For the past two years, much of the artificial intelligence conversation has focused on models, applications and use cases. Enterprises have asked which large language model to use, which workflow to automate or which chatbot to deploy. Those questions still matter, but they are no longer enough.
AI is increasingly becoming an infrastructure story as much as a software story. Moving from pilots to production places pressure on data centers, cloud architecture, chips, cooling systems, cybersecurity, network latency, energy availability and regulatory compliance. A proof of concept can run on borrowed capacity. A mission-critical AI system cannot.
Enterprise AI readiness therefore extends beyond compute. It includes the cloud environments where workloads operate, the integration layers connecting AI with existing systems, governance over how data is accessed and used, and the operational resilience required to keep digital services running.
This becomes more important as generative AI gives way to increasingly agentic systems. Rather than only generating answers, these systems can take actions, trigger workflows and interact with enterprise systems. Gartner has projected that by 2028, one-third of interactions with generative AI services will use action models and autonomous agents for task completion. When AI becomes embedded in payments, lending, logistics, public services or enterprise operations, downtime, latency and security gaps become business risks rather than isolated technical issues.
Infrastructure becomes a competitive issue
The scale of AI-related infrastructure demand is already visible. The International Energy Agency projects global data center electricity consumption will more than double to around 945 TWh by 2030, representing just under 3 percent of global electricity use. From 2024 to 2030, it expects data center electricity consumption to grow by around 15 percent annually.
This is not only a hyperscaler issue. Enterprises in banking, insurance, telecommunications, healthcare, manufacturing, retail and logistics are integrating AI into core processes. IDC forecasts that AI and generative AI investment across Asia Pacific will reach $175 billion by 2028 as organizations move from experiments toward enterprise-scale deployment.
Southeast Asia faces an additional layer of complexity. The e-Conomy SEA 2025 report from Google, Temasek and Bain & Company projected that the region’s digital economy would surpass $300 billion in gross merchandise value in 2025. It also said more than 4,600 MW of new data center capacity was planned across the region.
As digital transactions, financial services, mobility and cloud-native businesses grow, AI will increasingly operate on systems already expected to perform at speed and scale. This puts cloud, edge infrastructure, data centers and network design at the center of AI strategy. Companies need to determine which workloads belong in centralized cloud environments, regional data centers or closer to users and transaction points.
Southeast Asia requires local infrastructure choices
Southeast Asia is not a uniform market. Countries differ in geography, regulation, connectivity, energy systems and enterprise maturity. Singapore, for example, has said through its Green Data Centre Roadmap that it aims to provide at least 300 MW of additional data center capacity in the near term while pushing for higher energy efficiency and greener power.
Indonesia faces its own combination of scale, connectivity and enterprise demand. LG Sinar Mas is the design and operation partner for SMX01, an AI-ready data center in Jakarta’s central business district. Official project materials say the facility is designed for high-density racks of up to 130 kW, advanced liquid cooling, low power usage effectiveness and multiple carrier fiber paths. The project is targeting Ready for Service in the fourth quarter of 2026.
The broader issue is that AI adoption depends not only on ambition, but on where compute is available, how efficiently it can be powered and cooled, and how reliably it is connected. These constraints become more important when organizations attach AI to customer, transaction and operational systems.
Regulated industries raise the requirements
Banks, insurers, telecommunications providers, healthcare organizations and public institutions must integrate AI with core platforms, customer and transaction data, cybersecurity controls and compliance processes. A bank using AI for fraud detection, for example, needs secure data access, identity and access management, auditable decision trails and clear rules for where data is processed and stored.
As organizations operate across public cloud, private cloud, hybrid environments and on-premise infrastructure, they also need deliberate rules for workload placement, data residency, security, cost management, vendor dependencies and disaster recovery. The task is therefore broader than deploying isolated AI use cases. Enterprises need to modernize the foundation around them through cloud migration, system integration, API management, data governance, cybersecurity and business continuity planning.
Energy, security and governance shape AI readiness
AI infrastructure is capital-intensive, but sustainability and operational resilience are equally important. High-density workloads generate more heat and require more advanced cooling, encouraging liquid cooling, improved thermal management, higher-efficiency chips and more efficient data center design. Power availability, redundancy, data center location, cooling technology and workload management increasingly need to be considered together.
In our work at LGSM, enterprises may arrive with a clear AI use case but an incomplete understanding of the infrastructure dependencies beneath it, including data platforms, integration layers, governance controls and compute requirements. Those gaps can surface later as fragmented data architecture, unpredictable cloud costs, weak security controls or infrastructure that cannot meet production requirements.
Agentic AI raises the security bar further. An AI agent connected to enterprise systems may access customer data, payment workflows, documents, APIs and operational tools. Without appropriate governance, this can create new attack surfaces or decisions that are difficult to audit.
Governance therefore cannot be a policy document added after deployment. It needs to be embedded into architecture through identity and access management, audit trails, data classification, model monitoring, human escalation, incident response and compliance reporting.
Build the foundation before scaling the model
Before scaling AI, enterprises should assess whether their data is accessible and governed, where workloads should run, whether legacy systems can integrate securely, and whether AI can connect to customer platforms and operational systems without creating new risks. They also need governance over cloud costs, access rights, data residency, audit trails, latency, cybersecurity, disaster recovery and business continuity.
Models will improve and applications will become easier to build. Operating AI reliably, securely and sustainably at scale will remain difficult. In Southeast Asia, companies that succeed will likely combine innovation with infrastructure discipline. Enterprise AI will depend not only on models or use cases, but on the full stack beneath them: resilient data centers, efficient cloud architecture, secure networks, robust governance, reliable power, sustainable cooling and infrastructure capable of supporting real-world business continuity.
Han Donghyup is CEO and President Director of LG Sinar Mas (LGSM), a joint venture between LG CNS and SM+, the digital and data center arm of Sinar Mas. He has more than three decades of technology leadership experience, including roles supporting LG CNS’s expansion across high-growth Asian markets. He leads LGSM’s work across data centers, cloud, IT modernization and digital transformation in Indonesia.
Editor’s note: This contributed article has been lightly edited for clarity, length, and style. Where appropriate, TNGlobal may verify, qualify or omit factual claims that cannot be independently corroborated. The views and arguments expressed remain those of the author.
Share your perspective: TNGlobal welcomes contributed insights and expert commentary from across Asia’s technology and innovation ecosystem. Submit a contribution for editorial consideration, or explore more conversations in our TNGlobal INSIDER and TNGlobal Q&A and Interviews archive.
Sinar Mas, LG CNS form joint venture to provide digital solutions in Indonesia