Relay Weekly · No.11

APAC leaders on AI at the UN, Infineon’s new Thailand plant, and understanding AI “brain fry”

A busy week across AI, trade and cybersecurity, with AI partnerships in Thailand and Singapore.

Shane Neubronner · 16 October 2026

Subscribe to this newsletter or send story tips to hello@relay-asia.com.

This Week's Analysis

Asia-Pacific brought five different AI governance ideas to the UN

Welcome to Relay Weekly No. 11.

It has been a week with plenty to unpack across Asia-Pacific’s technology landscape. From the IMF’s warning that global trade is leaning heavily on the AI boom, to Singapore’s push on cybersecurity and AI safety, new chip facilities in Thailand, and efforts to bring AI tools to workers and small businesses, the developments span infrastructure, policy and everyday use.

That makes this week’s Analysis a good place to start. At the United Nations General Assembly in late September, six Asia-Pacific governments set out their visions for artificial intelligence. Their priorities differed, from Singapore’s emphasis on safeguards and human accountability to calls for national AI capabilities, broader access to AI’s benefits and a greater voice in setting the rules.

The speeches offer a useful snapshot of how the region’s leaders see their countries fitting into AI’s future, and what they want to influence as that future takes shape. This week’s Analysis examines what they said, and where their visions diverge.

Read on Relay Asia →

Short Reads

  1. Infineon opened a Thailand site for chip assembly and testing

    Infineon opened the first module of its new site in Samut Prakan, Thailand, on 1 October 2026. The facility tests wafers and assembles power modules, handling later stages of chip production rather than manufacturing the wafers themselves. Around 350 people work there today, with the company expecting that number to reach 1,000 as operations expand. Infineon has plans for five modules, which could make the site one of its largest backend locations. This Short Read separates what is already operating from what remains planned.

    Read on Relay Asia →
  2. Building with AI can tire the person who has to watch it

    AI can take work off your hands, but it can also give you a new job: checking what it has done. A study of 1,488 full-time workers at large US companies, published by Boston Consulting Group researchers in Harvard Business Review on 5 March 2026, found that 14% of employees using AI at work had experienced this kind of fatigue. The strain was particularly evident among people monitoring AI output and managing multiple tools. There is no equivalent public figure for workplaces across Asia-Pacific, so the US finding should not be treated as a regional measure. This Short Read examines the oversight burden, the lack of comparable regional data, and a measurement scale developed by researchers in Singapore.

    Read on Relay Asia →

Suggested Reads

  • The AI boom is carrying trade, and the IMF says the cushion is thinning. The IMF was in Bangkok this week with a slightly brighter growth forecast, but not an entirely reassuring one. CNBC reports that the Fund expects the global economy to grow by 3.1% in 2026, up from July’s 3% projection, before reaching 3.4% in 2027. Both remain below last year’s 3.5%. The more revealing figures concern trade. Reuters reports that IMF chief economist Silvana Tenreyro said global trade grew 8.3% between 2023 and 2025, while trade in AI grew 29%. Without AI-related trade, she said, overall trade growth would have been 4.6%. China, India and other Asian economies have benefited. The AI boom is doing a remarkable amount of work in the global numbers. It also makes the picture look healthier than it might if the benefits were spread more evenly.
  • PM Lawrence Wong is worried about the bottom rung of the career ladder. At Monday’s DBS Asia Future Forward conference in Singapore, interviewer Tan Su Shan, the CEO of DBS Bank, joked that Prime Minister Lawrence Wong had forgotten his guitar. The conversation soon turned to a less entertaining prospect: what happens to young workers if AI takes over the routine tasks through which people traditionally learn a job. In the Prime Minister’s Office transcript, Wong says that much of the learning for new hires happens through this everyday work. “We are concerned that there may be that bottom rung of the career ladder that is broken.” Singapore is examining the issue with industry and its tripartite partners, but has yet to announce specific programmes. On the same stage, Broadcom chief executive Hock Tan identified a different constraint on AI infrastructure: capital, which he put ahead of chips, memory or electricity. He also said demand for compute is running ahead of what the industry can supply, and that this imbalance could persist through 2029 and potentially into 2030.
  • Google is developing an on-device scam warning, rollout details to come. Singapore International Cyber Week brought two Google announcements. One concerns cyber defenders using Gemini 4 Argon and CodeMender within their own cloud environments through the Fairwind programme. The other could eventually reach a much wider audience. Google DeepMind and GovTech Singapore have adapted a lightweight Gemma model to recognise local scam narratives, with the aim of warning Android users without sending their conversations off the device. The blog says the rollout will be phased and will require user consent. Details on how people can install or access it are still to come.
  • Anthropic opens a Singapore office and signs a memorandum. Anthropic opened an office in Singapore, its fifth in Asia-Pacific, this week and signed a memorandum of understanding with the Ministry of Digital Development and Information. The agreement covers AI safety and security, model evaluation, research into AI adoption and its impact, and responsible use across the public and private sectors. The memorandum was signed by permanent secretary Kai Fong Chng and Anthropic co-founder Ben Mann during Singapore International Cyber Week, with Josephine Teo and Michael Sellitto witnessing. Anthropic says Claude is already used in services involving GovTech, Grab and DBS. The new office in Singapore gives the company a more substantial regional presence, while the memorandum establishes areas for cooperation.
  • 1.2 billion people in developing Asia-Pacific are still offline. GSMA Intelligence, in a report commissioned by the Asian Development Bank and released in Kuala Lumpur on 14 October 2026, estimates that 1.2 billion of the 4.3 billion people in developing Asia-Pacific remain offline. That represents about 46% of the world’s unconnected population. The striking detail is that 98% of people in developing Asia-Pacific live within reach of a 4G network. Around 1.1 billion people who could access mobile broadband still do not use it. The barriers include affordability, digital skills, safety and whether online content is relevant to people’s lives. The report also estimates a USD 184 billion investment gap. These are notable numbers as governments and businesses talk about AI reaching everyone.
  • An AI-assisted vocal is No.3 on Melon. The song "The Adult I Dreamed Of", by Meomuru, reached No.3 on South Korea’s Melon Top 100 on 14 October 2026, according to The Korea Herald. Released in June, the song gained traction through short-form videos. The voice was generated using Suno, according to the album description, while Meomuru wrote the lyrics and directed the production. So the chart position belongs to a song made with AI-generated vocals, rather than a conventional singer’s performance. While listeners seem happy to put the song on repeat, how the industry handles authorship, credit and disclosure is still catching up.
  • Beijing wants the boom, and it wants it controllable. China’s latest technology guidelines point in two directions at once. Reuters, citing Xinhua, reports that guidelines issued by the Communist Party and the State Council on 9 October 2026 call for advances in emerging technologies, including AI, while warning against investment bubbles. The plans cover computing power, algorithms, data and industrial pilot bases. AI development, meanwhile, is to remain “safe, reliable and controllable”. China seemingly wants the infrastructure and industrial gains from the AI boom, but not a rush of investment that becomes a problem of its own.
  • Grab and OpenAI take AI training to drivers and merchants in Thailand. Grab and OpenAI have launched “GO Forward with AI” in Thailand, offering ChatGPT courses through GrabAcademy for drivers, delivery riders and merchants to build confidence and practical AI skills. The initial target is more than 3,000 participants, within a regional partnership aiming to train over 30,000 people. Grab Thailand country head Chantsuda Thananitayaudom says a merchant assistant introduced in 2025 is used by more than 275,000 merchants, and that a driver assistant launched in September using OpenAI technology has been adopted by 120,000 drivers.

Each Friday we try to filter Asia's technology, business and policy stories into what changed, what it means and what to watch. If this edition helped, forward it to a colleague or friend. And if you spot something we should dig into, reply or write to hello@relay-asia.com.

← All Weekly

APAC leaders on AI at the UN, Infineon’s new Thailand plant, and understanding AI “brain fry” · Relay Asia