GPU Merdeka by Lintasarta: a sovereign AI cloud built in under three months
Using the NVIDIA Cloud Partner program's reference architectures, Indosat subsidiary Lintasarta stood up GPU Merdeka — a sovereign, NVIDIA-accelerated AI cloud — in under three months.
Key takeaways
- GPU Merdeka by Lintasarta is an NVIDIA-accelerated sovereign AI cloud built in under three months.
- It runs at a renewable-energy-powered BDx Indonesia data center.
- The build used NVIDIA Cloud Partner program reference architectures and technical support.
- Priority sectors include energy, financial services, and healthcare.
What GPU Merdeka is
GPU Merdeka by Lintasarta is a sovereign AI cloud — meaning the compute, data, and operations stay inside Indonesian infrastructure and governance boundaries — built on NVIDIA-accelerated hardware. It runs at a BDx Indonesia AI data center powered by renewable energy, and it was brought online in under three months by drawing directly on the NVIDIA Cloud Partner program's reference architectures, technical support, and access to NVIDIA's AI software stack rather than building a bespoke GPU cloud from scratch.
'Merdeka' — Indonesian for 'independence' — signals the positioning directly: enterprise and public-sector customers get NVIDIA-class AI infrastructure without routing sensitive data through offshore cloud regions.
Why three months matters
Standing up a GPU cloud that meets NVIDIA Cloud Partner reference standards — driver stack, DCGM telemetry, GPU Operator, storage, and network design — normally takes considerably longer than three months when a provider is designing the architecture from zero. The speed here is a direct product of the February 2024 NVIDIA Cloud Partner MoU: Lintasarta wasn't inventing an architecture, it was implementing one NVIDIA had already validated with its partner ecosystem.
Target workloads
Lintasarta positions GPU Merdeka around three priority sectors: energy, financial services, and healthcare — industries with real AI workloads (seismic and engineering analysis, fraud and risk modeling, imaging and clinical document search) and, often, regulatory or data-residency requirements that make an offshore GPU cloud a harder sell internally. The renewable-powered data center also gives sustainability-conscious buyers a concrete answer on power sourcing, not just a compute SLA.