Strongest APAC hyperscaler footprint. SG, Jakarta (3 AZs), KL, Bangkok, Manila. Best for China-SEA corridor workloads. DAMO AI research hub in SG.
https://alibabacloud.com โAlibaba Cloud launched a new public cloud region in Johor, Malaysia in July 2026, adding sovereign-adjacent compute capacity to the SingaporeโJohor corridor. The launch gives enterprises and AI workloads a Johor-domiciled alternative to the supply-constrained Singapore market.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-july-2026 โAlibaba Cloud opened two new data centres in Johor on 16 June 2026, bringing its Malaysian footprint to five facilities and activating the country's second public cloud region โ the largest Alibaba Cloud presence anywhere in Southeast Asia. The launch adds availability zones to the firm's ASEAN reach (104 AZs across 32 regions globally) and targets SME/MSME AI workloads at competitive price-performance versus rivals.
https://w.media/alibaba-cloud-launches-two-data-centers-in-johor/ โAlibaba Cloud launched a new public cloud region in Johor, Malaysia in July 2026, bringing its total Malaysia facilities to five and establishing its largest Southeast Asia footprint. The launch adds AI compute availability within the JohorโSingapore corridor, where colocation vacancy sits at just 0.7%.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-july-2026 โAlibaba Cloud launched a new public cloud region in Johor in early July 2026, adding two data centres and bringing its total Malaysia footprint to five facilities โ its largest presence in Southeast Asia; anchor customers at launch include TNG Digital and YTL AI Labs, consolidating Johor's position as ASEAN's fastest-growing alternative hyperscale hub.
https://w.media/alibaba-cloud-launches-two-data-centers-in-johor/ โp5 (H100) and p4d (A100) instances. Regions in SG, KL, Jakarta, Bangkok. SageMaker managed ML. Most enterprise ASEAN workloads run here. $33B SEA investment by 2039.
https://aws.amazon.com โCGK4 is Indonesia's first NVIDIA DGX-Ready-certified campus with up to 650 MW renewable-aligned capacity supporting H100 AI training; CGK3 brings liquid-cooled, high-density compute for GB200; CGK5 adds 300 MW hyperscale footprint. Total 1.2 GW power secured from PLN.
https://technode.global/2026/06/03/singapore-based-data-center-developer-bdx-secures-1-2-gw-of-power-in-indonesia-with-state-utility-pln/ โDayOne partnered with Australia's Firmus Technologies to develop a 360 MW AI data-centre campus on Indonesia's Batam Island, near Singapore, as part of a broader AI infrastructure buildout in the region; Firmus simultaneously secured a 12-year, 600 MW wholesale energy supply agreement with Gunvor Group for the next phase of its Project Southgate AI Factory programme.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-august-2026 โA3 (H100) and A2 (A100) instances. Cloud regions in SG, Bangkok (2025), Jakarta. TPU v5e for efficient training. Vertex AI. $1B Thai DC investment.
https://cloud.google.com โHuawei Ascend AI chips + GPU cloud. Singapore, Bangkok, KL regions. Strong for enterprises with China ties. ModelArts AI development platform.
https://intl.huaweicloud.com โSingapore data center. IBM watsonx AI platform. GPU cloud for enterprise ML workloads. Best for enterprises already running IBM software stack.
https://ibm.com/cloud โND H100 v5 series. Regions in SG, KL, Jakarta. $1.7B committed to Indonesia. Azure OpenAI Service. Best for enterprises in Microsoft ecosystem.
https://azure.microsoft.com โNaver announced 55MW Nvidia GPU expansion starting in 2027, with existing Singapore region operational and future Vietnam and Thailand regions planned.
https://www.datacenterdynamics.com/en/news/naver-to-deploy-55mw-of-nvidia-hardware-in-south-korea/ โBM.GPU.H100 bare-metal instances. SG and MY regions. Typically 30โ50% cheaper than AWS/Azure for same GPU. Strong for bare-metal HPC workloads.
https://cloud.oracle.com โSTT Global Data Centers announced in early July 2026 that it will expand its Jakarta campus by adding new facilities totalling 360 MW of AI-ready IT capacity, lifting the campus toward a combined 2.7 GW target by 2030. The expansion directly addresses surging hyperscaler demand in Jakarta, where a single incoming cloud customer is reportedly absorbing 300 MW โ equivalent to all of Jakarta's prior live capacity.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-july-2026 โSTT GDC announced a 360 MW AI-ready IT capacity expansion to its Jakarta campus this week, part of a trajectory to deliver a combined 2.7 GW on the same campus by 2030 โ one of the largest single-campus AI compute buildouts in Southeast Asia and a direct response to surging hyperscaler demand in Indonesia.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-july-2026 โStrong Singapore and Bangkok presence. GPU cloud via GN10 instances. AI platform TI Platform. Popular for SEA gaming and media AI workloads.
https://intl.cloud.tencent.com โThe TM Nxera joint venture between Telekom Malaysia and Singtel's Nxera unit completed the structural top-out of its Iskandar Puteri, Johor campus in June 2026, with the first 64 MW phase due live in 2026 as part of a total 280 MW campus. Separately, STT GDC (Singtel/KKR) secured a $1.37 billion green financing facility to fund STT Johor โ making Johor home to two of ASEAN's largest concurrently financed AI-grade campuses and cementing the corridor's role as Singapore's primary overflow compute hub.
https://www.datacenterdynamics.com/en/news/singtels-nxera-and-telekom-malaysia-top-out-data-center-in-johor/ โLargest chips ever built. CS-3 is a 4-trillion transistor wafer-scale engine. 1000x faster on sparse workloads than GPU equivalents. Used for specialized research.
https://cerebras.ai โMulti-vendor AI hardware cloud. Unique for offering AMD MI300X and Intel Gaudi 3 alongside NVIDIA. Good for teams benchmarking hardware alternatives.
https://cirrascale.com โSustainable GPU cloud using underutilized data center capacity. Carbon-neutral compute. B Corp certified. APAC-accessible. Good for climate-conscious ASEAN AI teams.
https://cudocompute.com โEquinix announced a $190 million investment in its KL2 data center in Cyberjaya, Malaysia โ renewable-energy-powered from day one and delivering over 2,200 cabinets upon full build-out, located less than 1 km from its existing Malaysian facility. The investment deepens Equinix's Malaysia footprint and provides AI-grade, carrier-neutral colocation capacity directly adjacent to the Johor hyperscale corridor.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-june-2026 โLarge-scale GPU cluster leasing for foundation model training. Focus on 1000+ GPU runs. Long-term contracts. Used by US and APAC AI labs for frontier training.
https://foundry.xyz โPan-European GPU cloud with Singapore node. Good APAC latency. Bare metal and virtual GPU options. Strong for media, gaming, and AI inference workloads.
https://gcore.com โEuropean cloud running on 100% renewable energy (Iceland). Good price for A100. GDPR compliant. Suits ASEAN teams with ESG reporting requirements.
https://genesiscloud.com โ200 MW AI data center campus in Korat, Thailand with six data halls supporting 150 MW of IT load and approximately 76,000 GPUs announced in June 2026.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-june-2026 โWorld's fastest LLM inference via custom LPU chips. 500+ tokens/sec on Llama 3. Zero memory bottleneck architecture. ASEAN startups use for real-time AI features.
https://groq.com โDecentralized GPU network aggregating idle ML compute. 40,000+ GPUs across 138 countries. Solana-based payments. Prices 50โ90% below cloud market rates. APAC nodes active.
https://io.net โGerman-based GPU cloud with EU data residency. Hourly billing, no lock-in. Strong for EU-compliant model training. Accessible from ASEAN with moderate latency.
https://leadergpu.com โMalaysian Resources Corporation Berhad (MRCB) and Taiwan's TECO Electric & Machinery signed an MOU on August 10, 2026, to develop and validate next-generation AI-ready Modular Data Centers in Malaysia, expanding on MRCB's previously announced 65 MW AI-ready facility. The modular approach lets MRCB unlock its existing landbank at variable densities โ air or liquid-cooled โ catering to Malaysia's growing pipeline of enterprise and hyperscale AI workloads as the government tightens approvals for non-AI data center projects.
https://technode.global/2026/08/10/mrcb-teco-team-up-to-develop-ai-ready-modular-data-centers-in-malaysia/ โNextDC celebrated the topping-out of KL1, its first international data center and Malaysia's first Uptime Institute Tier IV Design-certified facility in Kuala Lumpur, rated 65 MW and awarded Platinum by the Green Building Index. The facility is purpose-built for AI and hyperscale growth, positioning Malaysia's Kuala Lumpur corridor as a primary ASEAN AI-compute hub alongside Johor.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-june-2026 โNVIDIA's specialized cloud for drug discovery and life sciences AI. Used by ASEAN pharma and biotech. Pre-trained models for protein structure, genomics.
https://bionemo.nvidia.com โPremium AI supercomputing-as-a-service on Oracle, Microsoft, Google, AWS. Full DGX SuperPOD (256 H100s). For training frontier models. Most expensive but highest performance tier.
https://nvidia.com/dgx-cloud โPer-minute GPU billing with instant provisioning. No idle charges. API-first. Good for ASEAN teams with sporadic training and inference workloads.
https://oblivus.com โEuropean public cloud with Singapore region. AI Endpoints for model serving. Good price-performance. SOC2, ISO 27001. Popular alternative to US hyperscalers.
https://ovhcloud.com โProprietary Reconfigurable Dataflow Unit chips deliver fastest inference on large models. SambaNova Suite for enterprises. 8x H100-equivalent speed on LLM inference.
https://sambanova.ai โFrench cloud with strong EU data sovereignty story. Paris and Amsterdam GPU regions. APAC-accessible. Popular with ASEAN companies needing GDPR-compliant training.
https://scaleway.com โSTT Global Data Centers announced it will expand its Jakarta data center campus by adding new facilities totalling 360 MW of AI-ready IT capacity, bringing the campus toward a projected combined capacity of 2.7 GW by 2030. The expansion is engineered for next-generation AI workloads and targets a PUE of 1.25 with direct-to-chip liquid cooling.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-july-2026 โSTT Global Data Centers launched STT Jakarta 2 (24 MW) and energised an onsite high-voltage substation โ believed to be the first on any Indonesian data center campus โ advancing a 360 MW AI-ready pipeline toward a combined 2.7 GW campus target by 2030. Certified as an NVIDIA colocation partner with carbon-neutral operations via PLN collaboration. Adds the only NVIDIA Vera Rubin-certified colocation capacity publicly confirmed in Indonesia, directly filling the sovereign-compute gap Jakarta needs as hyperscaler demand accelerates.
https://w.media/stt-gdc-expands-jakarta-ai-dc-to-advance-360mw-pipeline/ โTrue Internet Data Center (Charoen Pokphand Group) broke ground on its seventh facility โ a 6-billion-baht ($182M) AI hyperscale data center in Northern Bangkok โ on 20 June 2026, targeting Q3 2027 launch. The facility is purpose-built for GPU and AI workloads, features a smart fan-wall liquid-cooling system, and connects to Thailand's largest core network hub, adding to Thailand's 944 MW under-construction pipeline.
https://technode.global/2026/06/22/true-idc-invests-182m-in-its-7th-data-center-in-northern-bangkok/ โTrue Internet Data Center, Thailand's largest DC operator under Charoen Pokphand Group, broke ground on its seventh facility in Northern Bangkok with a $182 million investment, designed as an AI hyperscale site connected to Thailand's largest core network hub and supporting real-time cloud, AI, and digital-platform workloads. The project is positioned as a key node in Thailand's bid to become ASEAN's digital and AI hub.
https://technode.global/2026/06/22/true-idc-invests-182m-in-its-7th-data-center-in-northern-bangkok/ โTrue Internet Data Center (Charoen Pokphand Group) broke ground on its 7th facility in Northern Bangkok on June 20 2026, committing over 6 billion baht ($182M) to an AI hyperscale campus designed for cloud and GPU workloads with liquid-cooling readiness and Thailand's best-in-class PUE. Operations target Q3 2027, adding capacity to CP's existing live AI hyperscale facility and deepening Thailand's push to become the digital hub of ASEAN.
https://technode.global/2026/06/22/true-idc-invests-182m-in-its-7th-data-center-in-northern-bangkok/ โSingapore-based neocloud deployed Nvidia GB200 NVL72 system in Malaysia in May 2026 to support AI infrastructure expansion.
https://www.datacenterdynamics.com/en/news/bitdeer-ai-deploys-nvidia-gb200-nvl72-system-in-malaysia/ โSingapore-based Bridge Data Centres IIO (Thailand) Co., Ltd. received Thailand BOI approval for a 134 MW data center in Chonburi with an investment value of 24.6 billion baht (~$746 million). The project adds significant AI-ready colocation capacity in Thailand's Eastern Economic Corridor, where Bridge already hosts TikTok's AI workloads in Malaysia.
https://technode.global/2026/05/07/thailand-approves-29b-investment-wave-as-data-center-demand-surges/ โLargest specialist GPU cloud. NVIDIA Elite CSP. 250,000+ GPUs. Kubernetes-native. Best for large-scale training runs. No APAC region โ high latency from SEA.
https://coreweave.com โAnnounced 6 August 2026, CoreWeave's first APAC footprint: three owned-and-operated facilities in Greater Jakarta totalling 360 MW of contracted IT power, targeting AI labs, startups, enterprises, and government customers across Southeast Asia, with online date of 2028. Moody's flagged the 360 MW as ~10% of CoreWeave's existing 3.5 GW contracted global base, noting no anchor customer disclosed yet.
https://technode.global/2026/08/06/coreweave-expands-cloud-ai-platform-to-indonesia-marking-first-move-into-asia-pacific-region/ โCoreWeave announced three facilities totalling 360 MW of contracted IT power in Greater Jakarta โ its first Asia-Pacific footprint โ expected online in 2028 and targeting AI labs, startups, enterprises, and government customers across Southeast Asia. The move, flagged by Moody's as CoreWeave's first expansion beyond the US and Europe, represents ~10 % of the company's existing 3.5 GW contracted global base.
https://technode.global/2026/08/06/coreweave-expands-cloud-ai-platform-to-indonesia-marking-first-move-into-asia-pacific-region/ โDecentralized cloud compute via Akash Network. Lowest spot prices globally. No SLA. ASEAN teams use for batch workloads and CI/CD ML pipelines.
https://akash.network โCoreWeave announced three Indonesian data centers totalling 360 MW of contracted IT power, its first Asia-Pacific footprint, owned and operated by CoreWeave and expected online in 2028; Moody's confirmed the move extends CoreWeave's addressable market into enterprise, sovereign, and AI-native customers across Southeast Asia.
https://technode.global/2026/08/06/coreweave-expands-cloud-ai-platform-to-indonesia-marking-first-move-into-asia-pacific-region/ โFinnish GPU cloud with European data privacy compliance. APAC-accessible. H100 cluster availability. Green energy powered. Good for GDPR/PDPA-compliant workloads.
https://datacrunch.io โJapan's Datasection began phased operations of its Bangkok-area AI data center in JulyโAugust 2026, deploying 4,696 NVIDIA B200 GPUs built on ASRock Rack infrastructure for an unnamed US-based technology company, with plans to ramp capacity further. The facility marks Japan's first major GPU-cloud footprint in Southeast Asia.
https://www.datacenterdynamics.com/en/news/japans-datasection-to-deploy-4696-b200-gpus-in-thailand/ โTokyo Stock Exchange-listed neocloud Datasection purchased 4,696 NVIDIA B200 GPUs from ASRock Rack for deployment in a Bangkok data center, with phased operations targeted for JulyโAugust 2026; capacity is pre-committed to an unnamed US-based technology company. Puts Japan-origin neocloud capital directly into ASEAN AI compute supply for the first time at this scale.
https://www.datacenterdynamics.com/en/news/japans-datasection-to-deploy-4696-b200-gpus-in-thailand/ โTokyo Stock Exchange-listed Datasection is deploying a 5,000-GPU NVIDIA B200 cluster in Bangkok โ hardware sourced under a December 2025 Inventec agreement โ operated in partnership with NVIDIA-certified cloud partner CUDO Ventures. Projected to contribute JPY 15.9 billion to Datasection's FY03/27 revenue; part of a ~20,000-GPU, 40 MW multi-country build-out targeting Japan, Australia, and Thailand.
https://w.media/japans-datasection-bets-on-dominance-with-gpu-supercluster-ambitions/ โTokyo-listed neocloud operator Datasection is deploying a 5,000-GPU NVIDIA B200 cluster in Bangkok, operated in partnership with NVIDIA-certified cloud partner CUDO Ventures; hardware sourced under a December 2025 Inventec agreement covering 10,000 B300/B200 GPUs across Japan, Thailand, and Australia. The Bangkok cluster materially adds sovereign-grade AI training capacity to Southeast Asia's tightest compute market.
https://w.media/japans-datasection-bets-on-dominance-with-gpu-supercluster-ambitions/ โTokyo Stock Exchange-listed neocloud Datasection is deploying a 5,000-GPU NVIDIA B200 cluster in Bangkok, operated in partnership with NVIDIA-certified cloud partner CUDO Ventures, with hardware sourced under a December 2025 Inventec agreement. The cluster is classified as a highly probable pipeline project targeting ~JPY 15.9B in FY03/27 revenue, adding Bangkok as a sovereign-AI-grade inference and training node for Southeast Asia.
https://w.media/japans-datasection-bets-on-dominance-with-gpu-supercluster-ambitions/ โDayOne Data Centers partnered with Australia's Firmus Technologies to develop a 360 MW data center campus on Indonesia's Batam Island, near Singapore, as part of a broader AI Factory buildout across the region. The campus is designed with liquid-cooling-enabled, high-density racks and sits at the center of the SIJORI corridor, giving hyperscale tenants low-latency access to Singapore interconnect while benefiting from Indonesia's lower construction costs.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-august-2026 โDeployed AI infrastructure in Singapore via ST Telemedia Global Data Centres; part of NVIDIA AI Cloud ecosystem announced June 2026.
https://blogs.nvidia.com/blog/ai-cloud-ecosystem/ โAustralia's Firmus Technologies and Singapore-headquartered DayOne are co-developing a 360 MW NVIDIA DSX AI Factory campus at Kabil Tech Park in Batam, deploying up to 170,000 Grace Blackwell, Vera Rubin, and Vera AI accelerators through 2027โ2028. DayOne holds a 450 MW PPA with PT PLN Batam, and the campus targets AI-native cloud customers across Asia-Pacific โ one of the largest GPU deployments on record in ASEAN.
https://www.datacenterdynamics.com/en/news/firmus-to-deploy-170000-gpu-cluster-in-batam-indonesia/ โAustralian neocloud Firmus Technologies, in an eight-year NVIDIA partnership, is deploying up to 170,000 NVIDIA AI accelerators (Grace-Blackwell, Vera-Rubin, Vera) at a 360 MW DayOne campus in Batam's Kabil Industrial Tech Park, rolling out through 2027โ2028; the campus is backed by a 450 MW PPA with PT PLN Batam and is among the largest AI-factory builds in Asia-Pacific, putting sovereign-grade GPU capacity within Singapore's backyard for the first time at this scale.
https://www.datacenterdynamics.com/en/news/firmus-to-deploy-170000-gpu-cluster-in-batam-indonesia/ โAustralian neocloud Firmus Technologies and Singapore-based DayOne are co-developing a 360 MW NVIDIA DSX AI Factory at the Kabil Industrial Tech Park on Batam Island, backed by an eight-year NVIDIA chip-and-cloud-revenue partnership. The campus โ among the largest AI infrastructure builds in Asia-Pacific โ targets 'AI Native' customers and sits within DayOne's SIJORI cross-border platform linking Singapore, Johor, and Batam.
https://w.media/firmus-dayone-to-co-develop-360mw-nvidia-ai-factory-in-batam/ โNVIDIA and Australian neocloud Firmus Technologies announced a partnership on ~June 29, 2026 to deploy up to 170,000 NVIDIA AI accelerators โ spanning Grace-Blackwell, Vera-Rubin, and Vera platforms โ at a DayOne data-center campus in Batam's Kabil Tech Park, rolling out through 2027โ2028 at up to 360 MW. The deal is structured as a revenue-sharing and credit-support model, making it one of the largest single AI-infrastructure commitments in Asia-Pacific and positioning Batam as a rival to Singapore for frontier GPU capacity.
https://technode.global/2026/06/29/nvidia-australias-firmus-to-jointly-deploy-170000-ai-accelerators-in-indonesia-targeting-30b-in-revenue/ โAustralian neocloud Firmus Technologies and Singapore-based DayOne are co-developing a 360 MW NVIDIA DSX AI Factory at DayOne's Kabil Industrial Tech Park campus in Batam, Indonesia, with up to 170,000 NVIDIA AI accelerators across Grace-Blackwell, Vera-Rubin, and Vera platforms to be deployed through 2027โ2028. The campus is backed by a 511 MVA (~450 MW) PPA with PT PLN Batam โ Indonesia's largest power deal โ positioning Batam as one of the largest AI infrastructure projects in Asia-Pacific.
https://w.media/firmus-dayone-to-co-develop-360mw-nvidia-ai-factory-in-batam/ โAustralian neocloud Firmus Technologies and Singapore-headquartered DayOne are co-developing a 360 MW NVIDIA DSX AI Factory campus in Batam under an eight-year NVIDIA partnership; the deal positions Batam among Asia-Pacific's largest AI infrastructure developments and DayOne has already signed a 511 MVA PPA with PT PLN Batam to power the campus.
https://w.media/firmus-dayone-to-co-develop-360mw-nvidia-ai-factory-in-batam/ โAggregates idle data center capacity globally. Strong H100 availability. Per-second billing. No long-term contracts. Popular with European and APAC ML teams.
https://fluidstack.io โVietnam's G-Group received its investment certificate on July 1, 2026 for a $300M AI data-center campus in Hanoi targeting 30 MW of AI and HPC capacity, co-developed with Oman's OTECH (Omantel Group) under a tripartite agreement with the Hanoi People's Committee. The campus is designed to serve Hanoi's stated goal of running a digital government on AI and big-data platforms by 2035, requiring at least two hyperscale data centers in the capital โ making G-Campus the first domestic-led AI-campus commitment at this scale in northern Vietnam.
https://technode.global/2026/07/01/vietnams-g-group-to-invest-300m-in-building-ai-data-center-campus-in-hanoi/ โLondon- and Singapore-headquartered Galaxy Data Center secured $250M in an initial strategic financing round from an unnamed international institution to fund multi-country GW-scale AI data centre campuses across Southeast Asia, using Singapore as its regional hub; the company separately committed $2B to a Rayong, Thailand EEC campus at Silicon Tech Park. The capital unlocks high-density, green AI infrastructure across multiple ASEAN markets simultaneously.
https://technode.global/2026/06/17/galaxy-data-center-secures-250m-to-build-next-generation-green-compute-infrastructure-in-sea/ โGalaxy Data Center closed a $250 million initial strategic financing round in July 2026 to fund its data center platform build-out across Southeast Asia. The raise signals a new ASEAN-focused neocloud entrant targeting the regional GPU and colocation gap at scale.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-july-2026 โGalaxy Data Centers closed a $250 million initial strategic financing round in July 2026 to fund development of its data centre platform across Southeast Asia; specific country footprints and GPU configurations are pending announcement, but the raise signals a new well-capitalised neocloud entrant targeting the region's under-served mid-market compute demand.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-july-2026 โGMI named as NVIDIA AI factory partner at GTC Taipei COMPUTEX 2026, building AI infrastructure capacity in the region.
https://blogs.nvidia.com/blog/nvidia-gtc-taipei-computex-2026-news/ โ200MW AI data center campus in Nakhon Ratchasima with 6 data halls, construction starts July 2026, first phase Q1 2027. Designed for DBOT model (design, build, operate, commercialize).
https://w.media/gorilla-technology-buys-40-acres-land-in-thailand-for-200mw-ai-data-center/ โNasdaq-listed Gorilla Technology acquired a 40-acre site in Korat (Nakhon Ratchasima) for a 200 MW campus with six data halls supporting ~150 MW net IT load and approximately 76,000 GPUs at full build-out. Construction starts July 2026, first phase targeted Q1 2027, serving both domestic Thai and broader ASEAN regional compute demand.
https://www.datacenterdynamics.com/en/news/gorilla-technology-eyes-200mw-data-center-project-in-thailand/ โUK-listed AI-infrastructure firm Gorilla Technology has purchased a 40-acre site in Korat (Nakhon Ratchasima), 260 km northeast of Bangkok, for a six-hall, 200 MW campus targeting ~150 MW net IT load and ~76,000 GPUs at full build-out. Construction starts July 2026; Phase 1 RFS Q1 2027. The campus follows a DBOT (Design-Build-Operate-Transfer) model serving governments, enterprises, and strategic partners across Southeast Asia.
https://www.datacenterdynamics.com/en/news/gorilla-technology-eyes-200mw-data-center-project-in-thailand/ โUK-listed AI infrastructure firm Gorilla Technology has purchased a 40-acre site in Korat (Nakhon Ratchasima), ~260 km northeast of Bangkok, for a 200 MW campus comprising six data halls (five ร 30 MW + one ร 50 MW) with ~76,000 GPUs at full build-out under an all-GB300 configuration. Construction breaks ground July 2026 with Phase 1 targeted for Q1 2027; campus is purpose-built to serve both domestic Thai demand and regional ASEAN GPU-as-a-Service customers via Gorilla's DBOT model.
https://www.datacenterdynamics.com/en/news/gorilla-technology-eyes-200mw-data-center-project-in-thailand/ โUK-based Gorilla Technology Group has acquired a 40-acre site in Korat (Nakhon Ratchasima) for a 200 MW AI campus with six data halls (five at 30 MW, one at 50 MW) and 150 MW net IT load; construction starts July 2026, Phase 1 targeted for Q1 2027. Designed under a Design-Build-Operate-Transfer model to serve both domestic Thai and broader Southeast Asian regional compute demand.
https://www.datacenterdynamics.com/en/news/gorilla-technology-eyes-200mw-data-center-project-in-thailand/ โUK-based Gorilla Technology broke ground this month (July 2026) on a 40-acre, 200 MW campus in Korat (Nakhon Ratchasima), 260 km northeast of Bangkok, comprising six data halls (5ร30 MW + 1ร50 MW) supporting ~150 MW of net IT load and up to 76,000 GPUs at full deployment; Phase 1 targets completion Q1 2027, positioning the campus as a regional compute node for both domestic Thai and broader ASEAN demand.
https://www.datacenterdynamics.com/en/news/gorilla-technology-eyes-200mw-data-center-project-in-thailand/ โUK-based Gorilla Technology secured a 40-acre site in Korat for a 200 MW campus (150 MW net IT load) across six data halls, targeting ~76,000 GPUs at full build-out; construction starts July 2026, Phase 1 complete Q1 2027. Gorilla is targeting ~$1.5 billion in annualised revenue from the campus by 2028, serving both Thai sovereign and regional Southeast Asian demand.
https://www.datacenterdynamics.com/en/news/gorilla-technology-eyes-200mw-data-center-project-in-thailand/ โUK-based Gorilla Technology has acquired a 40-acre site in Nakhon Ratchasima (Korat), 260 km northeast of Bangkok, for a 200 MW AI campus with six data halls (five ร 30 MW + one ร 50 MW) supporting ~76,000 GPUs at full build-out. Construction starts July 2026; Phase 1 targets Q1 2027 completion, serving both domestic Thai and regional Southeast Asian compute demand under a Design-Build-Operate-Transfer (DBOT) model.
https://www.datacenterdynamics.com/en/news/gorilla-technology-eyes-200mw-data-center-project-in-thailand/ โGorilla Technology Group announced a 200 MW AI data center campus in Korat, Thailand, comprising six data halls supporting 150 MW of IT load and approximately 76,000 GPUs at full deployment. Adds Thailand's first large domestic GPU-campus footprint outside Bangkok, diversifying the country's AI compute geography.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-june-2026 โUK-listed AI infrastructure firm Gorilla Technology has acquired 40 acres in Korat (Nakhon Ratchasima), ~260 km northeast of Bangkok, for a 200 MW campus comprising six data halls (five at 30 MW, one at 50 MW) with ~76,000 GPUs at full deployment. Construction starts July 2026, Phase 1 targeted for Q1 2027; the campus is designed under a DBOT model for governments, enterprises, and strategic partners across ASEAN.
https://www.datacenterdynamics.com/en/news/gorilla-technology-eyes-200mw-data-center-project-in-thailand/ โSimple GPU cloud with clean API. H100 and A100 on-demand. No minimum commitment. Suited for ASEAN startups building and testing AI models iteratively.
https://hyperstack.cloud โIndosat announced as NVIDIA AI factory partner at GTC Taipei COMPUTEX 2026, expecting ~3% EBITDA contribution from AI cloud by 2026.
https://blogs.nvidia.com/blog/nvidia-gtc-taipei-computex-2026-news/ โML-focused cloud with Jupyter-native workflow. Good JupyterLab integration. Budget option for research and development. APAC-accessible from Singapore.
https://jarvislabs.ai โBest on-demand H100 price globally. ML-optimized UX with pre-installed frameworks. No APAC region but good for training jobs where data center location doesn't matter.
https://lambdalabs.com โBare metal GPU cloud with data centers in Sรฃo Paulo, Dallas, Santiago, Warsaw. APAC expansion 2025. Predictable pricing, bare-metal performance.
https://latitude.sh โUK-based GPU cloud with H100 cluster focus. Good for EU/UK data residency. Expanding capacity in 2025. Used by European AI startups and research labs.
https://massedcompute.com โServerless GPU with sub-second cold starts. Python-native, zero DevOps. Ideal for ASEAN ML teams wanting no infrastructure overhead. Billed per GPU-second.
https://modal.com โRaised $22M Series A in May 2026; moved HQ from India to Singapore; building full-stack neocloud platform with GPU fleet expansion across India and Singapore.
https://www.datacenterknowledge.com/ai-data-centers/ai-cloud-startup-nava-raises-22m-in-series-a-funding-round โYandex spin-off cloud, fully EU/APAC compliant. H100 and H200 clusters. Strong for European data sovereignty. APAC-accessible with good latency from SG.
https://nebius.ai โAustralian operator NextDC celebrated the topping-out of KL1, its first international Tier IV facility in Kuala Lumpur, in June 2026. The 65 MW campus has already received Uptime Institute Tier IV Design Certification and a Platinum Green Building Index rating, purpose-built for AI and hyperscale growth โ marking NextDC's first footprint outside Australia and adding institutional-grade sovereign-adjacent capacity to Malaysia's fast-growing AI cluster.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-june-2026 โGradient ML platform. Jupyter-native workflow. Acquired by DigitalOcean. Free-tier GPU notebooks for experimentation. Good entry point for ASEAN ML students.
https://paperspace.com โSingapore-headquartered Racks Central, backed by a fresh investment from the US$1B China-ASEAN Investment Cooperation Fund II (CAF II / ESR), is building Southeast Asia's first dedicated AI-factory campus in Johor across four large-scale developments totalling 510 MW. Phase 1 (RCJM1) is RFS Q4 2026; each successive phase will upgrade to next-generation NVIDIA GPU generations through 2028, making it the region's only publicly confirmed Rubin-roadmap campus.
https://w.media/exclusive-johor-set-to-host-southeast-asias-first-ai-factory-campus/ โRun open-source models via API. Huge model library including Llama, SDXL, Whisper. Simple pricing, no infra. Popular with ASEAN startups building AI features quickly.
https://replicate.com โSecure Cloud (Tier 3+, SLA) and Community Cloud tiers. 18+ GPU families. Spot pricing available. Community GPUs in APAC. Popular with ASEAN ML developers.
https://runpod.io โMOU effective 1 August 2026: SC ZEUS sources, builds, and operates data-centre infrastructure; Robust HPC procures and manages high-performance GPU clusters; SB Telecom Singapore handles data-hall fit-out, ICT integration, and connectivity. The trio will develop AI Factory solutions with liquid cooling and 800 VDC architecture across Southeast Asia.
https://technode.global/2026/08/04/japans-softbank-telecom-unit-singapores-sc-zeus-malaysias-robust-hpc-sign-ai-data-center-pact-for-southeast-asia/ โSoftBank Corp. subsidiary SB Telecom Singapore, Singapore-headquartered SC ZEUS Data Centers, and Malaysia-based Robust HPC Group signed a three-way MOU effective August 1, 2026, to build and operate AI data centers and GPU clusters across Malaysia, Thailand, Indonesia, and Vietnam โ with SC ZEUS sourcing land and operating facilities, Robust HPC procuring and managing GPU infrastructure, and SB Telecom handling ICT integration and connectivity. The alliance is the most concrete new multi-country neocloud entry into ASEAN this fortnight, spanning four markets simultaneously.
https://technode.global/2026/08/04/japans-softbank-telecom-unit-singapores-sc-zeus-malaysias-robust-hpc-sign-ai-data-center-pact-for-southeast-asia/ โBangkok-based neocloud expanding offerings to NVIDIA H200 and GB200 Grace Blackwell Superchips as confirmed in NVIDIA partnership July 2025.
https://blogs.nvidia.com/blog/thailand-vietnam-sovereign-ai/ โSingtel Digital InfraCo's AI cloud brand RE:AI has partnered with WEKA to embed WEKA NeuralMesh high-performance software-defined storage into its GPUaaS platform, targeting storage-bottleneck elimination and idle-GPU reduction for sovereign AI workloads; the integrated offering will be sold to governments and enterprises across Singapore and ASEAN. Singtel also integrates NeuralMesh into its existing national-scale GPUaaS, extending the reach of Singapore's sovereign compute stack to the wider region.
https://www.datacenterdynamics.com/en/news/gmi-cloud-and-magna-ai-partner-on-global-network-of-sovereign-ai-factories/ โMarketplace model like Vast.ai but with stricter quality control. RTX consumer GPUs at low cost. Good for ASEAN dev teams on tight budgets.
https://tensordock.com โGB200 clusters for trillion-parameter model inference. Best for real-time large model serving. Serverless API plus dedicated clusters. Used by ASEAN LLM startups.
https://together.ai โUSD.AI (AI compute financing platform) and Singapore-headquartered BSQ Capital Partners announced a $300M initial allocation on July 16, 2026 to provide GPU-backed financing to AI compute operators across Asia-Pacific, structured as first-priority security over GPU assets and their cashflows; designed to close the capital-access gap for ASEAN neocloud operators who hold high-value GPU fleets but lack bank-ready underwriting. Directly enables the next wave of ASEAN GPU cloud builds.
https://w.media/usd-ai-and-bsq-capital-jv-allocates-us-300-million-gpu-backed-financing-for-apac-data-centers/ โSingapore-headquartered BSQ Capital Partners and US-based USD.AI launched a joint venture on 16 July 2026, allocating $300M in GPU-backed senior secured financing to APAC AI compute operators โ addressing the gap between high-value GPU fleets and banks unwilling to underwrite accelerator hardware as collateral. This is the first institutional GPU-collateralized lending vehicle headquartered in Singapore, directly enabling ASEAN neocloud scale-up.
https://w.media/usd-ai-and-bsq-capital-jv-allocates-us-300-million-gpu-backed-financing-for-apac-data-centers/ โOn July 16 2026, USD.AI and Singapore-headquartered BSQ Capital Partners announced an initial $300 million allocation of GPU-backed senior secured loans, collateralised by first-priority security over GPUs and their contracted cashflows, targeting AI compute operators across Asia-Pacific โ unlocking balance-sheet capacity for ASEAN neoclouds that lack traditional bank financing.
https://w.media/usd-ai-and-bsq-capital-jv-allocates-us-300-million-gpu-backed-financing-for-apac-data-centers/ โGPU marketplace โ rent idle datacenter and consumer GPUs. Cheapest option for non-critical workloads. Some APAC hosts. No SLA. Best for experimentation.
https://vast.ai โ32 cloud locations including Tokyo, Seoul, Sydney. Nearest APAC nodes to SEA. Bare metal GPU and cloud GPU tiers. API-first, developer-friendly. Good ASEAN latency.
https://www.vultr.com โMalaysia's first NVIDIA-powered AI data center in Johor, operational May 2025. RM10B ($2.38B) total investment; serving global hyperscalers.
https://w.media/ytl-power-completes-nvidia-powered-ai-data-centre-in-johor/ โDecentralized GPU cloud, Singapore HQ. 435,000+ GPU containers. Novel distributed model โ rents idle enterprise GPUs worldwide. Serves AI and Web3 clients.
https://aethir.com โMalaysia's leading carrier-neutral DC operator. AI-optimized colocation in Kuala Lumpur. Strong interconnect for Malaysia GPU buildout. MCMC compliant.
https://aims.com.my โIndonesia's first sovereign AI data center (Dec 2024). JV: Indosat + Lintasarta + BDx. NVIDIA-powered, renewable energy. Most interconnected AI campus in SEA.
https://bdxworld.com โHyperscale AI data center campuses in Singapore, Malaysia, Thailand. Built for AI GPU density. Directly connected to fiber submarine cables.
https://edgeconnex.com โNVIDIA Reference Platform Partner in APAC. AI Inference Engine for ultra-low latency. Elastic GPU scaling. Focused on LLM inference and training in APAC.
https://gmicloud.ai โ200 MW AI data center campus in Nakhon Ratchasima with six data halls (150 MW IT load). Construction commencing July 2026, first phase Q1 2027.
https://w.media/gorilla-technology-buys-40-acres-land-in-thailand-for-200mw-ai-data-center/ โ200 MW AI data center campus in Korat, Thailand with six data halls supporting 150 MW of IT load and approximately 76,000 GPUs at full deployment.
https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-june-2026 โPreferred NVIDIA CSP Partner in SEA. 20MW AI-ready capacity. Merged with VNG Cloud (Dec 2025). DC in Bangkok (STT GDC, Tier III) and Vietnam. First SEA-native AI cloud at hyperscale.
https://greennode.ai โGPU cloud with APAC availability zones. Singapore and Hong Kong nodes. Enterprise and startup plans. Good for ASEAN data residency requirements.
https://www.layerstack.com โNTT global data center footprint in Singapore, Thailand, Indonesia. AI-grade facilities. Colocation and managed GPU services for enterprise ASEAN clients.
https://hello.global.ntt โPan-Asian data center operator. Facilities in Singapore, Jakarta, Mumbai. AI-grade power and cooling. Backed by Warburg Pincus. Key ASEAN AI infra node.
https://pdg.com โEdge GPU cloud across Southeast Asia. Deployed in Malaysia, Singapore, Thailand, Indonesia. Low-latency inference optimized for ASEAN AI applications.
https://radiarc.io โEffective 1 August 2026, SoftBank's Singapore telecom unit, SC ZEUS Data Centers, and Malaysia's Robust HPC Group signed a three-way MOU to build AI Factory solutions across Southeast Asia: SC ZEUS builds and operates the data center shell, Robust HPC procures and manages GPU clusters, and SB Telecom Singapore handles ICT integration and connectivity.
https://technode.global/2026/08/04/japans-softbank-telecom-unit-singapores-sc-zeus-malaysias-robust-hpc-sign-ai-data-center-pact-for-southeast-asia/ โLEED Gold, Tier III data center in Bangkok. GreenNode hyperscale node. Thailand's largest carrier-neutral DC. AI-ready power density up to 30kW/rack.
https://sttgdc.com โState telco Telkom's AI data center initiative. NVIDIA partnership. National AI infrastructure for Indonesian government and enterprise workloads.
https://indihome.co.id โTelekom Malaysia's data-center and GPU-as-a-service arm (with Singtel's Nxera), building an AI-ready hyperscale campus in Johor for hyperscalers, enterprise and AI workloads.
https://www.tm.com.my โIndia-origin data center expanding in SEA. AI-first campuses. Green power focus. Targeting Indonesian and Malaysian AI data center demand wave.
https://yotta.com โ600MW NVIDIA-powered sovereign AI data center in Kulai, Johor โ partly solar (~500MW). Anchors YTL Power's $2.36B NVIDIA agreement and Malaysia's sovereign LLM (ILMU).
https://www.ytl.com โMalaysia's first NVIDIA-powered AI data center in Johor, completed and operational; 1,640 acres with liquid-cooled GB200 GPUs serving hyperscaler customers.
https://w.media/ytl-power-completes-nvidia-powered-ai-data-centre-in-johor/ โ