The disclosure and our analysis
Evidence dates
September 14 announcement and subsequent operations update, whose page is dated September 16 and whose release dateline is September 17, 2026.
Bitdeer AI announced a ten-year data center services agreement for A202 in Johor Bahru, Malaysia. The facility represents 65.1 MW of critical IT load, with energization targeted for Q3 2027. This is a commitment for future infrastructure, not confirmation of operating capacity. [Source 1]
What is contracted
The company's subsequent operations table separates its data center agreements from its customer cloud contracts. A202 has a signed data center contract, while its cloud contract and long-term coverage fields are blank. Those blanks do not establish an absence of customers, but they also do not substantiate a sold-out facility. [Source 2]
The same table lists the neighboring A201 facility at 21.7 MW of IT load, with January 2027 readiness and customer negotiations still in progress. Together, the two Johor facilities account for 86.8 MW of planned IT capacity. Their differing delivery dates matter to buyers evaluating a phased deployment. [Source 2]
Economics and material unknowns
The September 14 release does not name A202's facility counterparty or disclose its service charge. It describes infrastructure designed for liquid-cooled NVIDIA systems and says Bitdeer expects customer prepayments to help fund GPU deployment. These are plans, not a published customer price list. [Source 1]
Bitdeer also references expected revenue above $800 million for five-year commitments at a different facility, A102. That comparison is not A202's signed revenue or colocation rental rate. Site, service mix and contract duration can differ. [Source 1]
Buyer analysis
A buyer should separate three commitments: the provider's access to a building, the delivery of usable compute, and the buyer's own reserved service. A signed upstream agreement can support a deployment plan without guaranteeing when a specific customer workload will run.
Before relying on this option, request a phased acceptance schedule, rack density and cooling specifications, network requirements, and the remedies if delivery slips. GPU service pricing should identify hardware, support, networking and energy treatment so that comparisons with a colocation quote are meaningful.
No currently leasable quantity is established by the reviewed disclosures. A current written provider offer is needed to determine what a new customer can reserve and on what terms.