Amidst a significant surge in data center CPU demand and the 18A advanced chip manufacturing process entering a growth trajectory, Wall Street giants are showing increasingly bullish sentiment towards x86 architecture CPU leader Intel (INTC.US). Citigroup has sharply raised its target price for Intel from $95 to $130, while another prominent investment firm, Melius Research, has increased its target from $100 to $150. This underscores how the "data center CPU demand frenzy + improving profit prospects for the 18A advanced chip process" are jointly fueling and advancing Intel's super bull market narrative.
Reports indicate that Intel has requested PC manufacturers to adopt chip products based on its 18A advanced chip manufacturing process. Wall Street investment giant Wedbush Securities views this as a positive signal that the semiconductor manufacturing leader is prioritizing margin expansion. Wedbush Securities analyst Matt Bryson wrote in a report to clients on Wednesday, "In our view, this strategy is very reasonable because Intel should prioritize using its older, capacity-expanded process nodes for higher-margin Xeon data center-level CPU production. Intel's ability to activate newer capacity, due to its existing cleanroom space, is a strategic advantage. This approach allows it to leverage this strength."
"The question, in our view, is rather how strong the actual performance of the 18A node and chips manufactured using this process will be. CEO Pat Gelsinger recently hinted that yields are improving rapidly," Bryson added.
At the 54th Annual Global Technology, Media, and Communications Conference hosted by Wall Street giant JPMorgan, Intel CEO Pat Gelsinger stated that Intel 18A, the advanced chip process below 2nm (approximately 1.8nm level), already supports Panther Lake mass production, with yields improving by about 7% per month, exceeding internal expectations.
According to Gelsinger's latest update, Intel 14A's 0.5 PDK has been released, with plans to launch the 0.9 PDK to external customers in October. The team has also begun long-term advanced process planning for the 10A and 7A nodes.
Gelsinger further noted that as the focus of AI computing infrastructure shifts from training to inference, CPUs are becoming increasingly vital and indispensable in the AI era. The CPU-to-GPU configuration ratio is accelerating from 1:8 towards 1:1, potentially even reaching 4:1. Additionally, Intel's business plans show it is actively pursuing ASIC business, offering customized AI CPU or AI GPU chip solutions.
CPU Renaissance With the launch of Anthropic's Claude Cowork and the anticipated proliferation of autonomous AI agent tools like OpenClaw by 2026, the wave of AI agents is rapidly sweeping the globe. The bottleneck in AI computing architecture is shifting decisively from GPUs, centered on matrix multiplication and addition throughput, to data center CPUs focused on control flow, task orchestration, and memory/IO coordination. High-performance CPUs for hyperscale AI data centers are facing severe supply shortages.
Wall Street analysts are expanding the AI computing infrastructure narrative from "GPU dominance/single-core driven" to a "full-stack computing reassessment" involving "AI GPU/ASIC + CPU + HBM/DRAM/NAND memory chips + data center high-speed interconnect systems led by optical interconnects."
As AI agents gain global popularity, the main investment theme in AI computing is shifting from a "single-point computing race around GPUs" to an "AI agent-driven full-stack computing system." The next wave of excess alpha returns will no longer belong solely to the strongest leaders in the AI GPU/AI ASIC field but will systematically spread across the full-stack AI computing infrastructure layer, including CPUs, memory, PCBs, liquid cooling systems, ABF substrates, and a wide range of wafer foundries. In this shift of the AI narrative, CPUs, optical interconnects, and memory chips are likely to be the biggest winners.
For the past two years, the AI narrative was almost monopolized by GPUs, with CPUs seemingly playing a "supporting role" in the AI arms race. However, with the comprehensive growth of inference workloads, data orchestration, task scheduling, memory access, network communication, and multi-tool invocation driven by agentic AI workflows like the open-source OpenClaw, the market has come to a stark realization: without powerful CPUs as the system's central hub, GPU clusters cannot operate efficiently. This essentially marks a "Renaissance" for CPUs, returning them from "underestimated infrastructure" to the very center of the chip stage.
Early large model inference primarily involved "single request-single generation," with CPUs handling data movement, request routing, and basic scheduling—typical auxiliary control plane functions. But as we enter the era of AI agents and reinforcement learning, system loads are no longer just single forward inference. They have evolved into complex closed loops involving task planning, tool invocation, sub-agent coordination, environmental interaction, state management, and result verification. The "orchestration layer" inherently involves CPU-intensive tasks with strong control flow, branch decisions, system calls, and memory access, which cannot be efficiently replaced by GPUs. Consequently, CPUs are transforming from their past "supporting role" into the new bottleneck determining system throughput, latency, and resource utilization.
Data Center CPU Demand Surge + Advanced Chip Foundry Poised for Rise The 18A process serves as a manufacturing validation point for the narrative of "data center CPU demand frenzy + improving profit prospects for the 18A advanced chip process." Wedbush interprets Intel's request for PC manufacturers to adopt 18A chips as "margin protection." The key logic is that Intel's management is striving to allocate more of its limited older-node capacity to high-margin Xeon, server, and industrial clients, while using 18A for new client-side products, thereby optimizing capacity allocation and margin structure.
Around the same time, Tom's Hardware, citing a Nikkei report, stated that Intel has shifted its constrained Intel 7 capacity towards server and industrial clients due to higher margins in these segments, with AI-driven data center CPU demand expected to continue rising in 2025.
The core message from Pat Gelsinger's statements at the JPMorgan conference was to inform the market that 18A is not just a node on the technology roadmap but has already entered the mass production support phase for Panther Lake, with yields improving by approximately 7-8% per month, transitioning from an "engineering risk" to "commercially verifiable."
If 18A mass production stabilizes, it will not only support PC-side Panther Lake but also establish a trust foundation for subsequent server CPUs, AI head-nodes, ASIC foundry services, and 14A client adoption. This is precisely the core of the broader narrative that is repositioning Intel from a "lagging process company" to being repriced by the market as a "US advanced chip manufacturing counteroffensive asset."
Simultaneously, the shift of AI from training to inference and Agentic AI will significantly increase the strategic weight of CPUs in AI data centers. While GPUs handle large-scale matrix computations, CPUs are responsible for scheduling, I/O, memory management, task orchestration, security isolation, database access, network stacks, and multi-agent workflow execution. As AI applications evolve from "single-shot inference" to "continuously running software labor," CPU demand will shift from traditional server refresh cycles to AI infrastructure expansion cycles.
Citigroup's latest model echoes this view. It forecasts that the total addressable market (TAM) for data center server CPUs will expand from $29.3 billion in 2025 to $131.5 billion in 2030, representing a compound annual growth rate of approximately 35%. This outlook prompted Citigroup to raise its target price for Intel significantly from $95 to $130.
The highest Wall Street target price of $150 comes from Melius Research's star analyst Ben Reitzes. Melius incorporates Intel into an AI semiconductor "bottleneck asset" revaluation framework. Reitzes believes that as AI computing infrastructure demand continues to create supply bottlenecks, semiconductor companies like Intel, which focus on these AI computing bottlenecks, will gain more market capitalization or upside potential compared to traditional software companies and non-semiconductor "Magnificent Seven" tech giants.
Regarding Intel's fundamentals, Melius's logic primarily hinges on two points: First, Agentic AI is driving a reacceleration of x86 server CPU demand. As AI moves from training to inference and agent execution, more CPUs are needed for scheduling, I/O, memory management, workflow orchestration, and security control. Second, under Pat Gelsinger's leadership, Intel Foundry (its advanced process chip foundry business) has the potential to unlock significant growth value.
Comments