US AI chip export curbs spark concern in Thai tech and investment sectors

Although no official decision has been announced by the administration of Donald Trump, the mere speculation has already shaken confidence in Thailand’s ambition to become the “digital hub of ASEAN”.

Yupapin Wangviwat, Chief Financial Officer of Gulf Development (GULF), said the company had not yet been affected by the potential US policy. 

She confirmed that customer operations and the data centre business under its subsidiary, Gulf Siam AI Data Centre (GSA DC) — a joint venture with AIS and NVIDIA — were progressing as planned.

The first phase of operations is set to launch in mid-2025, with plans to expand to a second phase, totalling over 50 megawatts in service capacity, she said. Yupapin also noted that GULF is collaborating with Google Cloud to advance AI and cybersecurity capabilities.

Yupapin reiterated the company's target for 2025 revenue growth at 20-25% year-on-year, with its main income still from energy, but with a significantly rising share from new businesses such as data centres and digital infrastructure.

“So far, based on updates from our teams, we have not encountered any impact. However, we continue to closely monitor the situation,” she said.

“Despite geopolitical risks and Trump’s trade policy stance, we are pushing forward with our existing plans for the rest of the year. By positioning ourselves as a regional digital infrastructure provider, we are able to reduce external volatility.”

Too early to assess impact amid lack of clarity

A source in Thailand’s tech sector said it may be too soon to assess the full impact, as it remains unclear whether Trump will follow through with the restrictions, or what the exact conditions might be.

However, if Thailand only imports AI chips for domestic use and is not serving as an intermediary to China, it may avoid direct repercussions.

In the longer term, though, the source warned the policy could affect Thailand’s strategic image in the eyes of foreign investors — particularly as the country highlights its strengths in logistics, energy and macroeconomic stability as key investment draws.

If Thailand is seen as vulnerable to technology transfer to China, or lacking in sufficient export control mechanisms, major global tech firms may reconsider or delay investments, the source added.

Over the past two years, tech giants such as Microsoft, AWS and Google Cloud have invested in building data centres and cloud infrastructure in Thailand. But if the US enforces a rule requiring foreign data centre providers to obtain certification from Washington, Thailand’s digital infrastructure may face stricter scrutiny — reducing its investment appeal.

Limited direct impact

Some analysts believe the direct impact will remain limited for now, as Thailand’s data centres and cloud services are not aimed at transferring technology to China. Still, in terms of investor confidence and transparency, Thailand needs clearer policies and safeguards.

Thailand’s electronics sector is mostly involved in packaging and testing components, rather than directly importing high-end AI chips. Analysts say any significant impact would only arise if local firms are found to be indirectly involved in passing US-made chips to Chinese clients — a connection that remains difficult to trace.

Importing AI chips is not straightforward, as they are mostly made-to-order for specific applications. Orders often require production quotas from companies like NVIDIA, with months-long lead times, and must comply with US export regulations.

Thailand imports minimal volume of AI chips

Analysts noted that Thailand currently imports only a small volume of AI chips, primarily in the form of finished products such as smartphones and gaming consoles.

Nonetheless, this situation should serve as a wake-up call for both government and industry. Three key actions are recommended:

• Establish a national AI chip usage monitoring and control system;
• Deepen strategic cooperation with the US and its allies;
• Revisit and diversify the national digital and AI strategy to avoid overdependence on any single country.

AI chips come in various types. The most prominent in the market is the graphics processing unit (GPU), which is designed for parallel processing and widely used in training and running AI models, especially deep learning. These chips, such as the NVIDIA A100, are energy-intensive and expensive.

Another type is the tensor processing unit (TPU), a specialised chip developed by Google to accelerate TensorFlow and other machine learning workloads.

Meanwhile, central processing unit (CPU) — the standard processors used in general computers — are not optimised for AI workloads and operate much more slowly than GPUs or other specialised processors.