Test report DSG-5371 · Rev F · tested October 10, 2026
AI Datacenter InfrastructureDevice under test
Nvidia and Firmus to build 170,000-GPU data center in Indonesia
Nvidia and Firmus plan a 170,000-GPU data center in Indonesia, placing one of the largest GPU clusters worldwide in Southeast Asia, Quartz reports.
- Read
- 2 min
- Words
- 375
- Node
- 20nm
- Operator
- Amara Osei
Spec summary
- Nvidia and Firmus plan a data center in Indonesia with 170,000 GPUs.
- The project was first reported by Quartz.
- Timeline, GPU models, budget and site location remain undisclosed.
- The facility would rank among the largest single-site GPU installations worldwide.
Nvidia and Firmus plan to build a data center in Indonesia equipped with 170,000 GPUs, according to a report by Quartz.
The project places Indonesia on the map of large-scale AI infrastructure deployments. At 170,000 GPUs, the facility would rank among the biggest single-site GPU installations announced worldwide, putting compute capacity on par with the training clusters operated by the largest hyperscalers.
What do we know about the deal?
Quartz reports that Nvidia will work with Firmus, a data center operator, to deliver the Indonesian facility. The reported figure of 170,000 GPUs defines the scale of the build.
Beyond the headline specifications, the report offers few operational details. Key open questions include:
- The project's timeline and construction phases
- Which GPU models Nvidia will supply
- Total capital expenditure and financing structure
- The site location within Indonesia and available power capacity
- Target customers — domestic enterprises, regional cloud providers, or AI training workloads
Why does Indonesia matter for AI infrastructure?
Southeast Asia has attracted rising interest from compute providers. Land, power and regulatory conditions in the region can support facilities that would face constraints in more saturated markets.
A 170,000-GPU installation carries substantial power and cooling demands. Industry benchmarks for high-density AI clusters put per-GPU power draw, including overheads, in the range that would push a facility of this size toward hundreds of megawatts of capacity — an infrastructure commitment that requires dedicated generation and grid agreements.
For Nvidia, the deal extends a pattern of partnering with regional operators to deploy GPU capacity outside the traditional US and European data center markets.
What does this mean for the market?
The announcement signals continued expansion of large-scale AI compute into emerging markets. Projects of this scale affect:
- GPU supply allocation, with 170,000 units representing a significant share of Nvidia's production output
- Regional cloud and AI services, giving Indonesian and Southeast Asian customers local access to frontier compute
- Data center construction, from power procurement to high-density cooling design
Quartz's report did not specify when construction begins or when the facility reaches full capacity. Firmus and Nvidia have not publicly detailed the contract value.
Die Signal will track the project as operators release technical and financial specifics.
via Google News: GPU datacenter (Source)
More from Amara Osei
Same lot · LOT-C1C6
- DSG-628820nmFirmus to deploy 170,000-GPU cluster in Batam, Indonesia
- DSG-582820nmMeta Leases AI Compute Capacity at Firmus' Southeast Asia Data Centers
- DSG-95397nmNVIDIA and SB Energy Plan 10-GW AI Data Center Complex in Ohio
- DSG-462814nmAsian banks scale GPU financing for AI data centres
- DSG-977414nmNvidia backs Cloverleaf, a data-center-powered land company