Rick Villars, Global Vice President of IDC: In 2026, the growth of AI spending in China and the United States will have different focuses.

The corporate intelligence expert "Finance Graffiti" learned that on December 2nd, Rick Villars, the vice president of the IDC Global Research Group, delivered a keynote speech titled "Towards a New Era of Intelligence: Three Driving Forces Reshaping the Global IT Industry" at the "IDC FutureScape 2026: China ICT Market Forecast Forum".

He pointed out that the global technology industry is entering an expansion era. In 2027, the total expenditure in the server and storage market will exceed $700 billion, and software expenditure will exceed $1.67 billion. However, only 13.6% of North American enterprises and 2.4% of Asia-Pacific enterprises can obtain measurable benefits from most AI projects, highlighting that value realization remains a key challenge.

"To break through the next barrier of AI application, enterprises need to formulate an enterprise-level AI strategy, build an AI-ready workforce, and construct an AI-ready technology stack," Rick Villars pointed out.

Overall, China and the United States are driving the technological progress of AI. Among them, the United States is an important driving force for AI expansion, mainly driven by AI infrastructure and software, while China is more driven by AI infrastructure.

Breaking through the next barrier of AI application includes three directions

In terms of AI expenditure, it is expected that the growth focuses of China and the United States will differ in 2026.

Specifically, the United States will focus on building artificial intelligence agents to automate business processes, enhance network recovery and resilience, and modernize the infrastructure of enterprise data centers, with expected growth rates of 48%, 33%, and 31% respectively.

China will pay more attention to the modernization of core enterprise applications, migrating applications from public infrastructure to local infrastructure, and migrating applications from local to the cloud, with expected growth rates of 39%, 36%, and 34% respectively.

In terms of benefits, only 13.6% of North American enterprises can obtain measurable benefits from more than 75% of AI projects. In the past two years, the average proportion of AI projects that produced measurable results was 47%.

Among the major challenges hindering enterprises from fully realizing the value of AI investment, resource competition between AI and other IT and digital programs accounts for 36%, resistance to process changes required for AI integration accounts for 33%, 29% is due to regulatory uncertainties affecting AI investment decisions, and the difficulty of quantifying and demonstrating the return on investment (ROI) of AI to stakeholders accounts for 28%.

Compared with North America, only 2.4% of Asia-Pacific enterprises can obtain measurable benefits from more than 75% of AI projects, and the average proportion of AI projects that produced measurable results in the past two years was 38%.

The factors hindering AI value investment in enterprises in this region are slightly different. Among them, 31% is due to unclear ownership and accountability of AI results, 30% is due to the difficulty of quantifying and demonstrating the return on investment (ROI) of AI to stakeholders, 30% is due to resource competition between AI and other IT and digital programs, and 30% is due to resistance to process changes required for AI integration.

Rick believes that breaking through the next barrier of AI application will include three directions. First, formulate an enterprise-level strategy to identify core business areas and prioritize transformation; second, build an AI-ready workforce and plan and promote necessary change management; third, construct an AI-ready technology stack and optimize the technical architecture to support AI and agent workflows.

"In the future, CIOs are responsible for the highest proportion of corporate AI transformation, at 46%. While Chief AI Officers and CEOs account for only about 18% and 16% respectively," he pointed out. In 2026, corporate priorities will include identifying major business areas that need transformation, coordinating investments and roadmaps to ensure alignment with the strategy, and establishing a core team to coordinate various initiatives across the company.

The total number of active agents will exceed 1 billion in 2029

"Agent workflows are reshaping the employee lifecycle, and enterprises need to rethink future work models," Rick pointed out. By 2026, 40% of jobs will work in collaboration with AI agents, which will redefine long-existing traditional entry-level, mid-level, and senior positions.

By 2027, the usage of agents by the Global 2000 enterprises will increase tenfold, and the call load will increase 1000 times. The screening, orchestration, and optimization of agents will become core responsibilities. At that time, if enterprises cannot establish a high-quality, AI-ready data foundation, they will suffer a 15% productivity loss due to the poor operation of generative AI and agentic systems.

In terms of agent sources, 57% of US enterprises expect application providers to pre-build agents, and 54% of Chinese enterprises expect to customize and build agents.

"In 2025, there were approximately 28.8 million agents globally. By 2029, the total number of active agents will exceed 1 billion, a growth of more than 40 times compared to 2025, of which 39% are unique low-code/no-code custom agents."

Rick further pointed out that in terms of the number of actions, agents took 120 million actions per day in 2025, and this will approach 217 billion times in 2029, a nearly 1,798-fold increase compared to 2025, of which 38% are completed by custom agents.

In terms of the number of Tokens/Calls, the daily Token delivery volume will exceed 37 trillion in 2029, a growth of more than 2,626,000 times compared to 2025, of which 40% are completed by custom agents. By 2029, the average delivery cost of tokens/calls per agent action will be 87% lower than in 2025.

"In the new era of intelligence, design for orchestration, deliver for scale, and govern for trust are the core," Rick believes.

To this end, he suggests that enterprises focus on the following aspects. First, insist on data integrity and invest in AI governance, observability, and interoperability; second, embrace modularity and interoperability and cooperate across ecosystems to build an open agent framework.

Third, design for scale and sustainability and build an architecture that can handle the exponential growth of the number of agents, interactions, and Token/calls requirements. Fourth, reevaluate pricing and delivery models and move beyond traditional seat-based or license-based charging to a results-oriented model that reflects continuous autonomous operation; fifth, be responsibility-oriented and establish corresponding protective measures and compliance mechanisms to win the trust of enterprises and the public.

This article is from the WeChat official account "Finance Graffiti" (ID: caijingtuya), author: Soda, editor: tuya. It is published by 36Kr with authorization.