Goldman Survey: AI Drives Surge in Hardware Spending, Overall IT Budget Moderately Slows

Overview

This report is based on Goldman Sachs' May 2026 survey of IT spending among Fortune 2000 global CIOs. Core conclusions show that while AI-related infrastructure spending momentum is strong, driving significant growth in server, storage, and PC hardware spending, overall IT spending intent has moderated slightly. Both overall IT spending index and capital expenditure index have declined consecutively, reflecting cautious attitudes during macroeconomic uncertainty and AI transformation assessment periods. The report details spending trends and vendor competition landscapes in sub-sectors such as software, hardware, services, and security.

Core views

Overall spending intent cools moderately, but structural divergence intensifies. Both overall and capital expenditure indices decline: May 2026 survey shows, overall IT spending index dropped from 68.0 in Nov 2025 to 65.0, IT capital expenditure index dropped from 65.5 to 59.0. Although both are below 10-year average, they remain above 50 expansion/contraction critical point, indicating spending is still growing but momentum weakened. This slowdown may stem from enterprises' evaluation of AI impacts, leading to non-AI category spending squeezed out, or deferring major project commitments before the impact of AI on core budgets becomes clear. Hardware spending rebounded significantly, driven by AI and supply tightness: Server, storage, and PC spending expectations improved dramatically. 60% of respondents plan to increase server spending (previously 37%), 75% plan to increase storage spending (previously 47%), 69% plan to increase PC spending (previously 36%). This trend mainly driven by AI infrastructure demand, 2020-21 device aging refresh, and DRAM/NAND supply tightness leading to early purchasing and ASP (Average Selling Price) increases. In contrast, network equipment spending expectations deteriorated; only 35% of respondents plan increased spending, below 5-year average. AI budget share gradually increasing, funding sources mainly 'reallocation': In short term (next 12 months), 88% of CIOs expect GenAI accounts for less than 10% of IT budget (weighted avg 5%); but after 3 years, 42% of respondents expect GenAI share will exceed 10% (weighted avg 10%). Notably, AI inference costs currently account for only 1-5% of IT budget, and approx 2/3 of funds come from reallocation of existing budgets (mainly application software and efficiency improvement projects), only 1/3 is pure new budget. This indicates AI spending is squeezing short-term software and headcount budgets. Software Sector: 'Buy' preferred over 'Build', cloud migration proceeds steadily. Regarding market concerns view that 'AI reduces code costs leading enterprises to shift to self-built software', survey shows 56% of respondents plan to reduce in-house build, increase purchase packaged software, only 17% plan to increase in-house build, alleviating investor concerns about SaaS vendors. Public cloud workload penetration rate expected to rise from current 27% to 45-50% within 3 years, GenAI will become incremental catalyst for cloud migration, expected GenAI cloud workloads grow 2.5x within 3 years. Security Software: AI brings minor increment, platform vendors benefit. 69% of respondents expect GenAI to increase their cybersecurity budget, weighted average increase magnitude 3.28%. Firewall refresh cycles flattening, showing more multi-year distributed replacement characteristics. In vendor selection, Microsoft, Cloudflare and Palo Alto Networks considered vendors with highest budget growth in next 3 years due to platform integration capabilities, while traditional point solution vendors face pressure.

Analysis framework

This report adopts high-frequency questionnaire surveys, constructing an 'IT Spending Index' similar to ISM Manufacturing Index to quantify CIO spending intent. Index > 50 represents expected spending growth, < 50 represents contraction. Report identifies marginal changes in spending trends by comparing data changes between Nov 2025 and May 2026. Simultaneously, report applies 'Volume-Price Decomposition' logic to analyze hardware spending (e.g., distinguishing PC volume expectations vs ASP increase impact), and through 'Budget Source Analysis' (new vs reallocation) to assess AI spending sustainability and its crowding-out effect on other IT categories. In competitive landscape analysis, report combines 'Wallet Share' net change metrics to evaluate relative strengths of each vendor in niche fields.

Methodology notes

• Industry/Industrial Analysis FrameworkSupply and Demand Framework

  IT Spending Index Construction Method

  Similar to ISM Manufacturing Index, calculating weighted proportion of respondents expecting increased, unchanged, and decreased spending to derive a diffusion index between 0-100. Index > 50 implies industry expansion zone, < 50 implies contraction, used to quickly determine industry sentiment direction.

• Industry/Industrial Analysis FrameworkVolume-price decomposition

  Volume-Price Factors in Hardware Spending

  When analyzing PC and server spending, distinguish the impact of 'increased procurement volume' vs 'average selling price (ASP) increase'. For example, large expected PC spending increase is partly because enterprises purchased early to cope with price hikes caused by DRAM/NAND shortages, and ASP rises due to suppliers passing on costs, rather than simple demand surge.

• Company Fundamentals and Financial FrameworkOthers

  Budget Reallocation Effect (Cannibalization)

  When analyzing spending sources for new technologies (like AI), distinguish between 'incremental budget' and 'existing stock reallocation'. The report states most AI spending comes from compressing other software licenses or labor costs, meaning AI giants' high growth may correspond to revenue pressure on traditional software vendors, i.e., 'zero-sum game' characteristics.

Asset mapping & comparison

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

• Microsoft (MSFT)

  Benefit

  Strengths

  Maintains leadership in IaaS area (46% share); Copilot penetration low but sentiment improves; E7 upgrade cycle expected to bring multi-year revenue boost; considered one of vendors with highest budget growth in security area.

  Weaknesses

  Current Copilot penetration still low (≤10%); user feedback needs time to validate.

  Comparison

  Compared to other Hyperscalers, Microsoft has risk resistance advantage in ecosystem integration and fleet-level fungibility.

  Risks

  If AI application implementation falls short of expectations, E7 upgrade cycle may be delayed.

• Amazon (AMZN)

  Benefit

  Strengths

  IaaS share stable second (30%); expected to rise to 14% in 3-year expectation (strategic voting share).

  Comparison

  Jointly dominates cloud market with Microsoft, but direct reach capability at enterprise-level AI application layer slightly inferior to Microsoft.

• Google (GOOGL)

  Benefit

  Strengths

  Strategic voting rank rose to third; Gemini Enterprise made progress in enterprise search interface; AI positioning stronger than expected.

  Comparison

  Increased visibility at AI model level; may erode other vendors' AI budget share.

• Cisco (CSCO)

  Hurt/Neutral

  Strengths

  Still maintains slight net gain in network equipment wallet share (+3%).

  Weaknesses

  Network equipment overall spending expectations deteriorated; Cisco budget growth expectation in security area showed the largest deterioration.

  Comparison

  Compared to Palo Alto and Cloudflare, Cisco is at disadvantage in security platform integration trend.

  Risks

  Risk of cyclical downturn in network spending.

• Palo Alto Networks (PANW)

  Benefit

  Strengths

  Considered one of vendors with highest security budget growth in next 3 years; benefits from SASE/SOC integration trend.

  Comparison

  Compared to traditional firewall vendors, platform advantages are more obvious.

• Snowflake (SNOW)

  Benefit

  Strengths

  As data platform, benefits from AI-native dev tools reducing migration friction; considered 2nd tier beneficiary.

  Weaknesses

  Ranked behind Hyperscalers and frontier model vendors in survey on AI wallet share capture capability.

  Comparison

  Compared to Databricks and Palantir, Snowflake has slightly higher positive positioning in survey, but still belongs to 2nd tier.

  Risks

  If enterprises shift to self-built data stacks, may face competitive pressure.

Key data

• Overall IT Spending Index65.0Dropped from 68.0 in Nov 2025, but still above 50 expansion/contraction line
• IT Capital Expenditure Index59.0Dropped from 65.5 in Nov 2025
• Proportion of Respondents Planning to Increase Server Spending60%Significant increase from previous 37%
• Proportion of Respondents Planning to Increase Storage Spending75%Significant increase from previous 47%
• GenAI Share of IT Budget (3-Year Expectation)10%Weighted average, 42% respondents expect >10%
• AI Inference Cost Funding Source2/3 ReallocationMainly from application software and efficiency projects, 1/3 is new budget
• Respondents Preferring 'Purchase' over 'Self-Build' Software56%Only 17% plan to increase in-house build, alleviating SaaS vendor concerns
• Security Budget Increase Driven by GenAI3.28%Weighted average increase magnitude, slight increase from 3.05% half year ago

Impact & implications

For hardware vendors, server, storage, and PC segments see short-term sentiment upturn; enterprises with AI server supply capacity and storage solutions benefit from volume and price simultaneous rise. For cloud vendors, although GenAI hasn't immediately triggered large-scale budget expansion, it is becoming a long-term catalyst for cloud migration; Microsoft Azure and AWS continue to consolidate share with ecosystem advantages, Google Cloud status improves. For software vendors, the 'buy over build' trend is a positive signal, but risks of clients compressing traditional software license budgets due to AI spending must be monitored. For cybersecurity vendors, direct revenue increment from AI is small (~3%), but platform vendors able to provide more integrated security architectures will hold advantage in budget consolidation; traditional point-product vendors face greater competitive pressure.

Risks

• Macroeconomic uncertainty leads enterprises to further tighten non-AI IT spending.
• AI ROI validation falls short of expectations, leading to weak subsequent budget additions.
• Supply chain (e.g., DRAM/NAND) continued tightness causes hardware delivery delays or excessive costs.
• Geopolitical or trade tariff environment fluctuations affect global IT equipment procurement.

What to watch

• Actual growth speed of cloud workloads during GenAI transition from experimental phase to production phase.
• Actual adoption speed of Microsoft E7 upgrade cycle and its effect on M365 ARPU improvement.
• Whether AI inference costs continuously squeeze traditional app software budgets, causing SaaS vendor pricing power damage.
• In cybersecurity field, substantial security spending increment brought after Agentic AI enters production environment.