Sharon AI Holdings is rapidly scaling its AI infrastructure platform, with total contracted value reaching approximately $8.8 billion year-to-date as the Australian neocloud operator targets deployment of more than 64,000 NVIDIA GPUs by mid-2027.
The company has secured 212 MW of data center capacity across Australia and New Zealand, with 120 MW already contracted. Sharon AI said its deployment pipeline continues to grow as demand for high-density AI computing capacity exceeds available supply.
The contracted portfolio includes a six-year strategic compute collaboration with NVIDIA carrying approximately $4.9 billion of total contract value. Sharon AI has also signed a five-year take-or-pay agreement valued at $1.32 billion with a global AI lab, a five-year take-or-pay agreement worth $950 million with a global technology company, and a five-year $373 million agreement with a global AI platform involving 2,048 NVIDIA B300 GPUs.
The NVIDIA collaboration provides access to as many as 40,000 GB300 GPUs and represents a major anchor for Sharon AI’s infrastructure expansion. The company said the six-year master services agreement helps de-risk the hardware buildout while allowing it to sell additional reserved capacity to third-party customers.
Sharon AI’s contracting model is structured around take-or-pay arrangements under which customers pay monthly for reserved capacity once systems are deployed, regardless of whether the customer uses all of its available computing capacity. Customer prepayments can also help fund procurement of GPUs and networking equipment.
The company believes this model can provide additional value after the initial contract period. Sharon AI estimates that GPUs may have at least two years of useful economic life remaining after a typical contract expires, creating opportunities to re-contract the hardware, sell computing capacity through the on-demand market or redeploy the infrastructure for inference workloads.
Its infrastructure footprint has expanded quickly. Secured AI Factory capacity increased from 54 MW in February to 212 MW by August, while contracted capacity climbed to 120 MW. Sharon AI said contracting visibility now extends through 2031.
Storage infrastructure is also scaling alongside GPU deployments. Sharon AI expanded its partnership with VAST Data to a 600 PB commitment, which the company says is sized to support approximately 100,000 GPUs.
Funding for the buildout has accelerated as well. Sharon AI has secured approximately $2.25 billion of capital since December 2025, including a $1.6 billion strategic financing completed in June, $350 million of convertible notes in April and $125 million from its Nasdaq IPO earlier in 2026.
Although the contracted pipeline is substantial, the financial results still reflect an early-stage infrastructure ramp. Q2 revenue was $1.93 million compared with $380,000 a year earlier, while adjusted EBITDA improved to positive $591,000 from a $1.68 million loss. The company recorded a large GAAP net loss that included approximately $400.4 million related to changes in the fair value of convertible notes.
Sharon AI expects revenue to begin ramping materially from the third quarter of 2026 through 2027 as larger deployments enter service.
