Google raises capital expenditure again, benefiting the Asian server and AI infrastructure supply chain

Overview

Starting from Alphabet/Google's June quarter 2026 earnings, this report analyzes how its capital expenditure, cloud business, AI demand, and TPU deliveries affect the Asian technology hardware and server supply chain. The core view is that Google's higher capital expenditure, tight computing power supply, and strong AI cloud demand will benefit general servers, AI ASIC servers, GPU servers, and related component supply chains.

Core views

Google's capital expenditure plan is strong and has been raised continuously, indicating that demand for AI and cloud infrastructure remains far from fully satisfied. Server spending accounts for 60% of Google's Jun-Q capital expenditure, and with accelerating capital expenditure in the second half, this is expected to support revenue momentum for Google's server supply chain. Supply constraints and rising demand for third-party capacity are favorable for the pricing environment of NeoCloud vendors, server brands, and component manufacturers. TPU systems began delivery to customer data centers in 2Q26, but revenue contribution this year is limited; management expects acceleration by year-end and greater revenue contribution next year, so momentum for the TPU-related supply chain is more likely to be released in 2027.

Analysis framework

The report uses an earnings read-through approach, linking Google's Jun-Q capital expenditure, cloud revenue, backlog, AI usage, and TPU delivery pace with order and revenue prospects for Asian server ODMs, branded vendors, and component makers. At the same time, it combines J.P. Morgan's ratings and price information on covered stocks to identify potentially benefiting supply chain segments and individual names.

Methodology notes

• Earnings read-throughCapex-to-supply-chain read-through

  Map changes in hyperscale cloud vendors' capital expenditure to server supply chain demand

  Google's higher capital expenditure, the share of server spending, and data center capacity expansion are used as leading indicators for the revenue momentum of ODMs, AI ASIC servers, GPU servers, and component suppliers.

• Demand validationCloud backlog and AI usage indicators

  Use cloud revenue, backlog, and AI usage to validate infrastructure demand

  Google Cloud's high year-over-year revenue growth, backlog rising to US$514bn, and growth in Gemini DAU and token consumption are interpreted as evidence of strong demand for enterprise AI and AI infrastructure.

Asset mapping & comparison

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

• Wiwynn Corp (6669.TW)

  Top beneficiary among Taiwan ODM supply chain names

  Strengths

  Benefits from cloud capital expenditure expansion, AI server demand, and revenue momentum from Google's supply chain.

  Weaknesses

  The report does not provide company-specific earnings forecasts or valuation details.

  Comparison

  Compared with other ODMs, the report explicitly names it as the top pick in the Taiwan ODM supply chain.

  Risks

  If Google's capital expenditure pace slows, AI server deliveries are delayed, or price competition intensifies, revenue momentum may fall short of expectations.

• Hon Hai Precision (2317.TW)

  Potential beneficiary in the server supply chain

  Strengths

  Listed in the report as one of the key potential beneficiaries of Google's strong capital expenditure plan, with an OW rating.

  Weaknesses

  The report does not provide standalone order, revenue, or margin forecasts.

  Comparison

  Included among potential beneficiaries alongside server supply chain companies such as Inventec.

  Risks

  The extent of benefit may be affected by the pace of customer capital expenditure deployment, component costs, and supply chain competition.

• Elite Material Co (2383.TW)

  Beneficiary in server components

  Strengths

  Listed in the report as a beneficiary in the server component chain, with an OW rating.

  Weaknesses

  The report does not elaborate on specific products, capacity, or gross margin analysis.

  Comparison

  Along with Unimicron, ASPEED, Delta, and Lotes, it is one of the component manufacturers favored by the report.

  Risks

  If demand for AI server materials falls short of expectations or component prices fluctuate, earnings elasticity may be constrained.

• Unimicron (3037.TW)

  Beneficiary in server components

  Strengths

  Benefits from demand in both AI and non-AI server supply chains, with an OW rating.

  Weaknesses

  The report does not provide a company-level valuation methodology or target price.

  Comparison

  Along with EMC, it belongs to the beneficiary chain related to PCB/materials.

  Risks

  Demand realization, customer mix, and supply chain pricing pressure are the main uncertainties.

• ASPEED Technology Inc. (5274.TWO)

  Beneficiary in server components

  Strengths

  The report lists it as a favored server component manufacturer, with an OW rating.

  Weaknesses

  The report does not quantify its revenue linkage with Google's supply chain in the main text.

  Comparison

  Along with Delta, Unimicron, EMC, and Lotes, it is among the component-side beneficiaries.

  Risks

  Changes in server platform upgrade cycles and customer procurement timing may affect revenue.

• Delta Electronics, Inc. (2308.TW)

  Beneficiary in server components and power-related areas

  Strengths

  The report lists it as a potential beneficiary and favored component manufacturer, with an OW rating.

  Weaknesses

  The report does not break down Google's demand contribution to Delta's specific business lines.

  Comparison

  Compared with ODM manufacturers, its benefits are more tilted toward components and infrastructure.

  Risks

  Component costs, customer qualification timing, and uncertainty in capital expenditure deployment.

• Inventec (2356.TW)

  Potential beneficiary in the server supply chain

  Strengths

  The report lists it as one of the potential beneficiaries of Google's strong capital expenditure.

  Weaknesses

  It is rated N, indicating the report sees its relative investment appeal as weaker than OW-rated names.

  Comparison

  Weaker than the clearly preferred Wiwynn and several OW-rated component names in the report.

  Risks

  The neutral rating reflects that the potential benefit may already be partly offset by valuation or fundamental constraints.

• Lenovo Group Limited (0992.HK)

  Potential beneficiary among server brands

  Strengths

  The report believes server brands can benefit from strong NeoCloud demand and favorable pricing dynamics.

  Weaknesses

  It is rated N, and the report does not provide detailed company-level financial upside.

  Comparison

  Along with ASUSTek, it is included in the discussion of server brand beneficiaries, but with a different investment rating.

  Risks

  Competition in branded servers, price changes, and fluctuations in NeoCloud demand.

• ASUSTek Computer (2357.TW)

  Potential beneficiary among server brands

  Strengths

  The report believes server brands may benefit from strong NeoCloud demand and better pricing.

  Weaknesses

  It is rated UW, indicating that despite a positive industry read-through, the company-level relative rating is negative.

  Comparison

  Like Lenovo, it is on the brand side, but Lenovo is rated N while ASUSTek is rated UW.

  Risks

  Company-specific fundamentals or valuation pressure may offset industry tailwinds.

• Alphabet (GOOG)

  Source of the earnings and capital expenditure read-through

  Strengths

  Strong cloud business growth, improved margins, high backlog, and rapid AI usage growth, with an OW rating.

  Weaknesses

  The continued rise from an already high capital expenditure base may create pressure on investment intensity and return cycles.

  Comparison

  As a hyperscale cloud vendor, changes in its capital expenditure have leading-indicator significance for the Asian server supply chain.

  Risks

  Computing power supply, returns on AI investment, cloud revenue recognition pace, and capital expenditure efficiency.

Key data

• Google Jun-Q CapexUS$45bnUp 100% year-over-year and 26% quarter-over-quarter, in line with market expectations.
• Share of capital expenditure spent on servers60%The remaining 40% was used for data center construction and networking equipment.
• Google full-year 2026 Capex guidanceUS$195-205bnRaised from US$180-190bn; midpoint up about 120% year-over-year.
• 2H26 Capex implied growthabout 50% HoHThe raised guidance implies a significant acceleration in second-half capital expenditure versus the first half.
• Google Cloud Jun-Q revenue growth+24% QoQ / +82% YoYThe year-over-year growth rate further accelerated from 63% in the previous quarter and was 10% above market expectations.
• Google Cloud margin35.6%Higher than 1Q26's 32.9% and 2Q25's 20.7%.
• Google Cloud backlogUS$514bnIncreased by more than US$50bn quarter-over-quarter in 2Q26, and management expects 50% to be recognized as revenue over the next two years.
• Gemini usageDAU tripled year-over-year; 2Q26 token consumption rose 38% quarter-over-quarterReflects strong AI application usage, supporting demand for computing power and servers.
• Potential beneficiariesHon Hai, EMC, Unimicron, ASPEED, Delta, Inventec, Wiwynn, Lotes, Lenovo, ASUSTekThe report specifically highlights Wiwynn as the top pick among Taiwan ODM supply chain names.

Impact & implications

For investors, Google's continuous upward revisions to capital expenditure reinforce the view that the AI infrastructure cycle remains in an expansion phase, especially benefiting Asian supply chain players with capabilities in complete servers, AI ASIC/GPU servers, PCBs, connectors, power supplies, and thermal management. In the short term, accelerating capital expenditure in 2H26 is expected to improve order and revenue momentum; in the medium term, TPU system volume expansion next year may bring stronger opportunities for the specialized AI server supply chain.

Risks

• Google or other CSPs may see capital expenditure pace come in below current guidance.
• Strong AI infrastructure demand but supply constraints may delay revenue recognition for the server supply chain.
• Rising component costs may alter the server spending mix and compress margins for some suppliers.
• TPU business has limited revenue contribution this year; if next year's volume ramp falls short of expectations, momentum for the related supply chain may be weaker than the report suggests.
• Changes in NeoCloud demand or pricing dynamics may affect the degree of benefit for server brands and component manufacturers.
• The report does not provide target prices, earnings forecasts, or valuation sensitivities for each beneficiary in the main text, so stock-specific judgments need to be combined with dedicated company reports.

What to watch

• Google's 2H26 capital expenditure execution progress and whether 2027 capital expenditure expectations continue to be raised.
• The quarterly pace of Google Cloud backlog conversion into revenue, especially whether it reaches a quarterly revenue run-rate above US$30bn.
• The magnitude of Google's increase in third-party capacity usage in 3Q26 and its impact on NeoCloud pricing.
• Whether Gemini DAU and token consumption growth continues, validating the strength of AI inference demand.
• The ramp-up of TPU system deliveries by year-end and in 2027, and the corresponding revenue recognition of supply chain vendors.
• Changes in orders, capacity utilization, and gross margins for server ODMs, PCB, connector, power supply, and thermal management vendors.