The AI PC revolution: what on-device AI means for data center strategy

December 2025 Update: AI PCs reaching 31% of PC market (77.8M units) in 2025, projected at 94% by 2028. Qualcomm Snapdragon X2 delivers 80 TOPS NPU performance, nearly doubling previous generation. Windows 10 end-of-support in October 2025 forcing hardware refresh cycle. Dell's AI Factory with NVIDIA linking edge devices to large-scale training environments. On-device inference shifting workload distribution between edge and cloud.

AI PCs will represent 31% of the total PC market globally by the end of 2025, with shipments projected at 77.8 million units.¹ Eight out of ten IT decision makers plan to invest in AI PCs this year.² The AI PC market reached $50.68 billion in 2024 and will grow at over 42.8% compound annual rate through 2034.³ By 2028, AI PCs will constitute 94% of PCs in use.⁴ The shift from cloud-dependent AI to on-device processing reshapes enterprise infrastructure strategy in ways that extend far beyond the endpoint.

The transformation reflects a fundamental recalculation. Cloud AI deployments incur high costs and security risks that enterprises increasingly seek to avoid.⁵ On-device AI processing keeps sensitive data local, reduces latency for interactive applications, and eliminates per-query cloud charges for inference workloads. The AI PC becomes an intelligent edge node in a distributed AI architecture, not merely a productivity device that connects to cloud services.

NPU hardware reaches enterprise capability

Qualcomm's Snapdragon X2 delivers 80 trillion operations per second of NPU performance, nearly doubling the 45 TOPS of the first-generation Snapdragon X.⁶ The chip supports up to 128GB of RAM, enabling local processing of substantial language models. First PCs with Snapdragon X2 will ship in the first half of 2026, setting up competition with Apple's M5.

The Snapdragon 8 Gen 5 mobile platform, announced November 25, 2025, enhances the Hexagon NPU with 46% faster AI performance.⁷ Qualcomm claims 36% better performance than Apple's comparable chips while achieving 3nm efficiency. The on-device AI capabilities enable agentic AI assistants that deliver context-aware interactions and personalized suggestions without cloud connectivity.

The edge AI hardware market will reach $26.17 billion in 2025 and grow to $59.37 billion by 2030 at a 17.8% compound annual growth rate.⁸ NVIDIA, Intel, Qualcomm, Samsung, and Apple compete for position. The processors handle larger AI models locally without relying on cloud computation, marking a broader industry shift where energy-efficient on-device AI becomes the key differentiator for both performance and privacy.

Dell's device-to-data center architecture

Dell's strategy, unveiled at CES 2025, treats the PC as an intelligent edge node within a distributed AI ecosystem.⁹ The company introduced three product lines built around integrated NPUs for local AI processing. The Dell Pro AI Studio software platform enables users to find, train, and deploy AI models locally on AI PCs.¹⁰

The approach reflects Dell's shift toward unified device-to-data center AI architecture. By embedding AI capabilities into endpoints, edge devices, and data center infrastructure, Dell democratizes AI access while addressing regulatory and data sovereignty concerns.¹¹ The emphasis on hybrid and on-premises solutions acknowledges that enterprises cannot send all data to public clouds.

In May 2025, Dell unveiled the AI Factory with NVIDIA, a co-engineered portfolio combining data center servers, workstations, storage, and networking with NVIDIA's Blackwell architecture.¹² The partnership formalized Dell's position as both enterprise infrastructure provider and PC innovator, linking edge devices to large-scale model training environments.

Market forecasts predict AI PCs will capture 64% of the PC market by 2028.¹³ The Windows 10 end-of-support in October 2025 forces a hardware refresh cycle, accelerating enterprise AI PC adoption.¹⁴ Organizations must upgrade regardless; AI capabilities provide additional justification for the investment.

Implications for data center strategy

The AI PC proliferation shifts AI workload distribution between edge and cloud. Inference workloads that previously required cloud API calls can execute locally on NPU-equipped devices. The cloud data center retains responsibility for model training, complex inference requiring more compute than endpoint hardware provides, and applications requiring data aggregation across many sources.

The change affects infrastructure planning in several dimensions. Cloud AI inference capacity requirements grow more slowly as edge devices absorb workload. Training infrastructure requirements continue growing as model sizes increase. The balance between inference and training compute shifts, potentially affecting GPU deployment strategies in cloud data centers.

Data governance becomes more complex when AI processing occurs across distributed endpoints rather than controlled data center environments. Enterprises must ensure consistent model versions, maintain audit trails, and protect sensitive data whether processing occurs in the cloud, on premises, or on endpoint devices.¹⁵ The distributed architecture introduces management overhead that centralized cloud deployments avoided.

Goldman Sachs Research forecasts global data center power demand will increase 50% by 2027 and up to 165% by decade's end compared with 2023.¹⁶ However, edge AI processing offloads some demand from centralized facilities. The net effect depends on whether edge AI substitutes for cloud inference or enables additional AI applications that would not have run otherwise.

Enterprise deployment considerations

In the business segment, x86 on Windows will account for 71% of the AI laptop market in enterprise domains this year.¹⁷ The existing enterprise IT infrastructure, management tools, and support capabilities favor Windows x86 over ARM alternatives. Enterprises adopting Arm-based AI PCs must evaluate compatibility with existing applications and management systems.

Enterprises adopt AI PCs to support advanced productivity tools that leverage generative AI, transforming workflows and enhancing operational efficiency.¹⁸ The use cases range from document summarization and code generation to real-time translation and meeting transcription. Local processing enables these capabilities without sending sensitive business content to cloud services.

The high costs and security risks of cloud AI deployments motivate more companies to roll out AI on PCs rather than depending entirely on cloud APIs.¹⁹ The calculation favors on-device processing when usage volumes are high, data sensitivity prohibits cloud transmission, or latency requirements demand local response. Cloud APIs remain preferred when models exceed local compute capacity or when applications require capabilities only frontier models provide.

Gartner forecasts AI PC shipments will total 143 million units in 2026, representing 55% of the total PC market.²⁰ By 2029, AI PCs will become the norm. Organizations should plan for a future where all endpoint devices include AI processing capability, enabling distributed AI architectures that current infrastructure strategies may not anticipate.

Strategic implications

The AI PC revolution does not eliminate data center AI infrastructure requirements. Training frontier models still requires hyperscale compute capacity. Complex inference workloads exceed endpoint capabilities. Multi-user applications require centralized processing. The edge AI expansion complements rather than replaces cloud AI infrastructure.

The strategic question becomes how to balance edge and cloud AI capabilities. Organizations investing heavily in cloud AI infrastructure should evaluate whether on-device inference could serve some workloads more efficiently. Those planning endpoint refreshes should ensure new devices include NPU capabilities that enable future AI applications.

The 1.5 billion PCs expected to refresh over the next five years represents a massive infrastructure transition.²¹ Each device becomes an AI processing node. The aggregate capability across enterprise PC fleets rivals small data centers. Infrastructure strategy must account for this distributed compute capacity and the data governance challenges it creates.

The AI PC revolution extends AI capability to the edge while maintaining the need for centralized infrastructure. Organizations that develop coherent strategies spanning edge to cloud will capture the benefits of distributed AI architecture. Those that treat AI PCs as isolated endpoints separate from infrastructure planning will miss the optimization opportunities the new architecture enables.

Key takeaways

For infrastructure strategists: - AI PCs capture 31% market share in 2025 (77.8M units), reaching 94% of PCs by 2028; aggregate edge compute across enterprise fleets rivals small data centers - Cloud inference capacity requirements grow more slowly as endpoints absorb workload; training infrastructure continues growing as model sizes increase - Windows 10 end-of-support (October 2025) forces hardware refresh cycle; AI capabilities provide additional investment justification

For hardware planners: - Qualcomm Snapdragon X2 delivers 80 TOPS NPU performance with 128GB RAM support; Snapdragon 8 Gen 5 achieves 46% faster AI than predecessor - Edge AI hardware market reaches $26.17B in 2025, growing to $59.37B by 2030 (17.8% CAGR); NVIDIA, Intel, Qualcomm, Samsung, Apple competing - x86 on Windows accounts for 71% of enterprise AI laptop market; evaluate Arm compatibility with existing applications and management systems

For data governance teams: - On-device AI keeps sensitive data local, reduces latency, eliminates per-query cloud charges for inference workloads - Data governance becomes more complex with distributed processing; ensure consistent model versions, audit trails, and data protection across edge and cloud - Dell Pro AI Studio enables local model deployment; hybrid/on-premises solutions address regulatory and data sovereignty concerns

For capacity planners: - Goldman Sachs forecasts 50% data center power demand increase by 2027, 165% by decade's end; edge AI may offset some centralized demand - 1.5 billion PCs expected to refresh over 5 years; each device becomes AI processing node in distributed architecture - Balance edge and cloud: device handles local inference, cloud retains training, complex inference, and multi-source data aggregation

For enterprise deployment: - 8 out of 10 IT decision makers plan AI PC investment in 2025; market reaches $50.68B in 2024 with 42.8% CAGR through 2034 - Dell AI Factory with NVIDIA links edge devices to Blackwell-based data center training environments - Local processing favored when usage volumes high, data sensitivity prohibits cloud, or latency demands local response

References

1. Gartner. "Gartner Says AI PCs Will Represent 31% of Worldwide PC Market by the End of 2025." August 28, 2025. https://www.gartner.com/en/newsroom/press-releases/2025-08-28-gartner-says-artificial-intelligence-pcs-will-represent-31-percent-of-worldwide-pc-market-by-the-end-of-2025

2. IT Pro. "AI PCs will 'become the norm' by 2029 as enterprise and consumer demand surges." 2025. https://www.itpro.com/hardware/ai-pcs-will-become-the-norm-by-2029-as-enterprise-and-consumer-demand-surges

3. GM Insights. "AI PC Market Size, Share & Industry Analysis, 2025-2034." 2025. https://www.gminsights.com/industry-analysis/ai-pc-market

4. IT Pro. "AI PCs will 'become the norm' by 2029."

5. MarketsandMarkets. "AI PC Market Size & Share, Trends, 2025 To 2031." 2025. https://www.marketsandmarkets.com/Market-Reports/ai-pc-market-64905377.html

6. IEEE Spectrum. "Qualcomm's Snapdragon X2 Promises AI Agents in Your PC." 2025. https://spectrum.ieee.org/qualcomm-snapdragon-x2

7. Edge AI and Vision Alliance. "The 8-Series Reimagined: Snapdragon 8 Gen 5 Delivers Premium Performance and Experiences." December 2025. https://www.edge-ai-vision.com/2025/12/the-8-series-reimagined-snapdragon-8-gen-5-delivers-premium-performance-and-experiences/

8. Mordor Intelligence. "Edge AI Hardware Market - Companies, Trends & Insights." 2025. https://www.mordorintelligence.com/industry-reports/edge-ai-hardware-market

9. SiliconANGLE. "Personal computer evolves with AI: Dell's vision for PCs." December 4, 2025. https://siliconangle.com/2025/12/04/dell-transforms-personal-computer-ai-age-dellaipc/

10. SiliconANGLE. "Personal computer evolves with AI."

11. SiliconANGLE. "Personal computer evolves with AI."

12. Forrester. "Dell Tech World 2025: AI's Model T Moment?" 2025. https://www.forrester.com/blogs/dell-tech-world-2025-ais-model-t-moment/

13. theCUBE Research. "2025 is the Year of the AI PC." 2025. https://thecuberesearch.com/ai-pc-2025-is-the-year-ai-goes-beyond-the-browser/

14. IT Pro. "Global PC shipments surge in Q3 2025, fueled by AI and Windows 10 refresh cycles." 2025. https://www.itpro.com/hardware/global-pc-shipments-surge-in-q3-2025-fueled-by-ai-and-windows-10-refresh-cycles

15. Equinix Blog. "How AI is Influencing Data Center Infrastructure Trends in 2025." January 2025. https://blog.equinix.com/blog/2025/01/08/how-ai-is-influencing-data-center-infrastructure-trends-in-2025/

16. Goldman Sachs. "AI to drive 165% increase in data center power demand by 2030." 2025. https://www.goldmansachs.com/insights/articles/ai-to-drive-165-increase-in-data-center-power-demand-by-2030

17. Canalys. "AI-capable PCs forecast to make up 40% of global PC shipments in 2025." 2024. https://www.canalys.com/newsroom/ai-pc-market-2024

18. MarketsandMarkets. "AI PC Market Size & Share."

19. MarketsandMarkets. "AI PC Market Size & Share."

20. Gartner. "Gartner Forecasts Worldwide Shipments of AI PCs to Account for 43% of All PCs in 2025." September 2024. https://www.gartner.com/en/newsroom/press-releases/2024-09-25-gartner-forecasts-worldwide-shipments-of-artificial-intelligence-pcs-to-account-for-43-percent-of-all-pcs-in-2025

21. theCUBE Research. "2025 is the Year of the AI PC."