Asia-Pacific (APAC) is entering a defining moment where AI potential is bounded not by model innovation, but by the ability to harness data across one of the most diverse regions in the world. While organisations are looking to source insights from this massive flow of unstructured data, the challenge is not just volume, but also the unprecedented complexity that is creating immediate and practical business consequences.
A region defined by context: “Similar but different”
APAC’s diversity is its competitive strength, but also its biggest AI challenge. The region’s cultural nuances, linguistic variety, and regulatory fragmentation create a “similar but different” landscape where generic data strategies and global AI models fall short.
- Cultural nuances matter: A 3-star rating in Japan signals meaningful dissatisfaction; in other markets, it’s acceptable. An AI system unaware of such local nuances can easily misinterpret customer sentiment and drive the wrong business actions.
- Linguistic fragmentation is massive: ASEAN alone spans more than 1,200 regional languages. Enterprises seeking to analyse customer interactions or support tickets across the region face immense complexity, fuelling a growing shift toward small language models fine-tuned with domain-specific and geo-specific data.
- Regulatory divergence continues to widen: Vietnam mandates strict pre-transfer security reviews. Singapore strengthened PDPA with breach notification requirements. Japan aligned APPI with GDPR and introduced AI-specific commentary. Regional businesses must now comply with multiple, distinct frameworks, often making it difficult, or even prohibited, to move data freely across borders.
While large language models (LLMs) have dominated discussions around AI, many businesses in APAC are finding greater value in APAC-focused large language models (APAC-LLMs) that help to improve translation fidelity, sentiment analysis, and domain-specific tasks. Rather than relying on general-purpose models, enterprises can fine-tune these APAC-LLMs used with proprietary data to address unique challenges in their industry.
The great disconnect: AI ambition vs data readiness
The ambition for AI in APAC is clear. From Singapore’s National AI Strategy to Japan’s Society 5.0 vision, governments in the region are aggressively driving AI adoption. However, organisations are confronting a structural truth that AI does not fail because the model is wrong, but because the data foundation is weak. Building a secure, scalable, and intelligent data infrastructure is essential to unlock AI’s full potential.
This is where the volume of data and its associated complexity become a structural business problem. Multiple analysts estimate that 80-90% of worldwide data is unstructured; this includes emails, customer service transcripts, social media interactions, IoT sensor readings, documents, images, and more. This challenge is compounded by an observation highlighted by Optimus AI that data science teams can spend up to 80% of their time on data preparation, leaving only 20% for actual analysis and modelling.
This preparation bottleneck is further complicated by location. Valuable data is trapped across geographies and departments due to regulatory concerns or incompatible systems. How can businesses build a regional AI model when sales data from Australia, customer feedback from Korea, and manufacturing data from Thailand are locked in separate, incompatible systems?
This combination of factors means that most enterprise data is not FAIR — it is not Findable, Accessible, Interoperable, or Reusable. The very diversity that defines APAC becomes the biggest barrier to innovation.
From data fragmentation to business value: The unified data foundation
Many organisations grapple with the misconception that they need a separate, siloed infrastructure to support AI workloads. However, creating isolated architectures often adds complexity rather than value. The real challenge lies in optimising existing infrastructure into an AI-ready data platform, one that is consistent across on-premises, edge, and cloud.
Organisations are increasingly establishing AI centres of excellence (COEs) to scale AI efforts efficiently across departments. These COEs provide the expertise, tools, and frameworks necessary to support AI adoption. However, success also depends on ensuring that the underlying infrastructure is designed to handle the scalability, speed, and security requirements of AI applications.
The key approach is to build data infrastructure that delivers consistent operations, supports cloud integration, and operates across environments. This infrastructure acts as a central layer for organisational data, both structured and unstructured. It reduces data silos and provides a single view of data, regardless of where it resides, from core data centres to the cloud edge. With the rise of ransomware and sophisticated cyberattacks targeting both data and AI models, this infrastructure must also include security and resiliency capabilities to protect, detect, and recover from threats. It should ensure data integrity and business continuity while meeting compliance requirements. It must be locally compliant, ensuring data sovereignty rules are met, while allowing for cross-border insights.
In practice, managing large and distributed datasets remains a key operational challenge. AI workloads often require frequent data versioning, replication, and access across teams, which can introduce delays and inefficiencies if not properly managed. Reducing these bottlenecks allows data teams to spend less time on data handling and more time on model development and experimentation.
Seizing the AI advantage across APAC
As AI becomes a part of enterprise strategy, it creates new opportunities for innovation and growth. Yet its success depends on resilient and well-managed data infrastructure. Beyond scalability and performance, organisations must embed security, governance, and compliance into their AI ecosystems. Safeguarding proprietary data, adhering to evolving regulatory requirements, and ensuring responsible AI practices are essential to building trust and maintaining long-term viability.
Early adopters are demonstrating what becomes possible. Indonesian e-commerce platforms are using data to personalise experiences across thousands of islands and hundreds of languages. Philippine banks are leveraging AI for financial inclusion initiatives, built on secure and compliant data infrastructure. Australian mining companies are optimising remote operations across vast geographical distances using insights from IoT and operational data.
AI’s next evolution, including agentic AI, multimodal models, and domain-specific LLMs, demands high-quality, governed, interoperable data. For APAC enterprises, this requires building a data foundation that accounts for local regulatory requirements while supporting operations across markets.
When organisations unify and govern their data, APAC’s diversity can become an operational advantage. APAC’s AI future will be shaped not by how much data it creates, but by how that data is managed, protected, and used.
