Zankore Secures Senior Term Loan of up to $3.1 Bn for Nvidia GPU Build-Out
The five-week-old Indonesian AI infrastructure platform will use the debt to acquire and deploy Nvidia GPUs, initially in Indonesia before expanding across ASEAN.
Magnific
Opinions expressed by Entrepreneur contributors are their own.
You're reading Entrepreneur Asia Pacific, an international franchise of Entrepreneur Media.
Zankore, an Indonesian AI-infrastructure platform launched barely five weeks earlier, has signed an agreement for a senior term loan of up to $3.1 Bn to finance the acquisition and deployment of Nvidia GPU infrastructure, initially in Indonesia, before expanding across ASEAN.
The figure indicates the maximum loan available under the borrowing arrangement, not an amount Zankore has necessarily drawn. The public disclosures do not specify the interest rate, maturity, drawdown schedule or collateral supporting the loan.
Citi acted as exclusive debt advisor, while Citi, ING, Natixis CIB, Qatar National Bank Group and United Overseas Bank (UOB) were senior mandated lead arrangers, underwriters and bookrunners, according to Zankore’s September 9 announcement.
The Structure Behind the Number
The headline number is striking for a newly launched platform. Yet the more consequential part of the transaction is the structure that persuaded banks to finance expensive computing hardware whose value depends heavily on utilisation and can be eroded by rapid advances in chip technology.
Nvidia has attached a revenue-sharing and credit-support model to the project. Zankore said the arrangement would align infrastructure deployment with customer demand, giving the company what it described as a capital-efficient route to scale.
Zankore chairman Vikram Sinha told Bloomberg that Nvidia was helping to make the project financeable. He said Nvidia’s service-order structure would let computing capacity reach smaller enterprises buying it on demand, instead of limiting access to investment-grade customers.
The arrangement appears designed to give lenders greater confidence that the GPUs will generate revenue after deployment. But neither Zankore nor the lenders disclosed how much credit risk Nvidia has assumed, how the revenue-sharing mechanism will operate or whether its support covers the entire loan.
Five Weeks from Launch to Loan
The platform was launched on August 6 by Indonesian telecoms operator Indosat Ooredoo Hutchison alongside Qatar’s Ooredoo Group, Nvidia and Nokia. Ooredoo holds a 49% stake and has committed about $800 Mn towards the launch and initial development, while Indosat provides local infrastructure and operating capabilities and Nokia supplies networking technology.
Zankore said it was building an initial 100 MW of Nvidia AI infrastructure in Indonesia. At its launch, the company targeted about 200 MW of capacity in the first half of 2027 and a longer-term total of 1 GW. But these are essentially development targets, not operating capacity.
The financing, therefore, gives Zankore access to debt for the hardware build-out, but it does not establish that the capacity has been installed or that customers have contracted to use all of it.
The Technology in Focus
The debt will finance more than standalone GPU hardware. Zankore plans to operate the GPUs as high-performance cloud clusters that customers can access at scale and on demand. Built on Nvidia’s full-stack AI computing technologies, these clusters are intended to support model training and fine-tuning, large-scale inference and agentic-AI applications. Zankore describes them as dedicated and production-ready.
The approximately 200 MW planned for the first half of 2027 is expected to use Nvidia GB300 NVL72 systems. Zankore also aims to adopt Nvidia DSX, a reference architecture that brings accelerated computing, networking, power, cooling and operations together as one system. Its planned DSX MaxLPS technology dynamically redistributes unused power across the GPU fleet. Nvidia says this can provide up to 40% more computing capacity within the same power budget.
Nokia is providing the AI-native networking intended to connect the infrastructure securely and at high speed, according to the launch announcement.
However, the disclosures do not state how many GPU systems Zankore will acquire, which data centres will house them or how the loan will be divided between deployment phases. Capacity expressed in megawatts denotes the power available to the infrastructure, not a direct measure of GPU numbers or computing output.
A Financing Model under Scrutiny
The financing model is not without scrutiny. In late August, the Wall Street Journal reported that Nvidia had paused some credit-support and revenue-sharing transactions amid internal concerns over possible antitrust scrutiny and partner pushback over the customer restrictions it imposed. Nvidia said the programme remained in place and was continuing to evolve.
Zankore’s financing, announced less than two weeks later, shows that the structure is still being used for at least one large deployment. It does not establish whether Zankore’s terms are identical to those attached to the transactions Nvidia reportedly paused.
Zankore, an Indonesian AI-infrastructure platform launched barely five weeks earlier, has signed an agreement for a senior term loan of up to $3.1 Bn to finance the acquisition and deployment of Nvidia GPU infrastructure, initially in Indonesia, before expanding across ASEAN.
The figure indicates the maximum loan available under the borrowing arrangement, not an amount Zankore has necessarily drawn. The public disclosures do not specify the interest rate, maturity, drawdown schedule or collateral supporting the loan.
Citi acted as exclusive debt advisor, while Citi, ING, Natixis CIB, Qatar National Bank Group and United Overseas Bank (UOB) were senior mandated lead arrangers, underwriters and bookrunners, according to Zankore’s September 9 announcement.
