Bitdeer Signs 10-Year Deal for 65.1MW Johor AI Cloud Facility, Secured Capacity Reaches 206.5MW

Bitdeer (NASDAQ: BTDR) has signed a 10-year data center services agreement for a 65.1MW AI cloud facility at its Johor Bahru, Malaysia, campus, a move that lifts the company's secured data center capacity to approximately 206.5MW of critical IT load — equal to 59% of its target of up to 350MW by the first quarter of 2028. The facility, designated A202, is expected to be energized in the third quarter of 2027.

The agreement marks one of the largest single capacity additions for Bitdeer in Southeast Asia and underscores the accelerating demand for liquid-cooled, rack-scale AI infrastructure in the region. Bitdeer defines secured capacity as facilities it owns or has covered with signed data center services agreements. However, the capacity remains uncontracted by customers — as of Monday, the company had no signed offtake commitments for A202 or the neighboring A201 facility, even as its estimated active AI cloud pipeline exceeded $7 billion.

A202 will share the Johor campus with A201, a 21.7MW facility scheduled to enter service in January 2027. Together, the two facilities would provide Bitdeer with 86.8MW of critical IT capacity at the site. The company expects the shared campus to shorten the development schedule by extending existing power, cooling, networking and other infrastructure to A202. Bitdeer did not disclose the cost of the 10-year data center services arrangement or its expected GPU deployment capital expenditure.

"Johor is where we can move fastest," said Retainna Lin, Bitdeer AI's vice president of AI cloud. Lin noted that the company added 65.1MW without originating another site because A202 can leverage infrastructure already associated with A201.

A202 is designed to accommodate NVIDIA systems using liquid cooling at rack scale, including GB300 NVL72 and Vera Rubin platforms. Bitdeer plans to offer GPU cloud services and data hosting from the facility, with equipment purchases tied to executed customer contracts. The company said A201 capacity is in advanced negotiations and it expects to contract A202 before energization. However, Bitdeer cautioned that those discussions may not produce signed agreements before either facility enters service, and final pricing may differ from earlier contracts.

For A202, Bitdeer anticipates broadly similar per-MW economics to those of its 9.5MW Malaysia facility, A102. The five-year offtake commitments at A102 represent more than $800 million of expected revenue, implying over $84 million per MW across the contract term. Bitdeer fully committed A102 ahead of energization after initially contracting half of the facility under a five-year agreement carrying about $400 million of expected revenue. The company cautioned that economics differ by site, service mix, duration and final contract terms, meaning A102 does not establish A202's eventual revenue.

Bitdeer plans to finance A202's GPUs primarily through customer prepayments, which it generally seeks to structure above 50% of related capital expenditure, alongside financing secured by contracted cash flows and operating cash flow. "The signal from customers is unambiguous: demand for liquid-cooled, rack-scale AI Cloud capacity in 2027 is running well ahead of what the market can supply," said Chief Financial Officer Michael G. Potter.

The $7 billion active pipeline does not constitute contracted backlog. Bitdeer describes the amount as management's estimate of potential contract value associated with opportunities being discussed commercially, calculated using assumptions about capacity, contract length, pricing and service mix. The amount represents neither binding customer obligations nor recognized revenue. The planned third-quarter 2027 energization of A202 remains subject to site readiness, power availability, equipment deliveries and required approvals. The timetable also depends on third-party data center providers meeting their obligations, and customer contract conversion will determine when Bitdeer commits capital to the GPUs.