243 million Southeast Asians now use generative AI, but they use it far less than Americans: the region's AI paradox

How many people in Southeast Asia use AI? A lot more than most people realise, but not very intensively. Generative AI adoption in the region has reached 243 million people, or 58 percent of the working population, according to the e-Conomy SEA 2026 report by Google, Temasek and Bain and Company, as reported by The Straits Times. The technology reached that milestone in three years, compared with a decade for the internet.

But the same report found that rapid adoption has yet to translate into deeper usage. Measured by the amount of AI processing each person uses, Southeast Asians use about a tenth as much as Americans. At the same time, money is pouring into the region's AI infrastructure, with private funding for data centres hitting 16.3 billion dollars in just six months. That gap, between how many people use AI and how much they use it, is the region's AI paradox.

How fast AI spread in Southeast Asia

The adoption numbers are remarkable. Generative AI has reached 243 million people across Southeast Asia, 58 percent of the working population, in roughly three years. By comparison, the internet took around a decade to reach a similar share. Mobile first habits, young populations and widespread smartphone use helped AI tools spread quickly through messaging apps, search and productivity software.

Consumers are already using AI in everyday decisions. The same report found that 69 percent of consumers in six countries use AI to compare products when shopping online, and that sellers using AI tools see conversion gains of 10 to 15 percent. We covered the wider findings in our report on the e-Conomy SEA 2026 numbers.

The speed of adoption reflects how the region came online in the first place. Many Southeast Asians skipped desktop computers and went straight to smartphones, so new services spread through apps and messaging rather than through offices and corporate IT departments. AI tools arrived through the same channels, built into search, chat and shopping apps that people already used every day, which lowered the barrier to trying them.

Language is another factor that could shape how quickly usage deepens. Southeast Asia has dozens of major languages, and AI tools that work well in English may be less useful in Bahasa Indonesia, Thai, Vietnamese or Tagalog for complex tasks. Better local language performance could unlock deeper use in customer service, government services and education, which is one reason regional governments and companies are investing in local language models and data sets.

The usage gap with the US and Europe

Adoption is not the same as intensity. The report measured AI usage through tokens, the units of processing an AI model needs to understand a prompt and generate a response. In a study of six Southeast Asian countries, Indonesia, Malaysia, Singapore, Thailand, Vietnam and the Philippines, about 5 million tokens were used per person, roughly 10 percent of the United States' 60 million. Europe's figure was 14 million.

The gap suggests that most Southeast Asian users are still using AI for simple, occasional tasks, such as quick questions or translations, rather than embedding it deeply in work. Businesses in the region may be slower to integrate AI into complex workflows, and many users rely on free tools with lower usage limits.

Southeast Asia has won the race to try AI. The next race, using it deeply enough to change productivity, has barely started.

That gap is also an opportunity. If usage per person in Southeast Asia moved even halfway toward Europe's level, demand for AI computing in the region would multiply, which helps explain why investors are building data centres so aggressively.

Cost is part of the story. Paid AI subscriptions are priced mostly in US dollars, and for many workers in Indonesia, Vietnam or the Philippines a monthly plan represents a far larger share of income than it does in the United States. Free tiers come with tighter limits, which naturally caps how many tokens each person uses. As prices fall and local providers offer cheaper plans, usage per person could rise quickly even without new users.

The money flooding into AI infrastructure

Funding has jumped sharply. Private funding for data centres in the first half of 2026 reached 16.3 billion dollars, 16 times the 1 billion dollars recorded a year earlier. The total was driven by two large deals: an 11 billion dollar transaction involving ST Telemedia Global Data Centres and a 5 billion dollar deal involving DayOne Data Centers.

Venture money is shifting too. Of the 4 billion dollars in private funding for digital companies in the first half of 2026, 80 percent, or 3.2 billion dollars, went to AI applications. A year earlier, AI applications took only 30 percent, or 900 million dollars, of 3 billion dollars in digital funding.

The infrastructure boom has consequences. Malaysia's Johor is facing a power crunch from data centre demand, as we covered in our report on Johor's data centres and the grid, and Malaysia's prime minister has announced a 50 billion dollar AI and data centre investment from an unnamed global investor. The key forces shaping Southeast Asia's AI economy include:

• Mass adoption of AI tools, with 243 million users
• Low usage intensity compared with the US and Europe
• Record data centre investment, led by mega deals
• Venture funding shifting heavily toward AI applications
• Energy, water and grid constraints on data centre growth

Singapore, Johor and Batam form the region's main data centre corridor, but investors are also looking at Thailand, Vietnam and Indonesia's other islands as power and land constraints tighten. The location of new capacity will depend on access to electricity, water, undersea cables and government incentives, and on whether local communities accept the facilities.

What it means for businesses and workers

For businesses, the usage gap is the most important number. Companies that move beyond basic AI use to integrate it into customer service, operations, software development and analysis could gain a significant advantage in markets where competitors are still experimenting.

For workers, widespread adoption means AI skills are becoming baseline expectations. The challenge for governments and employers is training people to use AI productively, not just casually, so that the region captures the productivity gains rather than simply consuming AI services built elsewhere.

For investors, the picture is a classic early stage market: broad adoption, shallow usage and heavy infrastructure spending. Whether that spending pays off depends on how quickly usage deepens.

Southeast Asia has embraced generative AI at record speed, but the depth of use still lags far behind the US and Europe. Investors are betting that the gap will close, pouring billions into data centres and AI applications. Whether those bets pay off depends on turning 243 million casual users into a region that uses AI to work, build and compete.