Asia Pacific firms are transitioning from scaling to building autonomous AI agents

KPMG’s Q3 AI Pulse finds that Asia Pacific (ASPAC) companies are already benefiting from the meaningful business value that AI delivers towards greater productivity & efficiency, improved decision-making, reduced administrative burden, better customer experience and revenue growth.¹ Forty-two percent of ASPAC companies say that AI initiatives have delivered measured value in productivity & efficiency, followed by improved decision-making, reduced risk or adaptability.

About 7 in 10 (71 percent) companies in the region say AI will remain a top investment priority over the next 12 months, even if a recession hits. However, a slight dip in confidence is seen from the last quarter.

While confidence is high, it is slipping. 73 percent of ASPAC companies are confident of future-proofing their organizations’ AI trajectory, a 7 percent decrease from three months ago. This drop could mean that the early excitement and glow around AI is slowly fading as companies start facing the bills, in the form of power costs, talent availability, and shifting rules around data and chips across jurisdictions. Nevertheless, 34 percent in the region strongly agree that AI is already delivering meaningful business value in the form of productivity gains, cost savings and revenue growth. This is more than the global average in both the quarters, and the regional average in the last quarter. Within the region, India leads with 81 percent and is followed by China at 78 percent for companies who agree that AI is already delivering meaningful business value for their organizations.

More than a quarter (27 percent) of ASPAC companies have embedded AI into their daily workflows, encouraging its consistent use across their teams. This is more than the global average of 22 percent and last quarter’s ASPAC average of 20 percent. Among the 20 countries surveyed globally, more companies in Australia (34 percent) have moved beyond scaling and driving adoption of AI programs than elsewhere.

Data privacy concerns and cybersecurity continue to be the two greatest barriers for ASPAC businesses to successfully meet the goals of their AI strategy. In Q2, 39 percent of ASPAC firms were concerned with cybersecurity and 36 percent with data privacy. In Q3, data privacy and cybersecurity continue to worry about 4 in 10 ASPAC firms.

Even when demonstrating ROI related to AI, nearly half (47 percent) of ASPAC companies say that risk considerations such as data privacy and cybersecurity continue to pose the biggest challenge, 2 percentage points more than last quarter.

Over the next six months, risks such as those around data privacy and security, and macroeconomic factors such as GDP growth, inflation and tariffs continue to be the two biggest issues influencing ASPAC firms’ AI strategy.

This quarter, 42 percent of ASPAC companies say they will reevaluate their AI investment strategies given the total cost of AI ownership, more than last quarter’s 35 percent, but down from 52 percent in Q1 2026.

Over half of ASPAC companies understand the total cost of AI including infrastructure, tools and people-related costs, even understanding the cost of AI at the level of usage, such as per query. And 66 percent of companies in the region, compared with 59 percent globally, are using AI cost monitoring dashboards to manage AI usage costs. Meanwhile, other methods to manage AI usage costs include cost reviews, prompt design standards and token budgets.

Three quarters of ASPAC companies plan to invest more than US$50 million in AI over the next 12 months, in areas like training, technology, compliance and talent, compared with 69 percent globally. The share of companies planning to invest in AI under this bracket have increased since the last quarter.

More than 6 of 10 (63 percent) ASPAC companies are allocating their AI budget to IT infrastructure, particularly cloud and strengthening their cyber/data security (51 percent). In Q3, more ASPAC companies are spending on IT infrastructure than in the last quarter of 56 percent. 

Over the past 12 months, half of ASPAC companies realized an estimated financial value exceeding US$50 million from AI, in terms of cost savings, revenue generation, risk reduction and other measurable benefits, similar to the global average.

ASPAC companies are prioritizing human-AI collaboration and are actively redesigning roles and workflows to integrate AI into everyday work. Seventy percent of ASPAC companies, compared with 66 percent globally, are making good progress towards becoming a fully integrated AI-human workforce. Biggest share of this progress is seen among companies from China (78 percent), India and Singapore (77 percent each).

Companies, particularly from Australia (47 percent) and South Korea (45 percent), are prioritizing human-AI collaboration and fluency, building role-specific skills and redesigning processes so people and AI work effectively together.

Three quarters of ASPAC companies compared with 73 percent globally, say their employees are making strong and effective use of AI and are outperforming others in their organization. Last quarter this percentage was slightly higher — 79 percent regionally and 74 percent globally. Within the region there is a significant difference — 47 percent in India strongly agree to this vs only 11 percent in Japan.

However, talent continues to be a major hurdle. 39 percent of respondents in ASPAC say they are unable to use AI effectively due to insufficient staff expertise, compared to 35 percent globally. Meanwhile, one in four ASPAC and global companies say staff are resisting change to embed AI into everyday workflows and decision-making processes. Skills gap continues to worry 36 percent of ASPAC and 33 percent of global companies, hindering their ability to demonstrate ROI related to AI. In Q2, skills gap concerned 42 percent of ASPAC and 37 percent of global companies.

In response to this, ASPAC firms continue to redesign roles and workflows to integrate AI into everyday work (49 percent in Q3 vs 48 percent in Q2]. ASPAC firms are also upskilling/reskilling their current workforce and recruiting new roles for AI architects and prompt engineers (48 percent each). Sixty one percent of Indian firms, the most among all the countries surveyed, are redesigning roles and workflows to integrate AI into everyday work. 

About a third (29 percent) of ASPAC companies continue to focus on efficiency and revenue growth as part of their AI agents’ strategy. Last quarter 34 percent of ASPAC companies focused on this. Within the region, companies from Japan (36 percent) and Australia (29 percent) mostly focused on efficiency, with some exploration of new revenue opportunities; in India 39 percent of companies and in South Korea 33 percent focused equally on efficiency and revenue growth; while 29 percent of companies in China and 31 percent in Singapore, mostly focused on new revenue opportunities, with some efficiency gains.

Between Q2 and Q3, most companies have moved from scaling AI agents across multiple functions to building autonomous AI agents that can initiate actions and coordinate independently. Two-thirds (67 percent) of ASPAC companies vs 69 percent globally have deployed agentic AI in technology or IT, in operations (55 percent), and marketing and sales (46 percent). Within the region a staggering 81 percent of Chinese companies have deployed agentic AI in technology or IT, most among other surveyed countries.

Technical skill gaps and AI cost and economic literacy skills continue to be the biggest challenges ASPAC companies face in deploying AI agents to date. Both these challenges continue to impact over 4 in 10 companies in the region. 

Nevertheless, to ensure employees are ready to work with AI agents, more than 4 in 10 ASPAC companies are establishing clear protocols for when and how to use AI agents; developing role-specific guidelines for effective AI agent collaboration and delegation; and implementing “AI agent shadowing” programs where employees observe experts working with agents.

For risk mitigation efforts that organizations are putting in place with respect to AI agents over the next 6 to 12 months, about 4 in 10 (39 percent) ASPAC companies identified high-risk use cases where autonomous agent decision-making is not allowed. An equal percentage of companies are looking at AI agents as augmented support for their employees and providing training to their workforce.

There is a slight dip in owning of AI as a strategic business priority by ASPAC CEOs from the last quarter. But within the region, more than 8 in 10 CEOs in India and China actively own AI as a strategic business priority, more than other countries in ASPAC.

Among the options for who is to be held accountable for business decisions informed or executed using AI outputs, 36 percent (the highest response) ASPAC firms say that C-suite executives are to be held accountable. Last quarter, this figure was 26 percent.

Moving forward, companies in ASPAC will continue to integrate and invest in AI as a strategic business priority, ensuring human-AI collaboration and cybersecurity preparedness.