Edge AI: Why It Could Be the Next Strategic Frontier for Thai Enterprises

By Oravee Smithiphol, Tech Intelligence & Insights Manager, SCB 10X

 

 

Today's most capable AI systems largely depend on the cloud. While this architecture has powered the AI boom, it also creates practical limitations. Applications require internet connectivity, incur ongoing API costs, introduce latency, and may raise privacy or data sovereignty concerns.

A new generation of open-weight AI models is making it increasingly practical to run sophisticated AI directly on laptops, smartphones, industrial equipment, medical devices, and other edge devices. Rather than relying entirely on remote servers, organisations can increasingly choose where AI should run based on cost, performance, privacy, and operational requirements.

This does not mean cloud AI is disappearing. Instead, AI is becoming more flexible. Many organisations will adopt a hybrid approach, using cloud AI for compute-intensive tasks while deploying smaller models locally where speed, privacy, resilience, or cost matter most.

 

Why Edge AI Is Becoming Practical

Two developments are making edge AI increasingly viable.
The first is the rapid improvement of open-weight AI models.

Organisations such as Google, Meta, Mistral AI, Alibaba, and others now release powerful models that developers can download, fine-tune, and deploy themselves. This allows enterprises to adapt AI for their own languages, industries, and business processes instead of relying solely on proprietary cloud services.
 

 

The second is the rapid maturation of deployment tools. Platforms such as Ollama allow developers to run AI models locally with only a few commands. With nearly nine million users and adoption across approximately 85% of Fortune 500 companies, local AI deployment is rapidly entering the enterprise mainstream.

Together, these advances are shifting AI development from asking how large a model can be built to how much intelligence can be delivered efficiently on everyday devices.

 

Open Models Are Expanding Who Can Build AI

Open-weight models are lowering the barrier to AI development. Only a handful of organisations have the resources to train frontier foundation models from scratch, but many more countries, companies, and research organisations can build on existing models to address local needs.

The Typhoon family of Thai language models builds on open-weight foundations to improve Thai language understanding and support practical use cases for Thai users and enterprises. The project was recently featured by Google DeepMind as part of its Gemma ecosystem, demonstrating how locally developed AI can contribute to the broader global open-model community rather than simply consume it.

For Thailand, this illustrates an important opportunity: open models make it increasingly possible to build AI tailored to local languages, regulations, and business environments without developing a frontier model from scratch.
 

 

 

Edge AI Is Expanding Where AI Can Be Used

Running AI locally is not only about improving speed or reducing cloud costs. It also enables AI to operate where connectivity is unreliable, unavailable, or simply impractical.

In Uganda, Crane AI Labs has developed offline AI systems that support maternal healthcare in areas with limited internet access, allowing healthcare workers to access clinical guidance without continuous connectivity or API costs.

In Peru, developers are adapting open-weight models for Quechua, enabling schools and healthcare providers in remote communities to access AI-powered language tools despite limited internet infrastructure.

Agriculture presents another promising application. AI running directly on smartphones or field devices could help farmers identify crop diseases, assess plant health, and receive recommendations without depending on continuous internet access. For countries with large agricultural sectors, including Thailand, this could make advanced AI tools significantly more accessible.

 

Why This Matters for Thailand

Thailand's AI strategy has largely focused on cloud infrastructure, data centers, and GPU investment. These remain essential foundations for national AI capability. However, the next stage of AI adoption may depend just as much on deploying intelligence closer to where work takes place.

Manufacturers can embed AI into production equipment for quality inspection and predictive maintenance. Hospitals can process sensitive patient information locally to strengthen privacy. Financial institutions can deploy on-device AI to personalize customer experiences while reducing data exposure. Public agencies can build services that continue operating even when connectivity is limited.

These are not alternatives to cloud AI. They are complementary deployment models that allow organisations to balance performance, cost, privacy, and resilience.
For Thai enterprises, edge AI offers an opportunity not only to adopt AI more broadly but also to build solutions optimised for local industries, languages, and customer needs.

 

Oravee Smithiphol is Tech Intelligence & Insights Manager at SCB 10X, where she tracks emerging AI developments across startups, enterprises, and financial services organisations worldwide.

 

 

About SCB 10X

SCB 10X is the disruptive technology investment arm of SCBX Group. With an investment track record since 2016, SCB 10X has deployed over USD 500 million globally into startups in AI, blockchain, and fintech. SCB 10X has backed exceptional companies such as Together AI, Pagaya, Ripple, Fireblocks, Anchorage Digital

Beyond capital, SCB 10X partners with our portfolio founders to test, grow and scale their solutions through SCBX’s network, unlocking commercial opportunities into Thailand and Southeast Asia. Mandated as the group’s speedboat, we discover and ship state-of-the-art technologies and solutions into SCBX group. 
 
For more information, please visit https://scb10x.com/