Taiwan-based AI infrastructure company KONST is expanding into Southeast Asia through a new deployment at SMX01 in Jakarta. The company has signed a collaboration agreement with SM+ Data Centers and LG Sinar Mas to secure capacity at the AI-ready facility. The agreement gives KONST a physical compute base outside Taiwan for the first time and supports its GPU cloud and enterprise compute strategy. The move also places the company closer to customers across one of Southeast Asia’s fastest-growing digital markets.

KONST will deploy high-density, liquid-cooled infrastructure at SMX01 for GPU computing and related services. The company plans to combine GPU infrastructure, its Neo Cloud offering and AI applications within the Jakarta deployment. That model allows KONST to connect physical compute capacity with cloud and application services instead of treating GPU hosting as a standalone offering. The strategy also gives the company a platform for serving customers across Taiwan, Japan, Indonesia and other Asian markets.

SMX01 sits in Jakarta’s Central Business District and targets enterprise, hyperscale, cloud and AI workloads. LG Sinar Mas describes the facility as an AI-ready data center with advanced liquid cooling and support for high-density racks. The facility also offers multiple carrier fiber paths and a design focused on low-latency connectivity and modular expansion. SMX01 currently targets Ready for Service in the fourth quarter of 2026.

SMX01 Gives KONST High-Density Capacity

The choice of SMX01 gives KONST access to infrastructure built around demanding compute requirements. LG Sinar Mas says the facility can support high-density racks of up to 130kW per rack and uses advanced cooling systems for intensive workloads. Those specifications matter as AI clusters push rack power and thermal requirements beyond many conventional enterprise deployments. For KONST, the facility therefore provides a physical environment that matches the requirements of modern GPU infrastructure. SMX01 also brings connectivity into the equation. The facility uses multiple carrier fiber paths to improve network diversity and resilience, while its location near Indonesia’s largest internet exchange supports low-latency connectivity. These features can help GPU cloud operators connect compute resources with enterprises, networks and other digital infrastructure. The result is a deployment designed to support both compute density and the connectivity that AI workloads require. SM+ Data Centers, Korea Investment Real Asset Management and LG Sinar Mas developed SMX01 as a flagship Jakarta facility. The partners marked the project’s topping-off milestone in May 2026 as construction moved toward systems completion, data-hall readiness and testing. The project carries an investment value of about $300 million and aims to strengthen Jakarta’s position as a regional digital infrastructure market. The facility’s development also reflects growing demand for cloud, enterprise and AI capacity in Indonesia.

Liquid Cooling Becomes Central To AI Capacity

Liquid cooling sits at the center of KONST’s deployment strategy. High-density GPU clusters generate substantial heat, which makes thermal management a critical part of infrastructure planning. SMX01 uses advanced liquid-cooling systems specifically to support AI and other intensive workloads. That approach allows the facility to target higher-density deployments without relying solely on conventional air-cooling systems. For KONST, the value extends beyond simply obtaining rack space. The company needs power, cooling, networking and operational support to turn GPU hardware into usable compute capacity. SMX01 brings those elements together within a facility designed for high-performance workloads. The arrangement can therefore reduce the infrastructure gap between GPU deployment and commercial AI services.

KONST has built its business around that broader infrastructure model. The company provides GPU data-center construction, GPU cloud services, bare-metal server leasing, colocation services and its Horizon.ai ATP platform. It also works across infrastructure deployment and compute delivery, giving the Jakarta agreement a wider role than a conventional data-center capacity deal. Meanwhile, the Jakarta project adds a regional operating layer to that model. GPU servers for the deployment will come from Giga Computing, a GIGABYTE subsidiary, while Raydian Cloud will support GPU cluster deployment and operations. The partners will combine hardware, deployment and regional operating expertise to support the high-density clusters. That structure gives KONST an ecosystem around the compute layer rather than requiring it to build every capability independently.

KONST Targets Southeast Asia AI Demand

KONST’s Jakarta move comes as Southeast Asia attracts greater investment in AI-ready digital infrastructure. Indonesia offers a large domestic market and an expanding digital economy, while Jakarta has become a major destination for cloud and data-center investment. SMX01 adds another high-density facility to that ecosystem with an architecture aimed at enterprise, hyperscale and AI requirements. KONST can use that platform to place GPU capacity closer to regional customers. The company’s first deployment outside Taiwan also changes the scale of its regional proposition. KONST can connect resources in Taiwan and Jakarta while offering customers different deployment options across Asia. That flexibility becomes more valuable as enterprises seek access to AI compute without committing to their own GPU infrastructure. It also gives KONST a stronger position as regional customers move from experimentation toward sustained AI workloads.

Ben Chang, Chairman and Co-Founder of KONST, said: “Today marks a major milestone for KONST as we take the next step in expanding our AI infrastructure business into Southeast Asia. Indonesia is a key market for AI, and we are proud to work with SM+, LGSM, Giga Computing, and Raydian Cloud to build the next generation of AI data centers. This strong partnership is only the beginning. We believe KONST can become a leading and innovative brand in the AI ecosystem, and we look forward to deepening our collaboration.” The statement highlights the company’s wider ambition beyond a single Jakarta deployment. KONST wants to build an AI infrastructure brand with regional reach, rather than operate only as a Taiwan-based compute provider. The partnership gives that strategy a physical starting point in Southeast Asia. It also connects KONST with infrastructure, hardware and operations partners already active in the region.

SMX01 Builds A Broader Infrastructure Ecosystem

Herson Suindah, President Director & Group CEO of SM+ Holdings, said: “This collaboration turns our shared vision for SMX01—our flagship, future-proof data center—into reality. It is a true partnership that goes beyond a traditional buyer-seller relationship: SM+ and KONST share the same drive and determination to grow together. We are excited to support KONST and to grow alongside the company as AI and GPU development accelerates.” That partnership structure matters because AI infrastructure requires coordination across several layers. Data-center operators provide power, cooling, connectivity and physical space, while compute providers bring GPUs, cloud platforms and customer-facing services. Hardware and operations partners then help convert that infrastructure into a functioning AI cluster. SMX01 brings those roles together around a single deployment.

Raydian Cloud Founder and CEO Marcus Cheng said: “Having played a part in bringing these partners together, I am pleased to see the collaboration reach this milestone. High-density GPU clusters require specialized operating discipline from build-out through Day 2 operations. As trusted partners, Raydian Cloud looks forward to bringing its regional expertise to support KONST, SM+, and LGSM in setting a new benchmark for AI infrastructure in Indonesia.” The operating component could prove as important as the physical infrastructure. High-density GPU environments require careful deployment, monitoring and maintenance after the initial build-out. Raydian Cloud’s role gives the project a regional operations layer that complements KONST’s compute offering. That combination should help the partners address the practical demands of running AI infrastructure at scale.

Jakarta Becomes A Strategic Compute Node

SMX01’s development also strengthens SM+’s position beyond its existing Indonesian data-center network. The company says its data-center business operates and develops facilities across Indonesia, while SMX01 represents a move toward a flagship metro facility with hyperscale capability. The project targets cloud, enterprise and AI customers that need greater density and connectivity. KONST’s arrival adds an AI compute provider to that ecosystem before the facility reaches its planned service date. The facility’s timing could prove important for both companies. SMX01 expects to enter service in the fourth quarter of 2026, giving KONST a defined window to prepare its infrastructure and customer proposition. The project has already moved through its topping-off milestone and into the next stages of systems completion and commissioning. That progress gives the partnership a clearer path from agreement to operational capacity.

However, the larger significance lies in how the project combines data-center infrastructure with AI compute. Southeast Asia needs more than generic server space as GPU workloads increase power density and cooling demands. Facilities such as SMX01 can give compute providers the physical platform required to deploy these systems closer to customers. KONST’s agreement shows how that shift can create new partnerships between data-center developers and specialized AI infrastructure companies. For KONST, Jakarta now represents more than its first overseas compute deployment. It becomes a test of whether the company can replicate its integrated infrastructure model across new markets while maintaining the reliability and operational discipline required by AI customers. For SM+ and LG Sinar Mas, the partnership adds a dedicated compute provider to an AI-ready facility nearing completion. The combination could make SMX01 an important node in Indonesia’s expanding AI infrastructure landscape.

AI Infrastructure Moves Closer To The Customer

The KONST-SMX01 agreement reflects a broader change in the AI infrastructure market. GPU capacity increasingly needs to sit within data centers that can handle high power density, advanced cooling, resilient connectivity and sustained operations. That requirement is pushing compute companies to work more closely with specialized data-center operators. Jakarta provides KONST with an opportunity to build that model in a major Southeast Asian market. The immediate focus will remain on deployment and operational readiness as SMX01 moves toward its fourth-quarter 2026 target. KONST, SM+, LG Sinar Mas, Giga Computing and Raydian Cloud now have to translate their respective capabilities into a functioning high-density AI environment. The success of that process will determine how quickly the partnership can convert infrastructure capacity into commercial compute services. It will also provide an early indication of how KONST plans to scale its overseas footprint.

KONST’s move ultimately reflects a simple shift in the economics of AI infrastructure: compute needs to move closer to the markets that consume it. Jakarta gives the company a strategic base, while SMX01 provides the cooling, power and connectivity architecture needed for high-density deployments. Together, the partners are positioning the project around the next phase of AI demand rather than conventional data-center growth. For Southeast Asia, that makes the agreement another signal that the region’s AI infrastructure race is moving from plans and investment announcements toward physical compute capacity.