CoreWeave Enters Asia with 360 MW Data Center Cluster in Indonesia
AI cloud service company CoreWeave announced its entry into the Asian market by building three new data centers in Indonesia, marking its first infrastructure projects in the Asia-Pacific region.
The three data centers will have a total IT power capacity of 360 MW, expected to be operational by 2028. CoreWeave stated it will invest billions of dollars but did not disclose the specific amount.
The project targets AI labs, startups, and enterprise clients in Southeast Asia, with CoreWeave planning to build and operate the computing environment while forming a local team in Indonesia.
Source: Public Information
ABAB AI Insight
CoreWeave's expansion in Indonesia is not merely a geographical replication but an extension of "AI computing power supply" from North America and Europe to Southeast Asia, indicating that global GPU demand is no longer solely driven by American cloud providers but is being shaped by localized low-latency and data localization needs. Similar cases can be compared to CoreWeave's $2.2 billion investment in three data centers in Europe in 2024, and its ongoing expansion around Federals, agentic AI, and hedging tools in 2025-2026: the company has transformed from a "computing power rental business" to a "global computing power real estate developer." In terms of capital pathways, a scale of 360 MW means CoreWeave is not selling software but securing power, land, and grid resources, packaging these infrastructures into AI computing capacity to sell to labs, startups, and large enterprises, representing a typical heavy asset expansion. Essentially, this is a reconstruction of the industrial chain: the core of AI competition is shifting from model capability to the geographical location and supply speed of computing power. The underlying mechanism is that after the explosion of training and inference demand, those who can secure power and data centers earlier will be able to convert GPUs into sellable computing power faster.
ABAB News · Cognitive Laws
The competition in AI cloud ultimately hinges on power and land.
Computing power is not code; it is a heavy asset that can be rented.
Models are on the surface; the real moat lies in the infrastructure.