JPMorgan Research Report Analysis: AI Inference Sustains Servers, PCs Still Paying for Price Hikes
TechFlow Selected TechFlow Selected
JPMorgan Research Report Analysis: AI Inference Sustains Servers, PCs Still Paying for Price Hikes
Supply bottlenecks are the real constraining factor.
Navigating Web3 tides with focused insightsBy: Rita
TechFlow Digest
JPMorgan significantly raised its global server shipment growth forecasts for 2026 to 2028. The 2026 server shipment growth rate was increased from 15% to 22%, and the 2027 rate from 8% to 25%. The core driver is AI inference; as enterprises deploy AI models into practical applications, a large number of inference servers are required.
The PC market performed better than expected in the first half of the year, but this was the result of brands pulling in shipments before price hikes and exhaustion from Win10 upgrades. Demand remains fragile in the second half.
Supply bottlenecks are the real constraining factor. CPUs, substrates, memory, PCBs, and power components are all bottlenecked.
Servers Are Rising, PCs Are Falling
JPMorgan expects PC shipments to decline by 8% and servers to grow by 22% in 2026; in 2027, PCs will grow by 2% and servers by 25%.
This divergence is structural. Servers benefit from the continuous volume release of AI inference demand. JPMorgan estimates that AI accelerator shipments will have a CAGR of 50% from 2025 to 2028, and AI server management node CPU demand will have a CAGR of 74%. By 2028, global server CPU shipments will grow from 26 million units in 2025 to 68 million units, a three-year CAGR of 38%. Among these, Agentic AI server CPU demand will reach 53 million units, serving as the main driver of incremental growth.
PCs are suppressed by memory price hikes and exhausted demand. The BOM cost of high-end laptops has risen by 30% in a year, with memory proportion surging from single digits to 25%. Brands are raising prices to protect gross margins, at the cost of sales volume. JPMorgan predicts consumer PC shipments will decline by 14% in 2026, and commercial PCs by 4%.
Supply Bottlenecks and Nvidia Roadmap Are Two Major Variables
Demand has risen, but shipments cannot keep up. JPMorgan research shows server demand growing 35% to 40% year-over-year, but 2026 shipment growth can only reach 22%. CPUs, substrates, memory, PCBs, passive components, and power components are all bottlenecked.
In the US stock market, Dell (DELL)'s AI server backlog has reached $51.3 billion. JPMorgan analysts maintain an Overweight rating, raising the target price from $280 to $500. Hewlett Packard Enterprise (HPE) also had its target price raised by JPMorgan, reflecting confidence in AI and high-performance computing demand. Super Micro Computer (SMCI) also benefits from the continuous volume release of AI server demand.
ODM order visibility has already extended to 2027. Orders not delivered this year will be carried over to next year, so server shipment growth in 2027 is expected to accelerate further to 25%.
The Nvidia roadmap also holds suspense. Vera Rubin is scheduled for mass production in Q3 to Q4 2026, with a total of 70,000 to 80,000 NVL72 units for the full year, and 85,000 to 95,000 units expected in 2027. The next-generation Vera Rubin Ultra is expected to launch in the second half of 2027 to 2028. The Kyber rack architecture has encountered signal performance challenges with PCB and CCL materials, which may delay the Feynman generation.
Value Is Shifting Towards Components
Changes in BOM costs point to a clear direction: money is flowing to component manufacturers. GB300 and VR200 rack prices are 20% to 90% higher than GB200. Memory accounts for about 20% of VR200 BOM, compared to only 10% during the GB200 era. For component manufacturers, this is a window for simultaneous volume and price increases; for PC brands, this is the source of gross margin pressure.
In terms of US stock components, Arista Networks (ANET) is a core beneficiary in the data center switch segment. Amphenol (APH) has layouts in the high-speed connector field, benefiting from increased interconnect density within racks. Corning (GLW) reached a multi-billion dollar fiber optic supply agreement with Amazon, directly benefiting from the expansion of data center interconnect demand. Lumentum (LITE) has exposure in optical communication and optical engine directions. Micron (MU), as a memory supplier, is a direct beneficiary of rising BOM costs.
JPMorgan's allocation advice is straightforward: server components are superior to ODMs. Avoid PCs overall.
TechFlow Perspective
The most valuable part of this report lies in clarifying the roles of inference and training within the cycle. It does not stop at the conclusion that "servers are good, PCs are bad." The market has previously been trading training compute power, and only now realizes that inference demand has stronger sustainability.
The risk lies in the possibility that the implementation speed of Agentic AI may be slower than expected, making the demand forecast of 53 million CPUs too high. Additionally, Nvidia's Kyber architecture has encountered signal issues with PCB and CCL materials, and CPO mass production may be delayed until 2028. For the optical interconnect industry chain, this is a time window, not the end point.
Regarding PCs, supply chain data in the second half of the year may still fall below expectations, but the market has already partially priced this in.
Disclaimer
This article is a compilation and interpretation by TechFlow Research of JPMorgan's research report dated July 15, 2026. The ratings, target prices, earnings forecasts, and related judgments cited in the text are the views of the broker's analysts, representing only their institution's position, not representing the views of TechFlow Research, nor constituting any investment advice.
The market involves risks; decisions must be made independently. This article should not be used as a basis for buying or selling any securities.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@chaoxiangooo
