CoreWeave’s 360MW Indonesia Play Isn’t Just Expansion – It’s a Direct Shot at Big Cloud’s APAC AI Moat

Posted on by

(SeaPRwire) –

By: Ethan Gallagher

Let’s stop framing CoreWeave’s Indonesia announcement as just a regional expansion play. The 6.93% stock jump to $91.71 isn’t a reward for filling a geographic gap. It’s the market waking up to the fact that specialized AI cloud players are no longer niche side actors. Big Three cloud providers have long treated APAC AI compute as an afterthought. They overcharge for subpar density and force workloads to route through Singapore or Tokyo. That’s the unaddressed pain point CoreWeave just took a sledgehammer to. I’ve sat in meetings with Southeast Asian AI startup founders who say they pay 40% more per GPU hour than their U.S. peers. No amount of marketing fluff from AWS or Google Cloud fixes that gap.

The official release lays out clear, verifiable details. Three new data centers in Indonesia will add 360MW of contracted AI compute power. These are CoreWeave’s first physical data centers in the Asia-Pacific region, set to come online in 2028. As of March 31, 2026, CoreWeave ran 49 data centers across the U.S. and Europe, with 1GW active power and 3.5GW total contracted capacity. The company says the sites will serve local research labs, startups, enterprises and public institutions. It also plans to hire and train a local Indonesian operations team before launch. The project aligns with Indonesia’s public goals to expand digital infrastructure and build domestic AI capabilities.
CRWV Stock Card
CoreWeave, Inc. Class A Common Stock, CRWV

What the release doesn’t say is how this move undermines the Big Three’s APAC AI pricing power. Southeast Asian AI teams have complained for years about 30-40% higher compute costs compared to U.S. rates. Forced cross-border routing adds 200+ ms of latency for inference workloads that need to serve local users. Local data residency rules in Indonesia, Thailand and Vietnam also lock many teams out of using compute hosted in Singapore. CoreWeave’s specialty high-density GPU clusters already outperform generalist cloud instances on AI training and inference by 20-30% in independent tests. This footprint lets them undercut AWS, Azure and Google Cloud on price and performance for the region’s fastest growing workloads. They don’t have to carry the overhead of underutilized general compute servers weighing down their margins.

Specialized AI cloud providers will capture 40% of APAC AI compute spending by 2030, leaving generalist clouds fighting for leftover low-margin workloads.

Author bio: Ethan Gallagher, a Silicon Valley hardware architect and infrastructure strategist with 12 years designing large-scale AI compute clusters.