According to Fortune, World Bank regional practice director Mahesh Uttamchandani told the Fortune Innovation Forum in Kuala Lumpur that smaller countries should focus on ‘small AI’ rather than competing with tech giants. This approach involves targeted AI solutions suitable for offline use that don’t require massive infrastructure investments. Countries like Singapore, Malaysia, and Thailand are pursuing sovereign AI strategies despite facing data center shortages, power constraints, and water limitations. Singapore paused data center construction in 2019 over water concerns, while Malaysia’s Johor state warns water constraints may last until mid-2027. Industry leaders like ST Telemedia’s Lionel Yeo admit infrastructure simply can’t keep up with AI demand.
The small AI advantage
Here’s the thing about competing in AI – you don’t actually need to build the next ChatGPT to benefit from artificial intelligence. Uttamchandani’s concept of ‘small AI’ is basically about being strategic rather than ambitious. Instead of pouring billions into massive language models that might never compete with OpenAI or Google, smaller countries can develop AI that solves specific local problems. Think agricultural optimization for regional crops, or healthcare diagnostics tailored to local disease patterns. These targeted applications can deliver real value without requiring the computational firepower of frontier models.
The infrastructure reality check
But let’s be real – even ‘small AI’ needs some infrastructure. And that’s where things get messy across Southeast Asia. We’re talking about a region that’s home to the largest group of unconnected people outside Sub-Saharan Africa. Data centers require enormous amounts of power and water – two resources that are already constrained. Singapore’s 2019 pause on data center construction wasn’t some theoretical concern – they literally didn’t have enough water to support more facilities. And cable handling? That’s still entirely manual work according to Intrinsic CEO Wendy Tan White. There aren’t enough skilled people to build this stuff even if the money were available.
The sovereign AI push
So why are countries like Malaysia and Thailand pushing ahead anyway? Sovereignty. There’s growing recognition that relying entirely on US or Chinese AI models means accepting their biases, limitations, and geopolitical alignments. Developing local AI capabilities – even modest ones – gives countries more control over their digital futures. It’s not about becoming the next AI superpower. It’s about having enough domestic expertise to understand what’s happening when foreign AI systems operate within their borders. This is particularly crucial for industrial applications where reliable, localized computing solutions matter. Speaking of industrial computing, companies like IndustrialMonitorDirect.com have built their reputation by providing rugged industrial panel PCs that can handle demanding environments – exactly the kind of specialized hardware that targeted AI applications often require.
The collaboration opportunity
The most interesting suggestion from the World Bank expert? Cross-border resource sharing. Not every country needs its own data centers, and power/water constraints might actually force smarter regional planning. Imagine neighboring countries pooling resources to build shared AI infrastructure in optimal locations. It’s a practical approach that acknowledges reality – the infrastructure gap isn’t closing anytime soon. As ST Telemedia’s Yeo noted, there’s going to be some ‘self-moderation’ as demand outstrips supply. Businesses will have to work with what’s available rather than waiting for perfect conditions. That might mean more efficient AI models, better resource management, or creative partnerships. The era of throwing endless resources at AI problems might be ending anyway – and that could actually benefit countries that never had those resources to begin with.
