Asian enterprises raise AI budgets as security and costs constrain deployment

A regional survey by NielsenIQ reveals that while 95% of Asian businesses plan to increase artificial intelligence spending to support key operational functions, persistent concerns regarding data privacy, high implementation costs, and technical skill shortages continue to constrain full-scale enterprise rollout.

This survey commissioned by Alibaba Cloud reveals that 95% of enterprise respondents plan to boost their investment in artificial intelligence products over the coming period.

Among respondents, 69% expected IaaS spending to increase by more than 20%, compared with 61% for PaaS and 58% for MaaS. Conducted in mid-2025, the research gathered responses from 1,000 IT decision-makers spread evenly across eight key Asian markets, including Hong Kong, Indonesia, Japan, Malaysia, the Philippines, Singapore, South Korea, and Thailand. Participating firms represented a broad spectrum of industries, encompassing technology, manufacturing, financial and professional services, retail, and the public sector.

Infrastructure takes priority as adoption widens

The focus on underlying infrastructure coincides with a rapid expansion in the deployment of artificial intelligence tools and models across the region. Roughly 90% of surveyed decision-makers confirmed that their organisations have commenced using platform and model services, with 75% describing artificial intelligence as indispensable to business operations. Conversely, a mere 1% of respondents indicated that the technology was not a priority.

This financial allocation stems from a combination of operational and commercial objectives. Developing new business opportunities ranked highest among primary motivators at 64%, closely followed by cost savings and efficiency at 63%. Revenue growth motivated 48% of respondents, while 45% sought a competitive advantage and 43% aimed to improve customer experience. At present, deployments remain heavily concentrated in operational areas backed by established data sources and repeatable processes. Data analysis and decision-making led implementation at 66%, with customer service applications such as chatbots following at 64%. Marketing and content creation recorded 58% adoption, 55% applied the technology to product development and coding, and 38% incorporated it into human resources operations.

These spending plans place fundamental computing, storage, and networking resources ahead of development platforms and model access. These underlying systems are required to connect models with corporate data, apply security controls, and support applications as they move into regular business use.

Integrated systems attract interest as complexity increases

As managing infrastructure, platforms, and models together becomes increasingly complex, supplier selection is shifting. Integrated artificial intelligence and cloud offerings were preferred by 45% of respondents, whereas 34% favoured models capable of running across multiple clouds. A further 16% preferred standalone models without bundled cloud infrastructure.

Decision-makers associated integrated systems with centralised management, standardised data pipelines, and consistent governance policies. Consolidating these functions within one technology stack can simplify deployment, although it may also increase dependence on a provider’s services, pricing, and technical architecture.

Despite growing investment, security remains the most frequently cited barrier to wider adoption. Concerns regarding data privacy and security were selected by 48% of respondents, followed by high implementation costs at 42% and insufficient internal expertise at 37%. Regulatory and ethical concerns were identified by 31% of respondents, with the proportion reported to be higher in financial services, education, and the public sector. Integration difficulties affected 28%, while 23% struggled to access suitable products, 22% cited uncertain returns on investment, 19% reported difficulty identifying viable business cases, and 18% encountered resistance from management-level executives.

These barriers persist despite broad access to individual artificial intelligence products. Although 91% of respondents said artificial intelligence products were available in their markets, many wanted more complete systems designed around specific industries and capable of working with established processes. Customised solutions for industry use cases were requested by 54%, while 49% wanted better access to talent and skills, and 45% cited more affordable offerings. Respondents also identified successful case studies at 42%, vendor-provided training at 41%, and stronger board and management endorsement at 32% as requirements for larger deployments.

Singapore use concentrates on data and customer service

Singapore respondents reported particularly strong interest in artificial intelligence, with 96% describing their organisations as either very positive or cautiously optimistic about adoption.

Data analysis was the leading local use case, cited by 80% of respondents, representing the second-highest proportion among the eight surveyed markets. Customer service followed at 75%, while 54% reported using artificial intelligence for marketing and content creation.

This concentration on data analysis and customer-facing applications places added demands on information governance, system integration, and internal expertise. Higher infrastructure spending may provide the necessary capacity, but the survey’s security and skills findings indicate that technology budgets alone will not resolve deployment constraints.

Local-language availability affects regional deployment

Language support presents another limitation in markets including Japan, South Korea, Indonesia, and Thailand. Around half of respondents selected English as the primary language for their artificial intelligence applications, while the remainder relied on Bahasa Indonesia, Japanese, Korean, Thai, and other local languages.

Applications used in customer service, content creation, and internal workflows must account for the languages used by employees and customers. Limited access to suitable local-language models can restrict deployment even when general-purpose artificial intelligence products and cloud infrastructure are available.

Alibaba Cloud cited its Qwen models as an example of multilingual support, saying the latest generation covers more than 200 languages and dialects and has gained particular traction in Japan and South Korea. The claim forms part of Alibaba Cloud’s interpretation of the research rather than an independent comparison of model performance or market share.

Alibaba ties the findings to its infrastructure plans

Alibaba Group has committed at least RMB 380 billion, approximately US$53 billion, over three years to expand cloud computing and artificial intelligence infrastructure. The company said the planned investment exceeds its combined spending on artificial intelligence and cloud services over the previous decade.

Alibaba Cloud offers infrastructure, development platforms, and self-developed models, including the Qwen language model family and Wan image and video generation models. The survey’s preference for integrated artificial intelligence and cloud services closely matches the company’s own full-stack product strategy.

Dr Feifei Li, Chief Technology Officer and President of International Business at Alibaba Cloud Intelligence Group, said, “The research findings reinforce Alibaba Cloud’s belief that AI, delivered on top of scalable, secure cloud infrastructure and models, will be the defining technology platform for the next decade. As we enter the agentic AI era, our focus is shifting from producing efficient tokens to enabling actionable outcomes. By building a comprehensive agentic cloud, we provide the critical infrastructure, such as runtime sandboxes and orchestration, that empowers enterprises to seamlessly build the agent-native products of tomorrow.”

The investment gives Alibaba Cloud a direct commercial interest in the spending priorities identified by the survey. Infrastructure was the category most frequently expected to receive budget growth above 20%, followed by the platform and model services that make up the rest of the company’s artificial intelligence and cloud portfolio.