Right workload, right device: Dell and Intel make the case for distributed AI compute in APAC

With Southeast Asia outpacing the regional AI PC adoption average and 95% of organizations banking on workstations for AI, the intelligent endpoint era is no longer theoretical.

The cloud-first playbook is being rewritten. Across Asia Pacific, enterprises are waking up to a simple but consequential realization: not every AI workload belongs in the cloud, and not every employee should be waiting on it.

The shift is less about abandoning the cloud and more about getting smarter with compute. Sensitive data that cannot leave the building, latency-sensitive workflows that cannot afford a round trip, and AI development cycles that need sustained, predictable performance–these are workloads that belong closer to where work actually happens.

And that is precisely the argument Dell Technologies and Intel are making with two new IDC InfoBriefs released this week. The research paints a clear picture of where enterprise AI is heading: away from a single centralized model and toward a distributed architecture where AI PCs handle intelligent everyday workflows and workstations carry the heavier lifting of development, training, and specialized compute.

Dell calls it an AI compute continuum and the data suggests Asia Pacific organizations are already building it, whether they call it that or not.

Southeast Asia pulling ahead

Across Asia Pacific, 48% of organizations with more than 500 employees have already deployed AI PCs. Southeast Asia as a whole is outpacing that regional average by 6%, driven by leapfrog adoption strategies, Singapore's infrastructure maturity, and government-backed digitalization initiatives.

Malaysia is a meaningful contributor to that momentum, with 45% of organizations already deploying AI PCs. Marketing has emerged as the country's top anticipated impact area, spanning real-time content generation and privacy-preserving personalized messaging.

Singapore sits higher at 54%, with facility management singled out as an emerging focus area, rated 15% above the regional average, reflecting the city-state's push toward smarter building and operational infrastructure.

Interestingly, Thailand leads in Southeast Asia when it comes to AI PC adoption at 60% and the Philippines at 58%. While India is at 51% adoption, the size of its workforce would also mean that it has the most number of AI PCs in the region.

The productivity numbers make the investment case difficult to ignore. Employees in Southeast Asia are saving an average of 2.09 hours per day through AI-enabled PC features, 7% more than the regional average. Organizations that have crossed the 50% AI PC deployment threshold report saving 2.17 hours per employee daily–a 30% productivity gain over those still running AI on traditional PCs.

"AI PCs bring intelligence to everyday workflows, at the fingertips of employees where data is generated," said Jacinta Quah, vice president of Client Solutions Group and Modern Workplace, APJC at Dell Technologies. "Together [with workstations], they enable organizations to scale AI more effectively, strengthen security and privacy, and drive meaningful business outcomes."

Four out of five Asia Pacific organizations expect AI PCs to accelerate agentic AI adoption, while 78% cite security benefits and 77% highlight the cost advantages of running AI locally as compelling factors. Across the region, 65% of organizations are prepared to pay a premium of 10% or more for AI PCs as part of their device refresh cycle, a signal that intelligent endpoints are no longer treated as commodity hardware.

SEA organizations in particular place strong emphasis on manageability and support infrastructure when evaluating AI PC partners, reflecting the region's diverse and often fragmented IT environments.

Workstations: where the serious AI work gets done

While AI PCs address the workforce layer, workstations are handling the heavier lifting, and demand for them is climbing as organizations shift more AI development on-premise, driven by data sovereignty concerns, rising cloud costs, and the need for consistent, low-latency performance.

Across Asia Pacific, 95% of organizations expect workstations to be critical or important to their AI initiatives over the next two years. In Southeast Asia specifically, inclusive of Singapore, Malaysia, and Thailand, 92% of organizations report higher productivity among workstation users, and 52% expect their workstation fleet share to grow within five years.

The IDC workstation research, drawn from 960 IT and business decision-makers across the region, found workstations being used across the full AI development lifecycle: data preparation (62%), foundational model training (60%), model fine-tuning (59%), deployment (44%), and inference (29%). In SEA, organizations are especially active in data preparation (66%), model fine-tuning (62%), and foundational model training (55%).

"The speed at which AI models are being compressed to run on-device has been remarkably fast," said Bryan Ma, vice president of Client Devices at IDC. "In the next year or two, very robust models will run on PCs that far exceed today's capabilities. At the same time, organizations continue to depend on high-performance workstations for advanced AI development and specialized workloads."

Notably, IDC flags that the primary barrier to broader workstation adoption is no longer cost or capability, it is outdated IT procurement policy. The research argues that organizations need to modernize these policies to reflect total cost of ownership, accounting for device longevity, reduced cloud dependency, lower latency, and sustained performance over time.

On supply chain resilience amid rising memory costs and infrastructure pressures, Dell says its globally distributed supply chain positions it to absorb those pressures without disproportionate impact to customers. "We are confident in our ability to secure supply and adjust pricing as needed, leveraging our world-class supply chain to deliver the best outcomes for our customers and shareholders," the company said.

The broader argument is architectural. Enterprise AI cannot scale on cloud alone, and a distributed model–intelligent endpoints for the workforce, high-performance workstations for development–offers a more practical, more secure, and more cost-controllable path forward for organizations across the region.