Alejandro Reyes at the GBA Roundtable: Building a New Paradigm for AI Governance in Shenzhen on November 7, 2026
Given the imminent 33rd APEC Economic Leaders' Meeting in Shenzhen this November, all 21 APEC economies are anticipating a more prosperous and robust Asia-Pacific region. AI is expected to be a hot topic among government officials, business leaders and scholars during APEC week.
"A common AI law for 21 very different economies is unrealistic. What APEC can achieve is interoperability: common procedures, compatible standards and practical cooperation that allow economies to work together while each keeps its own rules," noted Alejandro Reyes, Senior Fellow of the Center on Contemporary China and the World at the University of Hong Kong, in an exclusive interview recently.
AI gap between advanced and developing economies in APEC
The AI gap among APEC members, in Reyes' eyes, is across the whole stack. AI hinges on electricity, chips, computing power, data centers, models, applications and above all, people with the skills to use it.
The leading AI powers dominate the most expensive layers. Many developing economies remain dependent on them. The cost of building frontier AI capacity is rising, even as the cost of accessing and using AI models is falling.
"That gives developing economies more opportunities, but not necessarily greater autonomy," Reyes warned.
His biggest angst lies in adopting AI. Can a small manufacturer in Vietnam or a public hospital in Indonesia use AI productively, in its own language, at a price it can afford?
"Access to technology does not by itself give a country the capacity to put it to productive use. Almost no country can build every layer itself," Reyes noted.
What countries can do is collaborate to achieve what they cannot accomplish on their own, which entails shared computing, open models for Asian languages, training for officials and workers and financing that makes digital infrastructure bankable.
Fortunately, there are some silver linings. As Reyes noted, open-weight and open-source models, many of them from China, have given developers in developing economies far more choice than they had two years ago.
AI Singapore, a national program, developed the SEA-LION models for Southeast Asian languages. Indonesian partners worked with it to develop Sahabat-AI for Indonesian and other local languages.
In October this year, the APEC Finance Ministers' Meeting in Hong Kong SAR endorsed voluntary approaches to making digital infrastructure investable, including risk-sharing arrangements, first-loss guarantees and financing involving development banks and long-term investors.
The real success of data centers
According to the AI Index 2026 published by Stanford HAI, the United States hosts over 5,000 data centers, more than ten times any other country, and consumes more energy than any other region.
Indeed, Reyes admitted that data centers are the factories of the AI economy. Where they are built and who controls them are now part of a country's economic sovereignty. However, they use a great deal of electricity and often require water, and the question for the Asia-Pacific region is how to build them without overloading grids or communities.
He set the example of Johor, in southern Malaysia, where foreign firms build data centers due to its land, power and connectivity, including access to undersea cables. Nonetheless, residents have complained about strained power and water, and the government has then tightened approvals.
Reyes proposed that the first step is to plan data centers and power grids together, as APEC energy ministers recommended in September. Malaysia is considering asking large projects to provide more of their own power and water.
He also highlighted the location and detailed that China's "East Data, West Computing" program moves some computing workloads to western provinces with more land and greater access to renewable energy.
Financing is part of the answer as well, as Reyes pointed out. The approaches endorsed by APEC finance ministers in Hong Kong could help smaller economies afford efficient facilities rather than settling for the cheapest ones.
Additionally, he suggested that communities and local governments need to be involved early, as public trust has to be built from the start. The environmental costs cannot simply be passed on to households and businesses through higher electricity prices or greater pressure on water supplies.
Counts of data centers vary by source. What matters more is capacity and power use. For developing economies, the objective should be more than building more facilities, he stated.
"The real measure of success is whether data centers create value for the local economy without imposing unacceptable costs on communities and the environment. Hosting AI infrastructure is only the first step towards building an AI economy," Reyes added.
Outcomes on AI expected in 33rd APEC Economic Leaders' Meeting in Shenzhen
For all the high expectations for the approaching 33rd APEC Economic Leaders' Meeting in Shenzhen, Reyes hoped that it could focus on AI adoption, digital infrastructure, skills and trust.
"In a sense, there is a regional roadmap already," he claimed.
The APEC AI Initiative runs from 2026 to 2030 with annual reviews. The 2026 APEC Digital and AI Ministerial Meeting in Chengdu this July added a cooperation agenda on connectivity, adoption, skills and trust, while APEC energy ministers adopted recommendations on AI and energy in September.
Also, safety collaboration remains relatively thin, as the Chief Strategy Officer of AI Safety Asia observed. Reyes proposed two practical steps, namely, common approaches to evaluating AI risks and a standing network of contacts for reporting and responding to serious AI incidents.
China's Global AI Governance Action Plan calls for mutual recognition platforms for safety assessments, information sharing and emergency response mechanisms. China's National Development and Reform Commission has announced plans to accelerate AI legislation and encourage developers to report safety incidents and vulnerabilities.
"The challenge now is to translate these aspirations into credible practices," he added.
The country has said that AI and digital economy outcome documents are being prepared, building on the 2026 APEC Digital and AI Ministerial Meeting on connectivity, adoption by smaller firms, digital literacy, trust and investment.
"The test will be whether a smaller economy, business or public agency can do something next year that it cannot do today. That is what will make a difference to developing economies," he projected.
Reporter | Zhang Ruijun
Photo | Zhang Ruijun
