Zankore, a startup supported by Nvidia and telecom giant Indosat Ooredoo Hutchison, has obtained a $3.1 billion loan to acquire and deploy Nvidia GPUs in Indonesia and the wider Southeast Asian region, aiming to rapidly expand AI compute capacity.
- Zankore raises $3.1bn loan to buy Nvidia GPUs for Southeast Asia
- Loan structured with revenue-sharing tied to demand and credit support
- Indonesia’s telecom infrastructure underpins rapid AI capacity buildout
What happened
Zankore launched in August 2026 and shortly thereafter secured a senior term loan facility worth up to $3.1 billion. This financing will enable the company to purchase and deploy Nvidia GPUs throughout Indonesia and the broader Southeast Asian region. The loan was arranged exclusively by Citi, in partnership with ING, Natixis CIB, Qatar National Bank, and UOB acting as lead arrangers, underwriters, and bookrunners.
The financing deal not only provides upfront capital but also includes innovative arrangements linking repayments to actual revenue generated from GPU capacity usage. Nvidia, one of Zankore’s investors and hardware suppliers, plays a central role in this ecosystem, supplying the GPUs that Zankore will deploy for AI workloads such as training and inference.
Why it matters
This transaction highlights a novel approach to funding AI infrastructure, where hardware providers like Nvidia collaborate with financial institutions and local telecom operators to accelerate deployment. The involvement of Indosat Ooredoo Hutchison, one of Indonesia's largest telecom operators, brings substantial operational and regulatory support necessary for building extensive infrastructure.
Furthermore, the model spreads financial risk across Nvidia, banks, and operators, closely linking investment returns to actual market demand for AI compute power. As Southeast Asia’s AI ecosystem grows, Indonesia emerges as a critical testbed for long-term AI data center capacity expansion, leveraging existing telecom assets to scale rapidly.
What to watch next
Zankore plans to commence operations with 100MW of Nvidia GPU capacity in Indonesia, targeting 200MW by mid-2027 and aiming for 1GW within three years. Observers should monitor how customer contracts and revenue streams evolve, as the facility's financial model depends heavily on demand fulfillment.
Another key aspect to watch is the collateral structure backing the loan, especially whether GPUs themselves act as security and how fluctuations in GPU market value might impact lender risk. Additionally, the broader Southeast Asian market’s response, including projects like Firmus with similar Nvidia-backed ambitions, will provide insights into the viability of these emerging AI infrastructure financing models.