2026 was the year when Asia Pacific moved from AI experimentation to pragmatic adoption. 2027 will raise a more difficult question: As AI crosses the boundaries that enterprises have traditionally relied on, who controls it and who is accountable for what it does?

AI will embed itself more deeply in the region’s economies, enterprises, and daily life, even as the conditions for expanding it tighten. Geopolitical tension, trade restrictions, energy constraints, sovereignty requirements, and low consumer trust will all raise the cost of deploying AI and the consequences of getting it wrong. APAC leaders will have to make explicit choices about where AI runs, what it can reach, and when a person must stay in control.

The defining question therefore shifts from how fast APAC adopts AI to how governments and businesses establish control, accountability, and trust as AI crosses three boundaries: between global and sovereign, between personal and corporate, and between human and machine. Three of our predictions for 2027 sit on those boundaries.

One in five large, regulated APAC enterprises will approve Chinese open-weight models.

Minimum viable sovereignty is expanding from data residency to control of models and inference, and holding a model’s weights gives enterprises greater control over where and how it runs. So a self-hosted Chinese model can satisfy operational sovereignty requirements even when its origin raises separate geopolitical concerns. What will separate adopters from the rest is engineering maturity. Enterprises that have built their own context, harness, evaluation, and governance layers around the model can swap the model underneath without breaking production. Technology leaders should build those layers now, because they preserve model choice whichever way the geopolitics move, and should decide whether model provenance is a measurable security risk, a geopolitical constraint, or both.

One in three large APAC enterprises will discover a ten-to-one shadow agent problem.

Employees will build agents, software vendors will embed them, business units will fund them, and country teams will pick their own AI platforms, all faster than the enterprise can find, govern, or retire them. APAC’s fragmented technology estates and market-by-market sovereignty rules will multiply the problem. Each unmanaged agent carries stale instructions, permissions nobody reviewed, and dependencies nobody documented, which adds up to a new form of technical debt. CIOs should establish an agent rationalization program and require every production agent to have a unique identity, an accountable owner, a declared purpose, observable activity, and a defined retirement path.

Ten of APAC’s largest consumer-facing brands will differentiate on human assurance.

As AI becomes ubiquitous, leading brands will decide deliberately where a person stays visibly in charge: wherever a customer faces financial risk, emotional vulnerability, or a complex decision. A “human assured” promise becomes a trust signal that customers can see, rather than a fallback for when automation fails. CX and marketing leaders should identify the moments where human judgment, expertise, or empathy create the most value, then organize governance and service delivery around those moments. The strongest brands will present human involvement as an intentional feature of a trusted experience, not as automation falling short.

In 2027, APAC leaders who draw these boundaries explicitly will scale AI with more resilience and more trust. Those who leave them blurred will inherit rising costs, unmanaged risk, and customer distrust.

Forrester clients can read our full Predictions 2027: Asia Pacific report to explore all five predictions and the actions leaders should take. Set up a Forrester guidance session to discuss what these predictions mean for your organization and its 2027 strategy.

If you aren’t yet a Forrester client, download one of our complimentary Predictions guides and access additional resources on the Predictions 2027 hub.

Share