Arm AGI CPU backlog doubles in 6 weeks, UBS maintains Buy and raises target price to $260

Overview

UBS’s earnings review report on Arm Holdings PLC (ARM.US) notes that Arm slightly beat revenue and EPS expectations in the quarter ended March, driven by royalty revenue growth from hyperscale cloud providers and Nvidia’s accelerated adoption of ARM CPUs. More importantly, the AGI CPU product backlog doubled from approximately $1 billion to $2 billion just 6 weeks after launch, indicating extremely strong demand. However, capacity constraints have prevented the company from raising revenue guidance, signaling supply is the bottleneck. UBS believes the long-term growth potential is underestimated. Accordingly, UBS maintains its Buy rating and raises the target price from $245 to $260.

Core views

Earnings Performance: Arm reported FQ4:26 revenue of $1.49 billion (+20% YoY) and Non-GAAP EPS of $0.60, slightly exceeding UBS and market expectations. Outperformance came mainly from strong licensing revenue ($819 million, +29% YoY), partly offset by royalty revenue ($671 million, +11% YoY) which was below the expected 13-14%. The royalty shortfall was due to MediaTek’s high prior year base and weakness in low-end smartphones. Core Growth Logic — Data Center Royalties: Data center royalty revenue doubled YoY, the largest highlight this quarter. Arm’s computing share among leading hyperscalers has reached approximately 50% (near zero at end 2020). Trends such as Google TPU paired with Axion CPU, AWS expanding Graviton, Microsoft advancing Cobalt, and Nvidia launching 256-core Vera CPU continue to strengthen Arm’s cloud penetration. UBS views Arm’s ~20% long-term royalty CAGR as possibly conservative, expecting CPU shipments and core counts to accelerate growth in FY27/28. AGI CPU Demand and Capacity Bottleneck: Arm management disclosed that since the late March launch, AGI CPU backlog doubled from about $1 billion to $2 billion. Yet the company still maintains revenue guidance at around $1.05 billion for the period, implying supply rather than demand is the limiting factor. The AGI CPU chip business is forecast to reach positive operating profit in 2027, with long-term operating margin around 35% (lower than IP model ~65%). The FY2031 framework of $15 billion AGI CPU chip revenue plus $10 billion IP revenue supporting over $9 EPS remains unchanged. Market Share and Long-Term Potential: UBS estimates a significant increase in Arm’s server CPU market share, with long-term IP income potential potentially exceeding the company’s FY31 target of about $10 billion. UBS is cautious about Arm achieving 15% CPU device market share by FY31, considering META accounts for less than 10% of global server shipments and OAI/Stargate deployments mostly use NVDA racks. UBS revises TAM to $170 billion, considering $15-16 billion revenue (~8-9% share) achievable.

Analysis framework

UBS’s analysis follows the main thread of “earnings validation → demand dissection → supply-demand conflict → long-term TAM and valuation”: First, from quarterly earnings, revenue is split into royalties and licensing, identifying that licensing exceeded expectations while royalties were hampered by mobile but with a clear core alpha signal from doubling data center royalties. Next, focusing on the AGI CPU backlog: comparing a backlog of $2 billion to revenue guidance of $1.05 billion reveals supply (not demand) as the bottleneck, forming the core bullish case for long-term potential. In demand logic, UBS introduces new perspectives on “cores per workload” and “cores per intelligent agent session,” arguing that traditional methods measuring demand by “CPU to GPU chip ratios” underestimate AI-driven CPU intensity. With the appearance of dedicated CPU racks and Vera architecture, previous assumptions of 4x CPU capacity increase may be conservative. Finally, valuation: considering Arm’s growth outpaces most semiconductor peers, UBS applies PEG instead of absolute PE valuation. Benchmarking against EDA/IP peers (e.g., SNPS, CDNS), UBS assigns a 1.6x target PEG multiplied by Arm’s 37% long-term EPS CAGR to derive a 61x PE, then multiplied by CY2028E EPS of $4.33, reaching a $260 target price.

Methodology notes

• Valuation MethodPE/PEG valuation

  PEG Valuation Method

  PEG (Price/Earnings to Growth ratio) divides PE by earnings growth rate, aiming to incorporate growth into valuation comparison. UBS believes Arm’s growth far exceeds peers, so using absolute PE isn’t fair. Thus it benchmarks against EDA/IP peers (e.g., SNPS, CDNS) using a 1.6x target PEG, multiplied by Arm’s 37% long-term EPS CAGR, resulting in 61x PE, then multiplied by CY2028E EPS $4.33 to reach $260 target price. This is lower than the prior 1.8x PEG but benchmarks a more reasonable peer group.

• Industry/Market Analysis FrameworkSupply and Demand Framework

  Backlog Analysis Under Supply-Demand Framework

  When a company’s backlog ($2 billion) far exceeds its near-term revenue guidance ($1.05 billion), it usually implies supply-side capacity is the bottleneck rather than demand shortfall. UBS therefore judges Arm’s AGI CPU true revenue potential is underestimated, and revenue should accelerate once capacity expands.

• Company Fundamentals and Financial FrameworkOthers

  Dual Engines of Royalties and Licensing Analysis

  Semiconductor IP companies’ revenues typically split into upfront licensing and later-stage royalties. Licensing revenue can fluctuate with timing of large deals while royalties better reflect actual end-product shipments. UBS’s breakdown reveals licensing beat due to large strategic deals, while royalty growth slowed by smartphone weakness, but doubling data center royalties reveals deeper product penetration.

• Industry/Market Analysis FrameworkPenetration S-curve

  Judging Technology Penetration Inflection Points

  Arm’s computing share among hyperscalers rose from near zero at end 2020 to ~50%, currently on a rapid climb along the S-curve. UBS focuses not only on chip shipments but introduces “core count” as a dimension replacing “chip count” to measure AI-era CPU demand intensity. As CPU cores evolve from 136 to 256/512, penetration upside further opens.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Arm Holdings PLC (ARM)

  Core coverage in report, bullish on data center royalty doubling and strong AGI CPU demand

  Strengths

  Rapid share gain among hyperscalers (from 0 to 50%), potential upside in long-term royalty CAGR, AGI CPU backlog doubling indicates strong demand, FY31 long-term framework supports over $9 EPS

  Weaknesses

  Royalty revenue dragged by mobile weakness, AGI CPU capacity may limit near-term revenue realization, chip business margin (~35%) lower than pure IP model (~65%), low free float (12%)

  Comparison

  UBS benchmarks against EDA/IP peers (e.g., SNPS, CDNS) rather than traditional semiconductor peers due to similarity in IP business model

  Risks

  Uncertainty if 15% CPU device market share is achievable by FY31; META’s limited global server share (<10%) and heavy NVDA rack deployment by OAI/Stargate may limit share growth

• AMD / INTC

  Beneficiaries: Arm’s bullish stance on data center CPU demand benefits overall CPU market

• AVGO / MRVL / ALAB

  Beneficiaries: Gain from broader AI network opportunity

Key data

• FQ4:26 Revenue$1.49 billion20% YoY growth, slightly above UBS/market expectation of $1.47 billion
• FQ4:26 Non-GAAP EPS$0.60Exceeds UBS/market expectation of $0.58, driven by strong licensing and lower operating expense
• Royalty Revenue$671 million11% YoY growth, below 13-14% expectation due to MediaTek’s high base and weak smartphones
• Licensing Revenue$819 million29% YoY growth, significantly above expectations, driven by multiple large strategic deals
• Data Center RoyaltyDoubled YoYGreater than 100% growth, core growth engine
• AGI CPU BacklogApproximately $2 billionDoubled from about $1 billion within 6 weeks of product launch
• FQ1:27 Revenue Guidance$1.21-$1.31 billionAbout 15% sequential decrease, royalties and licensing each expected to grow about 20% YoY
• FQ4:26 Operating Margin49.1%Operating expense $734 million below expectation, boosting margin
• FY2031 Long-Term Framework$25 billion total revenueIncludes $15 billion AGI CPU revenue + $10 billion IP, supporting over $9 EPS
• Long-Term EPS CAGR37%UBS estimated long-term compound annual growth rate of EPS
• Target PEG1.6xPreviously 1.8x, in line with average among EDA peers
• Target Price$260Previously $245, based on 61x CY2028E EPS $4.33

Impact & implications

UBS believes Arm’s ongoing penetration in data center CPUs is a positive signal for the entire data center CPU market and custom chip complexity. Specifically, Arm’s bullish stance on CPU demand benefits the broader data center CPU market including AMD and INTC, as well as AI network beneficiaries AVGO, MRVL, and ALAB. For Arm itself, UBS expects market focus to shift gradually toward CY2028 profit forecasts, with long-term revenue potential possibly exceeding the company’s own guidance.

Risks

• Arm has limited visibility on short-term shipment ramp from hyperscale data center customers; data is lagging and management is occasionally surprised by ramp magnitude
• AGI CPU revenue realization depends on securing sufficient capacity; supply side is constraint
• Uncertainty remains if Arm can achieve 15% CPU device market share by FY31; META accounts for less than 10% of global server shipments and OAI/Stargate deployments use mostly NVDA racks

What to watch

• Progress of AGI CPU capacity expansion and ability to convert $2 billion backlog into actual revenue
• Deployment pace and core count increases of ARM CPUs at hyperscalers (AWS, Google, Microsoft, Nvidia, etc.)
• Sustainability of data center royalty revenue’s doubling-level YoY growth
• Company’s roadmap advancement towards 256-core+ CPUs and validation of AI-driven CPU demand intensity