U.S. Cloud Giants Raise Capex Again; AI Optical Communications Upcycle Expected to Extend to 2028

Overview

By interpreting the 2Q 2026 capital expenditure, full-year guidance, and management commentary of four major U.S. cloud service providers—Alphabet, Amazon, Meta, and Microsoft—the report concludes that AI infrastructure construction remains in a multi-year expansion cycle. Total quarterly spending was in line with expectations, but forecasts for 2026 and especially 2027 continue to be raised. The shortage of AI computing power supply may persist into 2028, creating a structural tailwind for China's AI optical communications and related infrastructure supply chains.

Core views

The core conclusions include: first, the four major cloud service providers' quarterly capital expenditure grew 79% year over year, indicating that AI infrastructure investment has not slowed; second, the 2027 capital expenditure forecast was raised by 15.1% from the previous estimate, and all four companies expect spending to continue growing that year, reinforcing the view of a multi-year capex supercycle; third, current AI capacity remains broadly in short supply, and demand visibility has extended to 2028; fourth, the optical communications industry is expected to achieve a compound growth rate of about 65% over the next three years, and sector valuation re-rating may be jointly driven by mid-2026 results, September industry events, the 2027 outlook, progress in co-packaged optics and near-packaged optics, and sequential shipment improvement in 2H 2026.

Analysis framework

The report adopts a cross-market mapping approach, first summarizing the quarterly capital expenditure and year-over-year growth rates of the four U.S. cloud service providers, then comparing the latest full-year guidance with prior forecasts, and assessing cycle sustainability based on management commentary on capacity supply and demand, 2027 investment direction, and 2028 demand. At the stock level, the report uses P/E ratios, EPS, and enterprise value multiples for target price valuation, and maps cloud investment trends to the AI optical communications supply chain.

Methodology notes

• Industry Cycle AnalysisCapital Expenditure Cycle Analysis

  Assess upstream infrastructure demand based on major customers' capital expenditure

  Summarizes the quarterly spending, full-year guidance, and future-year forecasts of the four major cloud service providers to assess the intensity and duration of the AI infrastructure construction cycle.

• Industry Chain AnalysisCross-Market Mapping Analysis

  Map changes in cloud investment to the AI optical communications supply chain

  Increased investment by cloud service providers in data centers and AI computing power typically raises demand for high-speed optical modules, network interconnects, and co-packaged optics.

• Forecast AnalysisForecast Revisions and Management Guidance

  Validate demand visibility by combining forecast upgrade magnitude and management commentary

  The report compares new and old capital expenditure forecasts and cross-validates industry trends using company statements on supply shortages, 2027 growth, and 2028 demand.

• Valuation AnalysisRelative Valuation Method

  Determine target prices based on expected P/E ratios and enterprise value multiples

  The target prices for Alphabet, Amazon, Meta, and Microsoft are mainly based on future-year EPS and P/E ratios, supplemented by enterprise value multiples or cost of equity discounting.

Asset mapping & comparison

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

• China AI optical communications supply chain

  Upstream beneficiary of capital expenditure growth by U.S. cloud service providers

  Strengths

  Shortages in AI computing power supply, network bandwidth upgrades, and improved demand visibility for 2027 to 2028 are favorable for volume growth in high-speed optical modules and optical interconnect products.

  Weaknesses

  Orders and revenue recognition may lag cloud capital expenditure, and sector valuations are also easily affected by short-term shipment pace.

  Comparison

  Compared with cloud service providers, the optical communications supply chain has higher earnings sensitivity to capital expenditure changes, but operating volatility is usually greater.

  Risks

  Cloud spending slowdown, price competition, product iteration falling short of expectations, and delayed commercialization of co-packaged optics.

• Alphabet (GOOGL)

  One of the four major U.S. cloud service providers and an important driver of AI infrastructure demand

  Strengths

  Google Cloud growth has reaccelerated, and TPU, Gemini, and search query demand support capital expenditure and valuation premiums.

  Weaknesses

  The scale of capital investment is expanding rapidly, and returns depend on commercialization of cloud services and generative AI.

  Comparison

  Second-quarter capital expenditure grew 100% year over year, the highest growth rate among the four companies.

  Risks

  Slower advertising demand, intensifying competition, antitrust, and data privacy regulatory pressure.

• Amazon (AMZN)

  Directly drives demand for AI servers, data centers, and network infrastructure through AWS expansion

  Strengths

  AWS has a leading market position, second-quarter capital expenditure was the largest in absolute terms, and demand continues to exceed supply.

  Weaknesses

  An accelerating investment cycle may delay margin expansion and pressure near-term cash flow.

  Comparison

  Second-quarter spending was US$53.0 billion and 2026 guidance was US$220.0 billion, both the highest among the four companies.

  Risks

  Weakening macro consumption, cloud computing price competition, insufficient investment returns, and regulatory risks in the United States and Europe.

• Meta Platforms (META)

  Investment in AI recommendation, advertising technology, and foundation models drives data center and network demand

  Strengths

  User scale, engagement levels, advertising innovation, and AI opportunities support growth, and the target price implies relatively high potential upside.

  Weaknesses

  Large-scale investment in AI and non-advertising businesses may raise costs and affect overall profitability.

  Comparison

  Second-quarter capital expenditure was US$31.0 billion, up 83% year over year, and the lower end of 2026 guidance was further raised.

  Risks

  Slower advertising spending, competition, insufficient returns from metaverse investment, and antitrust, data, and privacy regulation.

• Microsoft (MSFT)

  Azure and AI capability buildout are important sources of cloud capital expenditure and optical interconnect demand

  Strengths

  Solid enterprise customer base, leading position in Azure and AI, and reaccelerating growth.

  Weaknesses

  The recent rapid increase in capital expenditure will increase free cash flow volatility.

  Comparison

  Second-quarter capital expenditure was US$41.0 billion; the decline in the 2026 forecast is mainly caused by lease accounting reclassification, with no reduction in actual investment willingness.

  Risks

  AI adoption slower than expected, Azure growth deceleration, intensifying cloud market competition, and earnings dilution from large acquisitions.

Key data

• Second-quarter capital expenditure of the four major cloud service providersUS$170.0 billion, up 79% year over yearBroadly in line with expectations, reflecting still-strong momentum in AI infrastructure construction.
• 2026 capital expenditure forecastUS$754.0 billion, up 83.7% year over yearRaised by 4.5% from the previous forecast of US$721.0 billion.
• 2027 capital expenditure forecastUS$1.112 trillion, up 47.6% year over yearRaised by 15.1% from the previous forecast of US$966.0 billion.
• Google second-quarter capital expenditureUS$45.0 billion, up 100% year over year2026 guidance was raised from US$180.0 billion to US$190.0 billion to US$195.0 billion to US$205.0 billion.
• Meta second-quarter capital expenditureUS$31.0 billion, up 83% year over year2026 guidance was adjusted from US$125.0 billion to US$145.0 billion to US$130.0 billion to US$145.0 billion.
• Microsoft second-quarter capital expenditureUS$41.0 billion, up 69% year over yearThe decline in 2026 data to US$175.0 billion is mainly due to lease accounting reclassification rather than a decline in investment willingness; third-quarter guidance exceeds US$50.0 billion.
• Amazon second-quarter capital expenditureUS$53.0 billion, up 69% year over yearIt has the largest absolute spending scale, and 2026 guidance was raised from US$200.0 billion to US$220.0 billion.
• AI optical communications growth assessmentCompound growth rate of about 65% over the next three yearsDemand visibility extends to 2028, supporting the view of a structural industry upcycle.
• META target price and reference share priceTarget price US$800; reference share price US$556.71Corresponds to potential upside of about 43.7%, with valuation based on about 22x 2027E EPS.

Impact & implications

Continued upward revisions to capital expenditure mean that demand for AI data centers, network interconnects, and high-speed optical modules still has a long growth runway, and China's AI optical communications supply chain is expected to continue benefiting in 2027 to 2028. For cloud service providers, large-scale investment helps consolidate competitive positions in AI and cloud computing, but may increase near-term pressure on free cash flow, margins, and investment returns. The change in Microsoft's spending forecast is mainly an accounting classification adjustment and should not be interpreted as a weakening of actual investment willingness.

Risks

• AI application adoption slower than expected may weaken the economic returns of large-scale capital expenditure.
• Cloud service supply and demand shifting from shortage to surplus may lead to downward revisions to capital expenditure forecasts after 2027.
• Price competition in high-speed optical modules, customer concentration, and poor execution of product upgrades may limit supply chain profit growth.
• Technical validation or commercialization progress of co-packaged optics and near-packaged optics may fall short of expectations.
• Weakening macroeconomic conditions and advertising demand may affect Alphabet's and Meta's profitability and investment capacity.
• Intensifying cloud computing competition may pressure pricing and margins for AWS, Azure, and Google Cloud.
• Antitrust, data privacy, and other regulatory pressures may affect the operating performance of relevant platform companies.

What to watch

• Subsequent quarterly capital expenditure, 2027 budgets, and demand guidance from the four major cloud service providers.
• Whether AI computing power supply shortages can persist, and whether 2028 demand visibility is further confirmed.
• Whether AI optical module shipments can achieve significant sequential growth in 2H 2026.
• Changes in orders, revenue, and margins of optical communications companies after the release of mid-2026 results.
• Industry signals released by the China International Optoelectronic Exposition, Apsara Conference, and European Conference on Optical Communication in September.
• Customer validation, mass production timetables, and revenue contribution of co-packaged optics and near-packaged optics.
• Whether the 2027 AI optical communications industry growth outlook can maintain the expectation of a compound growth rate of about 65% over the next three years.