Aolani and NVIDIA Plan 22,000-GPU AI Factories Including Philippines

Singapore-founded AI cloud provider Aolani said on September 22, 2026 it is working with NVIDIA to deploy twenty-two thousand Blackwell Ultra GPUs across new artificial-intelligence factories in Malaysia and the Philippines, with deployments expected to begin in early 2027. Aolani claimed the expansion would lift its total past forty-eight thousand GPUs and more than one hundred megawatts of AI-factory capacity, and described the Philippine site as among the country’s first large-scale accelerated-computing platforms—though exact GPU splits, campus addresses and capital expenditure were not disclosed. The deal uses a revenue-sharing and credit-support model meant to align build-out with customer demand rather than speculative empty halls.

Filed under Infrastructure and dated September 26, 2026, this AI4Philippines briefing treats the Aolani plan as Filipino GPU-capacity news distinct from yesterday’s policy-coherence debate. It lands beside the $34.4 billion AI+ Infrastructure Masterplan’s early private-capital tests and PLDT’s Cavite hyperscale path, offering another bidder for Luzon power and interconnection while hyperscalers still only weigh sites.

Why it matters: Filipino startups and agencies still rent scarce GPUs abroad. Local Blackwell Ultra clouds can cut latency—but only if power, residency SLAs and exit rights are contractual.

What it means in practice

Aolani and NVIDIA Plan 22,000-GPU AI Factories Including Philippines — contextual photo

Philippine CIO and procurement leads should shortlist workloads eligible for Aolani tenancy; demand named megawatt and interconnection dates; assign an owner for partner due diligence; run time-boxed proofs on non-critical inference; and prefer contracts that keep humans on spend caps. Connect the announcement to the $34 billion AI masterplan and hyperscalers weighing Luzon sites.

Caveats come first. Press claims are not energized campuses; GPU counts can slip with supply; and “first large-scale” branding is the company’s own. AI4Philippines therefore presents the 22,000-GPU plan as directional infrastructure context until site permits and customer offtakes appear.

What to watch next: Philippine site selection; early-2027 rack energisation; and whether DICT masterplan incentives attach. Readers can continue on the AI4Philippines homepage, or browse the Newsroom for additional briefings.

Bottom line: treat this update as orientation, not instruction. Philippine AI infrastructure is attracting mid-tier GPU clouds under the masterplan umbrella and remains contingent. Organizations that benefit most will verify megawatts, keep humans on spend authority, and refuse to confuse a regional press note with finished clusters.