Susquehanna Investment Research raised Intel’s price target to $80 per share on April 23, ahead of the company’s Q1 2026 earnings release, citing strong server CPU demand driven by agentic artificial intelligence workloads. The new target represents a 25% upside from Intel’s Tuesday close of $63.81 per share. Supply constraints are expected to peak in Q1 2026 and ease in Q2 2026, supporting above-seasonal performance for the remainder of the year, according to analyst Christopher Roland’s customer report.
Analyst Upgrade Rationale: Server Strength vs. PC Weakness
Susquehanna expects Intel’s Q1 2026 results to meet or slightly exceed expectations, primarily driven by stronger server CPU demand but partially offset by weak PC original design manufacturer (ODM) shipments. Roland noted that Q1 ODM shipment trends are weaker than previously anticipated, creating downside risk to the client computing group (CCG) segment. The analyst’s CCG forecast is below the market consensus expectation of a 13% sequential decline.
Despite strong server performance, Roland maintains a “neutral” rating on Intel, citing memory chip shortages that are constraining PC assembly. He expects ODM shipments to decline by double-digit percentages throughout 2026 as storage chip shortages persist. On Intel’s foundry business, Roland characterized the company’s decision to join the Terafab project with SpaceX and Tesla as “interesting” and expressed optimism about external customer adoption of Intel’s 14A manufacturing process, describing some partnerships as “very active.”
The CPU Demand Inflection: Why Agentic AI Shifts the Calculus
For the past two years, the artificial intelligence industry narrative has been dominated by GPU capabilities—a dynamic that propelled NVIDIA’s stock to historic highs. CPUs played a supporting role in AI data centers, primarily handling generic control and basic scheduling during the training phase, where GPU parallel computing power handled the most computationally intensive matrix operations.
This dynamic is shifting fundamentally with the emergence of agentic AI and reinforcement learning workloads. Unlike simple text generation, agentic AI breaks a single user request into a complete workflow; the model executes an entire process rather than generating a single answer. When AI transitions from “compute once” to “execute a workflow,” system dependency on CPUs increases significantly. Many critical tasks are ill-suited for GPU execution: task orchestration, thread scheduling, process management, sandbox execution, preprocessing and postprocessing, cache coordination, and state maintenance are all traditional CPU workloads. In multi-agent scenarios where multiple agents run concurrently, invoke tools, and share state, demands on CPU core count, thread count, single-core performance, and memory management capabilities increase substantially.
SemiAnalysis chief analyst Dylan Patel stated in an April 8 interview that AI workload paradigms are shifting from simple text generation to complex agentic and reinforcement learning applications, and CPUs face “extremely severe capacity shortages.” This assessment is corroborated by TrendForce research: current AI data center CPU-to-GPU ratios stand at approximately 1:4 to 1:8, but in the agentic AI era, this ratio is expected to shift to 1:1 to 1:2.
Market Size Expansion and CPU Revaluation
The structural shift is driving substantial market growth. Creative Strategies forecasts that the data center CPU market will expand from $25 billion in 2026 to $60 billion by 2030; when agentic AI-related demand is factored in, the addressable market could approach $100 billion.
This revaluation has triggered supply-side responses. Both Intel and AMD announced price increases on select CPU product lines at the end of Q1 2026. More significantly, NVIDIA and Arm both announced entry into the server CPU market in March 2026—a GPU giant and an IP licensing company making identical strategic moves in the same month is not coincidental but rather a concentrated market signal.
Competitive Landscape: Intel’s Eroding Dominance
Intel’s Xeon processors once commanded over 95% of the data center CPU market. This dominance began eroding in 2021 when Intel 7 process yield issues caused Xeon Sapphire Rapids to delay nearly two years, opening a market gap for AMD’s EPYC Milan.
Intel plans two flagship products for 2026. The Xeon 6+ (Clearwater Forest), based on the Darkmont architecture, will offer 288 cores and 288 threads with a thermal design power (TDP) of approximately 450 watts. The Xeon 7 (Diamond Rapids), based on Panther Cove-X architecture, will reach up to 256 cores and 256 threads with a TDP of 650 watts. Both products will be manufactured on Intel’s most advanced 18A process and will introduce Foveros Direct hybrid bonding technology for the first time. However, TrendForce warns that ongoing 18A process yield challenges may push both products’ mass production timelines to 2027.
AMD’s competitive trajectory appears more stable. Its 2026 flagship EPYC Venice will employ TSMC’s N2 process with Zen 6 architecture and advanced CoWoS-L and SoIC packaging, delivering 256 cores and 512 threads through simultaneous multithreading (SMT)—the highest thread count in the current market. TrendForce expects AMD to continue gaining market share from Intel throughout 2026.
New Entrants Reshape the Competitive Landscape
Beyond Intel and AMD, a wave of non-traditional competitors is entering the server CPU market at unprecedented speed, seeking to fundamentally reshape competitive dynamics.
In March 2026, NVIDIA announced the Vera CPU as a standalone product to address customer demand for more flexible CPU-to-GPU configurations. Vera employs NVIDIA’s proprietary Olympus architecture, built on TSMC’s N3 process with CoWoS-R packaging, and delivers 88 cores and 176 threads with 1.8 terabytes per second of NVLink-C2C interconnect bandwidth, enabling shared memory with NVIDIA GPUs. NVIDIA also introduced the Vera CPU rack, integrating 256 CPUs per rack for a total of 22,528 cores, 45,056 threads, and 400 terabytes of aggregate memory.
Also in March 2026, Arm unveiled its first in-house CPU product, the Arm AGI CPU, ending 35 years of pure licensing business. Built on TSMC’s N3 process with the Neoverse V3 architecture, the AGI CPU provides 136 cores and 136 threads at 300 watts TDP, supporting DDR5-8800 memory and PCIe Gen6. Arm simultaneously launched two rack configurations: an air-cooled version integrating 60 AGI CPUs (8,160 cores, approximately 180 terabytes of memory) and a liquid-cooled version supporting 336 CPUs (45,696 cores and 1 petabyte of memory).
Major cloud service providers are accelerating their own CPU development. Amazon Web Services released Graviton5 (192 cores and 192 threads on TSMC’s N3 process) in December 2025, deploying it alongside its custom Trainium 3 AI ASIC to reduce AI compute costs. Microsoft launched Cobalt 200 (132 cores and 132 threads on N3 process) in November 2025. Google plans to release bare-metal Axion C4A.metal and next-generation Axion N4A in 2026, emphasizing cost-performance optimization.
Valuation Concerns Temper Enthusiasm
While Intel’s recent stock performance is striking and agentic AI’s strong server CPU demand provides a substantive rationale for a positive outlook, analyst sentiment remains cautious amid intensifying competition and Intel’s ongoing operational challenges. Of 52 analysts covering the stock tracked by media sources, only 10 assign a “buy” rating, while 6 assign a “sell” rating—a sell-rating concentration more than double the S&P 500 average. Intel’s current trading price commands a 27% premium to the consensus analyst target price, suggesting the stock has appreciated too rapidly. The stock’s current price-to-earnings ratio exceeds 90x, an all-time high and 50% above the peak valuation during the dot-com bubble; the chip sector average P/E ratio is approximately 21x.
Frequently Asked Questions
Q: Why is server CPU demand surging if GPUs have dominated AI infrastructure for the past two years?
AI workloads are shifting from simple text generation (dominated by GPU matrix operations) to agentic AI and reinforcement learning workflows that require extensive CPU involvement for task orchestration, thread scheduling, process management, and state maintenance. This structural shift is expected to increase CPU-to-GPU ratios in AI data centers from 1:4–1:8 to 1:1–1:2.
Q: What is Intel’s main competitive risk in the server CPU market?
Intel faces execution risk on its 18A process node, with TrendForce warning that both the Xeon 6+ and Xeon 7 flagship products planned for 2026 may not reach mass production until 2027 due to yield challenges. Meanwhile, AMD’s EPYC Venice is on track for 2026 using TSMC’s more mature N2 process, and new entrants including NVIDIA, Arm, and cloud providers are all launching competitive products on TSMC’s N3 process.
Q: Why does Susquehanna maintain a neutral rating on Intel despite raising its price target to $80?
While server CPU demand is strong, Susquehanna notes that memory chip shortages are constraining PC assembly and expects ODM shipments to decline by double-digit percentages throughout 2026. The neutral rating reflects a balanced view: positive server fundamentals are offset by PC segment weakness and competitive execution risks.
