A shrinking global market can still generate more sales dollars
IDC’s June outlook projected global traditional-PC shipments would decline 11.3% across 2026. It also warned that the final quarter could see a year-over-year fall of as much as 20%. This is a shipment forecast, not a measure of consumer purchases alone or of vendor revenue. PC shipments can represent products moving into distribution channels as well as systems reaching end users, and the traditional-PC category does not include tablets or x86 servers.
That distinction matters. Unit shipments and revenue answer different questions:
- Shipments show how many PCs are moving through the global traditional-PC market.
- Revenue reflects both the number of PCs sold and their selling prices, plus the mix of products and customers.
- Profitability depends on costs, pricing, operating expenses and other factors; revenue growth by itself does not prove that margins are improving.
IDC expects average PC selling prices to rise sharply in 2026. Its accessible June materials give slightly different estimates—17% in one publication and 18.3% in another—rather than a single definitive 20% figure. Either estimate describes a substantial shift. If prices rise materially faster than unit volume falls, industry revenue can grow even as the global traditional-PC shipment market weakens.
The arithmetic is straightforward, but the underlying conditions are not benign. A vendor selling fewer machines at far higher prices may post a strong top-line quarter while buyers delay upgrades, lower-cost configurations become less attractive, and institutional purchasers absorb higher acquisition costs.
Memory is the central pressure point, but not the whole story
IDC identifies the persistent memory shortage as a major force behind the global traditional-PC market’s volatility and expects no meaningful relief before the end of 2027. Higher memory costs affect a broad set of PCs, from low-end notebooks to workstations. They can push manufacturers to raise list prices, alter configurations, limit certain models, or emphasize products where a higher selling price better absorbs cost increases.
It would be too simple, however, to assign every price increase or shipment decline exclusively to demand from AI infrastructure. The broader global market outlook also identifies storage-component constraints and geopolitical issues as headwinds. Those factors can affect availability, manufacturing decisions, pricing and enterprise purchasing independently of AI data-center investment.
The near-term prognosis is therefore not simply “AI makes laptops expensive.” It is a supply-constrained global traditional-PC market in which multiple component and policy pressures are meeting a demand environment that may itself be distorted. IDC has said earlier growth was partly driven by orders pulled forward ahead of expected price increases and constrained availability. Pull-forward demand can make one period look stronger while leaving less demand for later quarters.
For buyers, this means a high price should not automatically be read as evidence of a better or more future-proof PC. Some of it may reflect a superior display, processor, memory capacity, warranty or build quality. Some may reflect higher input costs and constrained supply. Those are very different sources of value.
Vendor results demonstrate the disconnect, not a single cause
Recent results from major PC makers show how revenue can rise during a difficult global traditional-PC unit environment.
HP reported fiscal third-quarter 2026 Personal Systems revenue of $11.8 billion, up 18% year over year, even as total Personal Systems units fell 16%. Commercial revenue increased 22%, while consumer revenue rose 10%. HP also highlighted stronger sales and share in premium products and AI PCs. Its Personal Systems operating margin, however, was 4.6%, an important reminder that a revenue gain should not be treated as proof of unconstrained profitability.
Dell reported that its Client Solutions Group generated $15.034 billion in revenue in the quarter ended July 31, 2026, a 20% year-over-year increase. Commercial revenue rose 22% and consumer revenue 7%; Client Solutions Group operating income increased 42%. Dell’s reported operating margin for the group was 7.6%.
Lenovo reported that its Intelligent Devices Group revenue climbed 27% to $17.1 billion in fiscal first-quarter 2026/27, with revenue from PCs and smart devices up nearly 30%. The group’s operating margin was 7.1%.
These reports support a meaningful conclusion: major manufacturers can produce higher reported revenue despite weak broad global traditional-PC shipment expectations. They do not prove that AI PCs alone caused the gains.
The reported categories are not directly interchangeable. HP’s Personal Systems reporting, Dell’s Client Solutions Group reporting and Lenovo’s Intelligent Devices Group reporting cover different mixes of products, services and customer segments. Lenovo’s cited number, in particular, includes PCs and smart devices rather than a pure-PC figure. Their fiscal calendars and reporting periods also differ.
Commercial demand, premium product mix, regional share changes, enterprise refresh cycles and price increases may all contribute. Dell’s much faster commercial than consumer revenue growth, for example, is consistent with an important business-market contribution. It does not identify how much of the increase came from AI features, memory-driven price changes, conventional corporate replacements or a combination of all three.
Premium mix is plausible, but the evidence has limits
Manufacturers have clear incentives to prioritize products that can command higher prices when components are scarce. Premium laptops and desktops typically leave more room for higher memory capacities, stronger graphics, better displays, extended support, higher-end processors and business-oriented configurations. AI branding may further help vendors differentiate those products.
That is a plausible explanation for a portion of the current revenue performance. It remains an inference, not a verified industry-wide allocation finding. The available company reports show strong revenue results and some premium or AI-PC momentum, but they do not establish that manufacturers are systematically diverting scarce memory from entry-level PCs into AI PCs.
Nor is it established that every vendor is preserving or expanding margins through premiumization. HP and Dell’s segment results confirm that both revenue and operating performance need to be examined, but one company’s result cannot be generalized to the entire global traditional-PC industry. Supply contracts, product portfolios, channel inventory, commercial exposure and component hedging can produce very different outcomes.
This distinction matters because “AI PC” is a broad commercial label. A laptop may be sold as AI-ready because it has a neural processing unit (NPU), but its price can still be shaped more by memory, storage, display, chassis, CPU tier, GPU, sales channel or enterprise support bundle than by the NPU itself.
What local AI on Windows actually requires
The strongest practical case for an AI-oriented Windows PC is local processing: certain tasks can run on the device rather than sending input to a remote service. Microsoft says Copilot+ PCs contain built-in AI components intended for local operation on dedicated AI hardware, including an NPU. Microsoft also describes lower latency, less reliance on cloud services and keeping sensitive data on the device as potential benefits of local processing.
But Windows local AI is not exclusively an NPU story. Microsoft’s documentation says its local inference tooling can select among an NPU, a DirectX 12-capable GPU, an NVIDIA GPU through CUDA, or a CPU fallback. The best available hardware path depends on the workload and the software.
That creates a more nuanced buying rule than “always buy the highest NPU rating.”
An NPU can be particularly relevant when a buyer wants Copilot+ capabilities or frequent, power-efficient on-device AI tasks on a portable PC. A capable GPU may matter more for workloads that are designed to use graphics acceleration. And for occasional, modest AI use, a CPU fallback may be sufficient—though it may not match the responsiveness or power efficiency of dedicated acceleration.
Privacy also deserves precise treatment. Local processing can keep the specific data processed by a supported local feature on the device rather than sending that input to Microsoft’s servers. That does not mean every AI application, feature, account service or model is automatically local and private. Buyers should check the behavior of the particular Windows feature or application they intend to use.
How Windows buyers should respond to higher prices
The global traditional-PC shortage outlook gives buyers a reason to plan rather than assume that next month’s sale will reset the market. IDC expects pricing to remain elevated and has said prices are not expected to return to 2025 levels even after memory capacity expands. Yet that is not the same as evidence that waiting can never save money. IDC also notes competitive pressure from Apple’s MacBook Neo is helping keep some low-cost notebook choices available.
For a personal Windows purchase, the practical approach is to separate urgent need from speculative fear:
- Replace now if the existing PC is unreliable, unsupported for your needs, or blocks paid work or study. The productivity cost of waiting may outweigh a possible future discount.
- Set a specification floor before comparing AI labels. Memory capacity, storage, display quality, battery life, ports, repairability and warranty may matter more than an AI badge. In a memory-constrained market, a configuration with adequate RAM deserves particular scrutiny.
- Pay extra for local AI only when the intended software benefits. Buyers who need Copilot+ features, private on-device tasks or sustained AI use have a clearer rationale. Those who mainly browse, stream, write documents and use conventional desktop applications may gain little from a large AI-specific premium.
- Compare complete configurations, not starting prices. A low advertised price can conceal a memory or storage tier that will be limiting over the expected life of the machine.
- Treat discounts as model-specific. A promotion can be worthwhile even in an inflationary market, but it should be compared against equivalent specifications and not presumed to reflect an industry-wide price reversal.
For organizations, the issue is broader than a purchase decision. Higher average prices in the global traditional-PC market can increase the cost of Windows refresh projects, particularly when fleets need more memory or higher-performance configurations. Procurement teams should identify which roles genuinely need local AI acceleration, negotiate component and configuration flexibility, and avoid applying a premium AI standard to every employee by default. A targeted deployment may preserve budgets while still providing capable hardware to developers, designers, analysts and other users with defined local-AI workloads.
A market reset, not an AI-PC victory lap
The evidence supports a guarded conclusion. PC makers can report rising revenue while the wider global traditional-PC shipment market contracts because higher selling prices and a richer product and customer mix can outweigh lower volume. Memory scarcity is a major documented contributor, with storage constraints and geopolitical pressures adding to the uncertainty. Major vendors’ results show the pattern in practice, but their disclosures do not isolate AI PCs as the decisive driver.
For Windows users, that means resisting two equally unhelpful narratives: that every expensive AI PC is essential, or that all price increases are temporary hype that can safely be ignored. The global market may remain costly and constrained for some time, but low-cost options have not vanished entirely. The sensible purchase is the PC whose memory, performance, support life and local-AI capabilities match a real workload—not the one carrying the most fashionable label.