Google raises 2026 capex, benefiting AI and non-AI server supply chains

Overview

Based on Alphabet/Google’s June 2026 quarter results, this report analyzes how its capex, cloud business growth, AI demand, third-party compute usage, and TPU delivery pace affect the Asian server supply chain. The core conclusion is that Google’s higher full-year capex and continued compute supply constraints will drive revenue momentum for servers, AI ASICs, GPU servers, and key components in 2H26 and 2027.

Core views

Google’s capex intensity and cloud business demand both exceeded or met market expectations, showing that AI infrastructure investment remains in an upward cycle. With server spending accounting for about 60% of capex, alongside increased third-party capacity usage, Gemini DAU tripling YoY, and token consumption rising 38% QoQ, JPMorgan judges that pricing and order momentum for server brands, ODMs, and component makers will benefit. The report is especially positive on Wiwynn, while also favoring Delta, ASPEED, Unimicron, EMC, and Lotes, and believes Lenovo and ASUSTek may also benefit from NeoCloud demand.

Analysis framework

The report uses an earnings read-through approach, deriving the revenue and pricing impact on the Asian technology hardware supply chain from indicators such as Google’s capex, cloud revenue, order backlog, AI product demand, TPU deliveries, and capacity constraints, and mapping these to server ODMs, server brands, AI ASIC/GPU servers, and component companies.

Methodology notes

• Earnings interpretationEarnings read-through

  Deriving supply chain conditions from Google capex and cloud business metrics

  Rather than simply evaluating Alphabet’s share price, the report channels changes in Google’s capex, cloud revenue, and compute demand into the server supply chain.

• Supply chain analysisCapex-to-supply-chain mapping

  Mapping capex to server and component revenue

  Servers account for about 60% of Google’s capex, so the capex increase is seen as a leading signal for order growth for server ODMs, brands, and key component suppliers.

• Demand and capacity analysisCapacity constraint analysis

  Compute supply-demand constraints and third-party capacity usage

  Google still sees supply constraints and expects to increase third-party capacity usage in 3Q26, which is interpreted as a signal of an improving pricing environment for NeoCloud vendors, server brands, and component suppliers.

Asset mapping & comparison

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

• ALPHABET INC / Google

  Source of results and demand-side driver

  Strengths

  Higher capex, accelerating cloud revenue, expanding backlog, and strong AI product demand.

  Weaknesses

  The report does not discuss Alphabet’s valuation or target price in detail here; the focus is on supply chain read-through.

  Comparison

  Google Cloud YoY growth accelerated from 63% in the previous quarter to 82% and exceeded market expectations.

  Risks

  The sustainability of capex under a high base, the pace of AI demand realization, and the speed of easing supply constraints still need to be monitored.

• Wiwynn Corp (6669.TW)

  Top beneficiary in Taiwan’s ODM supply chain

  Strengths

  Explicitly identified in the report as the top pick in Taiwan’s ODM supply chain, benefiting from accelerated Google server capex.

  Weaknesses

  Dependent on order timing from large CSPs and AI servers.

  Comparison

  The report assigns it a higher preference relative to other ODMs.

  Risks

  Changes in customer capex timing, order delivery, and component cost fluctuations.

• Hon Hai Precision (2317.TW)

  Potential beneficiary in the server supply chain

  Strengths

  Listed as one of the companies that could benefit from Google’s strong capex plan.

  Weaknesses

  The report does not provide new company-level earnings estimates.

  Comparison

  Like Inventec and Wiwynn, it is a server supply chain-related company.

  Risks

  Server demand transmission falling short of expectations or changes in customer order allocation.

• Elite Material Co / EMC (2383.TW)

  Server component beneficiary

  Strengths

  Listed as a potential beneficiary, and the report also states a preference for server component supplier EMC.

  Weaknesses

  Sensitive to demand for high-end server materials and customer qualification timing.

  Comparison

  Like Unimicron, ASPEED, Delta, and Lotes, it belongs to the report’s favored component names.

  Risks

  Component costs, yield rates, demand volatility, and intensifying competition.

• Unimicron (3037.TW)

  Server component beneficiary

  Strengths

  Benefiting from demand for AI and non-AI server-related components.

  Weaknesses

  The report does not disclose a new target price or earnings forecast in this report.

  Comparison

  Along with EMC, ASPEED, Delta, and Lotes, it is listed as a favored component maker.

  Risks

  Capex delays, product mix changes, and pricing pressure.

• ASPEED Technology Inc. (5274.TWO)

  Server component beneficiary

  Strengths

  Listed as a potential beneficiary and as a favored server component supplier.

  Weaknesses

  Company performance depends on server platform shipments and product penetration.

  Comparison

  It is disclosed as OW in the company list in the report.

  Risks

  Platform transitions, customer inventory, and changes in server shipment timing.

• Delta Electronics, Inc. (2308.TW)

  Beneficiary in server power and data center-related components

  Strengths

  Listed as a potential beneficiary, and the report explicitly favors Delta.

  Weaknesses

  Demand transmission is affected by the pace of data center construction and product mix.

  Comparison

  It is disclosed as OW in the company list in the report.

  Risks

  Delays in data center construction, cost fluctuations, and competitive pressure.

• Inventec (2356.TW)

  Potential beneficiary in the server supply chain

  Strengths

  Included in the list of potential beneficiaries from the strong capex plan.

  Weaknesses

  The report discloses a rating of N, indicating lower preference than Wiwynn.

  Comparison

  Also part of the server supply chain, but not the report’s top pick.

  Risks

  Order wins and margin improvement falling short of expectations.

• Lenovo Group Limited (0992.HK)

  Server brand beneficiary

  Strengths

  May benefit from strong NeoCloud demand and favorable pricing dynamics.

  Weaknesses

  The report discloses a rating of N.

  Comparison

  Like ASUSTek, it is a server brand that may benefit.

  Risks

  NeoCloud demand volatility, price competition, and product mix changes.

• ASUSTek Computer (2357.TW)

  Potential beneficiary among server brands

  Strengths

  May benefit from strong NeoCloud demand and the pricing environment.

  Weaknesses

  The report discloses a UW rating, indicating a relatively cautious overall investment view.

  Comparison

  Like Lenovo, it is in the server brand beneficiary category, but with a lower rating.

  Risks

  Uncertainty in demand realization, competitive pressure, and valuation risk.

• Lotes (3533.TW)

  Server component beneficiary

  Strengths

  The report explicitly says it also favors server component suppliers such as Lotes.

  Weaknesses

  This report does not provide detailed company-level estimates.

  Comparison

  Like Delta, ASPEED, Unimicron, and EMC, it belongs to the favored component category.

  Risks

  Customer order timing, product upgrades, and changes in the competitive landscape.

Key data

• Google June quarter capexUS$45bnUp 26% QoQ and 100% YoY, in line with market expectations.
• Server spending as a share of capex60%The remaining 40% is used for data center construction and networking equipment.
• Full-year 2026 capex guidanceUS$195-205bnRaised from the previous US$180-190bn, with the midpoint up about 120% YoY.
• Implied 2H26 capex growth约 50% HoHThe higher full-year guidance implies a clear acceleration in the second half versus the first half.
• Google Gemini DAU同比增长 3 倍Reflecting strong AI application demand.
• Token consumption rate环比增长 38%Token consumption accelerated in 2Q26, supporting AI compute demand.
• Google Cloud revenue growth环比增长 24%，同比增长 82%The YoY growth rate accelerated significantly from 63% in the previous quarter and exceeded market expectations by 10%.
• Google Cloud margin35.6%Higher than 32.9% in 1Q26 and 20.7% in 2Q25.
• Google Cloud backlogUS$514bnIncreased by more than US$50bn QoQ in 2Q26, mainly driven by enterprise AI product demand.
• Expected revenue recognition from backlog50% recognized over the next 2 yearsThis implies Google Cloud’s quarterly revenue run-rate may rise from the current roughly US$25bn to above US$30bn.

Impact & implications

Google’s capex increase reinforces the durability of demand for AI infrastructure and data center hardware. From an investment perspective, the most direct beneficiaries include general servers, AI ASIC servers, GPU servers, server brands, ODMs, and high-value components. Supply constraints and increased third-party capacity usage may also improve the pricing environment for NeoCloud-related suppliers and server brands.

Risks

• Although Google has raised capex, supply chain revenue momentum may come in below expectations if AI demand, cloud revenue, or the pace of order recognition slows.
• If compute supply constraints ease quickly, the favorable pricing dynamics for NeoCloud and server brands may weaken.
• TPU systems are expected to contribute limited revenue this year, and more visible TPU-related supply chain revenue momentum mainly depends on a ramp in the coming year.
• Rising server component costs could drive changes in capex mix and may also compress margins for some suppliers.
• The report is mainly an earnings read-through and does not provide full valuations, target prices, or earnings forecasts for all beneficiary companies.

What to watch

• Whether Google’s actual 2H26 capex execution reaches about 50% HoH growth.
• The pace at which Google Cloud backlog converts into recognized revenue, especially whether quarterly revenue run-rate rises above US$30bn.
• The extent of increased third-party capacity usage in 3Q26 and its impact on NeoCloud pricing.
• Whether Gemini DAU, token consumption, and enterprise AI product demand can continue to grow.
• The ramp-up pace of TPU systems from late 2026 into 2027 and the related supply chain revenue contribution.
• Changes in orders, revenue, and gross margin for suppliers such as Wiwynn, Delta, ASPEED, Unimicron, EMC, and Lotes.