW of capacity announced comes from a new partner and is located in Australia, with some initial megawatts potentially available as early as late 2026, and further delivery expected through 2027. SharonAI began 2026 with 54 MW of secured capacity and has now established its first New Zealand facility to support sovereign infrastructure requirements for customers with data residency and regulatory needs. The company is currently working on B300 and GB300 cluster deployments and expects first material revenue to begin in the fourth quarter of 2026 as large-scale deployments come online.

On the capital front, SharonAI has raised approximately $2.2 billion since December 2025, including a $1.6 billion oversubscribed strategic financing round completed in June and a $350 million convertible note completed in April. The company is also advanced in discussions concerning debt facilities. SharonAI's capital strategy relies on partnerships with data-center operators rather than developing greenfield facilities, with NVIDIA as its primary compute supplier, NEXTDC as its primary data-center capacity provider, VAST Data as a storage partner (with an expanded commitment to 600 petabytes to support as many as 100,000 GPUs), and World Wide Technology as its exclusive APAC procurement, testing and implementation partner. Lastly, SharonAI announced leadership changes, including Anuj Goel, formerly of Macquarie Group, as chief financial officer; Melissa Anastasiou as chief legal officer; and Andrew Penn as non-executive chairman.

Context & Analysis

SharonAI's earnings call reveals a company scaling rapidly at a critical juncture for AI infrastructure deployment. The expansion from 100 MW of capacity and $2.2 billion in contracts just three months prior to 212 MW and $8.8 billion represents acceleration driven by large-scale enterprise demand for GPU compute. The NVIDIA partnership is especially significant: a $4.9 billion minimum revenue commitment with an anchor pricing model positions SharonAI to capture upside as GPU demand allows higher pricing—a dynamic the company is already experiencing, as evidenced by the $4 per GPU hour rate for its most recent AI platform agreement. CEO James Manning emphasized that demand for GPU capacity remains constrained, suggesting pricing power and customer desperation for compute resources.

The company's capital strategy reflects confidence in this market: SharonAI has raised $2.2 billion since December 2025 and is pursuing additional debt facilities, but it is not building greenfield data centers itself. Instead, it partners with operators like NEXTDC while securing GPU supply from NVIDIA and storage from VAST Data. This capital-light model allows SharonAI to scale without the balance-sheet drag of infrastructure construction. The deployment timeline is clear: initial megawatts from a new Australia partner could arrive as early as late 2026, with the majority of capacity online through 2027. Contracting visibility extends to 2031, indicating long-term customer lock-in through multiyear take-or-pay agreements that reduce execution risk.

FAQ

When will SharonAI start generating revenue from these GPU deployments?
SharonAI expects its first material revenue to begin in the fourth quarter of 2026 as large-scale deployments come online. The 40,000 GPUs included in the NVIDIA-backed program are expected to be deployed during the first half of 2027.
How much capital has SharonAI raised recently?
The company has raised approximately $2.2 billion since December 2025, including a $1.6 billion oversubscribed strategic financing round completed in June and a $350 million convertible note completed in April.
What is the most expensive GPU pricing rate SharonAI has achieved?
A B300 deployment for a global AI platform was priced at more than $4 per GPU hour, which Chief Executive Officer James Manning described as a record rate for the company.
Yahoo Finance AIRead Original Article

Get the latest AI Stocks & Markets news every morning

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · takes 30 seconds · unsubscribe anytime

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articleScientists create first AI-generated viruses using DNA model

The AI news that matters, in one minute each morning.

Sign up free