Durable AI Demand Amid Supply Constraints: BofA Raises Target Prices for Multiple IT Hardware Stocks

Overview

This report summarizes key takeaways from the IT Hardware sector at the BofA 2026 Global Technology Conference. The overall tone is optimistic, viewing AI infrastructure demand as not only strong but also becoming broader, deeper, and more durable. On the demand side, activity is spreading from pure GPU training to Inference and Agentic AI workloads, directly benefiting traditional CPU servers, enterprise storage, and the HDD market. On the supply side, restricted availability of key components like memory and HDDs is driving increased industry pricing power and business model optimization. Consequently, the report raised target prices for IBM, NetApp, Seagate, and Western Digital, maintained positive views on companies including Dell, HPE, and DigitalOcean, while flagging cost pressures in the PC segment and governance risks at specific firms.

Core views

Shift in AI Demand Structure and Traditional Computing Recovery: The report emphasizes that Inference and Agentic AI are becoming new drivers for traditional computing. These workloads shift some compute capacity from parallel GPU processing to serial CPU processing, benefiting traditional server vendors like Dell and HPE, as well as IBM's mainframe platform. Dell management noted that its F27 revenue guidance is limited by supply rather than demand, with customer planning extending to 2028; driven by AI inference use cases, generational growth for the IBM Z17 mainframe has accelerated to approximately 135%. This indicates that AI tailwinds have spilled over to non-GPU hardware, and demand exhibits multi-year durability rather than being a short-term pull-forward. Structural Tightness in Storage and HDD Markets: Storage has become a strategic bottleneck in AI architectures. Western Digital (WDC) raised its HDD EB growth forecast for the next 3-5 years to over 25%, while Seagate (STX) confirmed order visibility of 4-5 quarters, with nearline capacity allocated through 2027. Both companies emphasized improving areal density via technology (e.g., HAMR) rather than simply adding unit capacity to maintain pricing discipline and margin expansion. SanDisk is reducing NAND cycle volatility significantly by locking in long-term volume and price contracts through 'New Business Models' (NBMs), which include fixed pricing and floor clauses. In enterprise storage, both NetApp and Everpure reported significant increases in AI-related wins, driven by strong all-flash upgrade and hybrid cloud demand. Supply Chain Tightness and Channel Value Re-rating: The overall component environment is tightening, with signs of cost inflation across memory, CPUs, and HDDs. This complexity enhances the strategic value of distributors (Arrow, Avnet) and EMS/ODM vendors (Flex, Celestica). Recovery at Arrow and Avnet is driven by unit volumes rather than mere price increases, with significant backlog growth; as a leading white-box networking ODM, Celestica sees robust demand for 400G/800G and upcoming 1.6T switches, achieving ROIC near 50%. Flex's planned SpinCo focuses on AI data center power and cooling infrastructure, with FY27 booking rates already at 90% and growth expected around 70%, reflecting the profound impact of rising AI compute density on power architecture. Stock Divergence and Specific Risks: Despite the positive industry backdrop, individual stock performance diverges. Although HP Inc. has an AI PC lineup, it faces severe component cost inflation pressure in the second half, potentially causing Personal Systems margins to hit a trough; rated Underperform. Super Micro Computer (SMCI), despite being in the AI server sweet spot, is also rated Underperform due to unresolved export compliance reviews and material weaknesses in internal controls. Conversely, DigitalOcean, leveraging high retention among AI-native customers and higher ARR per MW, is viewed as being in the early stages of a revenue acceleration cycle; Nutanix continues to benefit from VMware displacement opportunities, although near-term revenue recognition is weighed down by server hardware shortages.

Analysis framework

The report employs a 'bottom-up validation of top-down themes' analytical approach. It first establishes the macro assumption that AI demand is 'broad, deep, and durable,' then cross-validates this through management feedback from over a dozen supply chain companies at the conference. The analytical logic follows a 'Demand Diffusion → Supply Constraints → Earnings Quality' trajectory: focusing not just on revenue growth, but emphasizing how order visibility (Backlog/Book-to-Bill), pricing power (ASP trends), technological iteration (HAMR/1.6T), and business model transformation (NBMs/Servicization) support margins. Regarding valuation, the report favors assigning higher multiple premiums to companies demonstrating structural growth and earnings resilience, while applying discounts to those facing cost squeezes or governance flaws.

Methodology notes

• Industry/Sector Analysis FrameworkSupply-demand framework

  Supply-side constraints as the core driver of earnings improvement

  In mature industries like HDDs and storage chips, the report no longer tracks shipment growth in isolation but focuses on how 'Supply Restricted' conditions grant pricing power to leading vendors. When demand exceeds supply and capacity expansion is limited by technology (e.g., HAMR yields) or capital discipline, companies can achieve profit growth outpacing revenue growth through price hikes and product mix optimization—a key signal for cyclical stock valuation re-rating.

• Industry/Sector Analysis FrameworkPenetration S-curve

  Migration of AI workload penetration from training to inference/agentic stages

  The report breaks down AI demand into different stages, noting that the market is currently at an inflection point penetrating from 'model training' to 'enterprise inference and agentic applications.' The technical characteristics of this stage (serial processing, frequent data read/write) determine that beneficiaries expand from pure GPUs to CPUs, storage, and networking equipment, forming the industrial logic basis for understanding IT hardware sector rotation and value re-rating.

• Company Fundamentals & Financial FrameworkEarnings Quality Analysis

  Distinguishing 'volume-driven' vs. 'price-driven' recovery quality

  When assessing the cyclical position of electronic component distributors (e.g., Arrow, Avnet), the report specifically distinguishes whether recovery is driven by 'unit sales' or 'price increases.' Volume-driven recovery typically signals genuine end-demand warming and more sustainable earnings improvement; whereas recovery driven solely by shortage-induced price hikes often carries subsequent inventory impairment risks. This distinction is a critical methodology for determining the margin of safety when timing cyclical stock entries.

Asset mapping & comparison

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

• IBM (IBM)

  Beneficiary: AI inference and modernization tools accelerate demand for Z-series mainframes and storage; vertically integrated stack possesses pricing power

  Strengths

  Z17 generational growth accelerated to 135%; double-digit storage hardware growth; Red Hat and software subscriptions improve revenue quality

  Comparison

  Compared to pure x86 server vendors, IBM offers higher stickiness and better unit economics in mission-critical AI inference scenarios

  Risks

  Failure to execute growth roadmap; M&A integration risks; currency fluctuations

• Western Digital (WDC)

  Beneficiary: Structural tightening in HDD supply-demand; AI data creation cycle drives EB growth above expectations

  Strengths

  EB growth forecast raised to >25%; gross margin guidance 51-52%; net cash position; rising UltraSMR penetration

  Comparison

  Similar to STX in benefiting from HDD oligopoly, though slightly different pace in ePMR/HAMR transition

  Risks

  Cloud/data center spending slowdown; HAMR certification delays; severe NAND price volatility

• Seagate Technology (STX)

  Beneficiary: HDD order visibility reaches 4-5 quarters; HAMR technology leadership supports margin expansion

  Strengths

  Nearline capacity allocated through 2027; 40TB HAMR certified; quarterly free cash flow near $1bn

  Comparison

  Leads peers in HAMR mass production progress, but requires continuous validation of yield and cost reduction curves

  Risks

  Deterioration in industry pricing; NAND substitution effect; decline in desktop/notebook unit sales

• Dell Technologies (DELL)

  Beneficiary: Dual engines of AI servers and traditional computing; storage mix optimization offsets gross margin dilution

  Strengths

  Record ISG orders; visibility to 2028; IP storage outperformed market for 5 consecutive quarters; operating profit growing faster than revenue

  Weaknesses

  F27 guidance constrained by supply, potentially capping near-term revenue upside

  Comparison

  Clearer advantages in proprietary storage IP and enterprise direct sales channels compared to HPE

  Risks

  Global economic slowdown; tariffs and trade wars; Intel processor supply issues

• DigitalOcean (DOCN)

  Beneficiary: High growth from AI-native customers; transition from bare metal to high-value inference services

  Strengths

  AI customer revenue up >200%; higher ARR per MW; demand is 3-4x capacity under construction; 2027 growth outlook 50%+

  Weaknesses

  Scale remains far smaller than hyperscale cloud providers

  Comparison

  Offers cost-effectiveness and ease-of-use advantages over AWS/Azure for SMB AI developers and inference scenarios

  Risks

  AI adoption returns below expectations; capacity ramp execution risks; graduation/churn of large customers

• Super Micro Computer (SMCI)

  Impaired/Risk: Despite being in the AI track, valuation is suppressed by governance and compliance issues

  Strengths

  Large-scale AI rack manufacturing; strong customization design capabilities; balanced liquid/air cooling capacity

  Weaknesses

  Unresolved export compliance review; material weakness in internal controls; gross margin pressure; working capital intensive

  Comparison

  Disadvantaged in enterprise services and compliance trust relative to Dell/HPE

  Risks

  Export controls and litigation shocks; component availability limiting revenue; financing needs diluting equity

• HP Inc. (HPQ)

  Impaired: Component cost inflation erodes Personal Systems margins; AI PC monetization is premature

  Strengths

  Global scale and supplier diversification; expanded AI PC product line; relatively stable printing business

  Weaknesses

  Significant memory/storage cost surge in 2H26; PC unit sales may decline high-teens; limited gross margin contribution from AI PCs

  Comparison

  Weaker enterprise hedging mechanisms and high-end storage mix compared to Dell

  Risks

  Demand destruction due to failed pricing pass-through; structural decline in printing market; unfavorable FX

Key data

• WDC HDD EB Growth Forecast>25% CAGR3-5 year CAGR, significantly raised from previous mid-teens expectations
• Dell ISG Demand VisibilityExtends to 2028Covers GPUs, core servers, storage, and networking; F27 guidance constrained by supply
• Seagate HAMR Product Roadmap40TB Certified / 50TB Late 2027Expected ~25% annual capacity growth per drive without increasing unit output
• DigitalOcean AI Customer Revenue Growth>200%Zero churn in $1mn+ customer base; 2027 growth outlook raised to 50%+
• Flex SpinCo FY27 Growth Expectation~70%FY27 booking rate ~90%; FY28 booking rate ~70%
• Celestica ROIC~50%Significantly above 12-13% cost of capital, reflecting higher mix of high-return ODM business
• IBM Z17 Generational Growth~135%Accelerated from 110%-125% in prior generations, driven by AI inference use cases
• SanDisk NBMs Contract Coverage>$11bn Financial GuaranteeIncludes $400mn prepayment; 1/3 of F27 revenue already locked via new model

Impact & implications

The report suggests the IT hardware industry is entering a structural upcycle driven by AI, rather than a transient inventory restocking. For investors, this implies stock selection logic must shift from pure 'AI concepts' to 'earnings realization capability': companies with differentiated IP, pricing power amidst supply constraints, and long-term contract coverage will command valuation premiums. The storage and HDD sectors are viewed as the most definitive sub-sectors currently due to optimal supply-demand dynamics; traditional server and networking vendors are poised for a second growth curve driven by incremental Agentic AI demand. Meanwhile, persistent supply chain tightness triggers a value re-rating for channel partners with complex logistics management and inventory service capabilities. However, profit erosion for downstream terminal brands (e.g., PCs) due to cost inflation, along with governance concerns at select companies, remain structural vulnerabilities to monitor.

Risks

• Macroeconomic slowdown or recession leading to reduced enterprise IT spending
• Component costs (memory, CPU) inflating beyond expectations without ability to pass through downstream
• Demand correction triggered by lower-than-expected returns on AI capex
• Geopolitical and trade tariff policies disrupting supply chains
• Governance and compliance risk materialization at specific companies (e.g., SMCI)
• Mass production or yield of new technologies like HAMR falling short of expectations
• NAND/HDD prices retreating due to overly rapid supply release

What to watch

• Quarterly ASP and gross margin trends for HDD and storage vendors to validate pricing discipline
• Growth in non-AI/GPU product lines for traditional server vendors like Dell and HPE
• Book-to-Bill ratios and inventory turnover changes for distributors (ARW, AVT)
• Results of SMCI's independent compliance review and progress on internal control remediation
• Pace of new MW deployments at DigitalOcean and realization of ARR per MW
• Contracting ratio and evolution of terms for SanDisk's New Business Models (NBMs)
• Progress of Flex separation transaction and financial performance of SpinCo/RemainCo post-independent operation