AI infrastructure drives a major cycle in Japanese technology; cloud and data centers remain preferred in China internet

Overview

This Goldman Sachs The 720 report covers multiple themes including Japanese technology components, China internet, JX Advanced Metals, Panasonic, China basic materials, China financials, Bank of Ningbo, Fisher & Paykel, India CRO/CDMO, Micron, and global energy Top Projects. The report's main theme is that the diffusion of AI infrastructure and AI applications is reshaping the cycles of hardware, cloud, materials, semiconductors, and energy capital expenditure, while also emphasizing fundamental divergence and valuation selection across China internet, financials, and commodities.

Core views

The core views include: first, Japanese technology components such as MLCCs, substrates, and Power Semis are entering a new AI-driven upcycle, but growth divergence across companies will widen due to differences in AI exposure and material costs; second, within China internet, cloud and data centers are the top pick for 2H26 due to cloud price increases and token growth, gaming and entertainment rise to second preference, while e-commerce and mobility fall to third; third, AI server materials strengthen the investment case for JX Advanced Metals and Panasonic; fourth, coal and copper are preferred in China basic materials, while brokers are preferred over banks in China financials; fifth, tight DRAM supply-demand improves Micron's earnings visibility, but the rating remains Neutral.

Analysis framework

The report uses a combination of thematic rotation and bottom-up company research: it first identifies cross-sector drivers such as AI, cloud, materials, financial capital returns, and energy capital expenditure, then maps them to company earnings forecasts, ROE/CROCI, valuation multiples, target prices, and rating changes. For the internet section, the report is structured around five investor debates on AI model competition, price competition, token growth, hyperscaler capex/free cash flow, and agentic AI form factors; for India CRO/CDMO, it uses the 4Cs framework to track orders, capacity, AI, and margins.

Methodology notes

• factor frameworkGS Factor Profile

  Compares stocks with the market and peers across four attributes: Growth, Financial Returns, Multiple, and Integrated.

  Growth is based on forward sales, EBITDA, and EPS growth; Financial Returns is based on ROE, ROCE, and CROCI; Multiple is based on valuation metrics such as P/E, P/B, P/D, EV/EBITDA, and EV/FCF; Integrated is a composite percentile of Growth, Financial Returns, and value factors.

• M&A probability frameworkM&A Rank

  Uses a 1 to 3 ranking to assess the probability that a covered company becomes an acquisition target.

  Rank 1 represents high probability, Rank 2 represents medium probability, and Rank 3 represents low probability; when the Rank is 1 or 2, Goldman Sachs may incorporate M&A factors into the target price.

• industry research framework4Cs

  Four focus areas used in India CRO/CDMO research.

  The report plans to assess industry trends around CDMO inventories, CRO headcount and AI drivers, slower operating profit growth from newly commercialized facilities, and the timeline for capex conversion into revenue.

Asset mapping & comparison

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

• Japanese technology components (MLCCs, substrates, Power Semis)

  Directly benefits from AI infrastructure build-out and later from the diffusion of edge AI and physical AI devices.

  Strengths

  Earnings, CROCI, and ROE are expected to exceed historical highs, and target prices for multiple AI beneficiaries have been raised significantly.

  Weaknesses

  Companies differ in AI exposure and material cost impact, and growth divergence within the sector is expected to widen.

  Comparison

  The report believes this cycle could be one of the largest and longest in the industry's history.

  Risks

  A slowdown in AI capex, rising material costs, customer inventory adjustments, or product price declines could weaken the earnings upgrade thesis.

• China internet and AI model ecosystem

  Affected by AI model competition, token growth, cloud service pricing, and the evolution of agentic AI form factors.

  Strengths

  Cloud and data centers become the top choice due to price increases and a token boom, while Tencent and NetEase look attractive on trough-cycle valuations.

  Weaknesses

  Price competition among Chinese model players is intensifying, and hyperscaler capex and free cash flow remain under scrutiny.

  Comparison

  Cloud and data centers are the first preference, gaming and entertainment the second, and e-commerce and mobility the third.

  Risks

  AI price wars, agent-to-agent disruption, macro weakness, regulatory changes, and lower-than-expected returns on capital expenditure.

• JX Advanced Metals

  Benefits through AI server materials such as semiconductor targets, titanium copper, high-purity tantalum powder, and InP substrates.

  Strengths

  AI-related volume and price growth drive earnings forecast upgrades, while higher copper price assumptions also benefit the resource business.

  Weaknesses

  Earnings are relatively sensitive to demand for AI server materials and copper price assumptions.

  Comparison

  Reinitiated with Buy and a 12-month target price of ¥4,900.

  Risks

  Weaker-than-expected AI server demand, falling copper prices, delayed capacity ramp-up, or slower material qualification progress.

• Panasonic

  High-performance circuit board materials, conductive capacitors, and AI data center capacity backup units are tied to generative AI infrastructure.

  Strengths

  Management plans to more than double AI-related supply capacity from FY3/26 to FY3/31, while valuation is below the sector average.

  Weaknesses

  The investment thesis depends on ramp-up of new AI products, portfolio review, and execution of layoffs and cost reduction.

  Comparison

  Maintained at Buy with a 12-month target price of ¥4,220.

  Risks

  Uncertainty around capacity expansion execution, AI data center demand, product competitiveness, and restructuring progress.

• China basic materials

  Coal and copper are supported by supply-demand fundamentals, while earnings forecasts for steel, cement, gold, and paper have been cut.

  Strengths

  Coal chemical demand and slower output recovery after copper mine disruptions support the preference for coal and copper.

  Weaknesses

  Price changes in some commodities year-to-date have led to downward earnings revisions.

  Comparison

  Yankuang-A is downgraded to Sell, while Zijin, JXC, Chinacoal, Muyuan, and YTH remain Buy-rated.

  Risks

  Spot price declines, faster-than-expected supply recovery, weak demand, or policy changes.

• China financials and Bank of Ningbo

  Brokers benefit from capital replenishment, high-return opportunities in Hong Kong markets, and AI company IPO underwriting; for banks, balance sheet quality matters more.

  Strengths

  Broker ROE guidance may be revised up, while Bank of Ningbo has high-quality growth, relatively high ROE, and solid capital levels.

  Weaknesses

  Bank loan growth is slowing, and valuation assessment is shifting from EPS growth toward balance sheet strengthening.

  Comparison

  The report prefers brokers over lenders, but Bank of Ningbo still maintains a Buy rating.

  Risks

  Net interest margin pressure, deteriorating asset quality, insufficient AUM growth, regulation of non-compliant accounts, and capital market volatility.

• Micron and the DRAM cycle

  Strong conventional DRAM pricing and tight supply-demand drive higher revenue, margin, and EPS forecasts.

  Strengths

  The report expects 3Q results to come in significantly above market expectations and company guidance, and tight supply-demand may persist into CY27.

  Weaknesses

  Despite the sharp increase in target price, the rating remains Neutral, indicating valuation or cycle risks still warrant attention.

  Comparison

  The 12-month target price is raised from US$400 to US$900.

  Risks

  A reversal in the DRAM price cycle, faster-than-expected supply release, slowing end demand, and changes in capital expenditure.

• Global energy Top Projects

  AI and digitalization help rebuild reserve life in deepwater exploration and global shale development.

  Strengths

  Oil and gas capex is expected to return to double-digit growth in 2027, with Big Oils maintaining leadership in high-return new projects.

  Weaknesses

  Project returns depend on FID timing, oil prices, and execution efficiency.

  Comparison

  TotalEnergies, ExxonMobil, and ConocoPhillips are listed as Top Projects beneficiaries.

  Risks

  Falling oil and gas prices, project delays, cost overruns, regulatory pressure, and energy transition constraints.

Key data

• Report date2026-06-09Published by Goldman Sachs Equity Research.
• Revision to Japan technology FY26-FY28 operating profit forecasts+3% to +17%The upward revision mainly reflects expectations for AI infrastructure build-out and subsequent diffusion of edge AI/physical AI.
• China AI token forecast350 trillion tokens per day by year-endSupports cloud and data centers as the top pick within China internet for 2H26.
• JX Advanced Metals target price¥4,900Reinitiated with Buy, with FY3/27 and FY3/28 operating profit forecasts raised by 16% and 13%, respectively.
• Panasonic target price¥4,220Maintained at Buy, with generative AI-related products expected to contribute more than 30% of adjusted operating profit by FY3/28.
• Copper price assumptionUS$13,400 to US$13,800/ton in 2026-2028Used to raise earnings forecasts for JX Advanced Metals' resource business.
• Bank of Ningbo target priceRmb41.29Maintained at Buy due to resilient high-quality growth, ROE, and capital levels.
• Fisher & Paykel target priceA$42.30Maintained at Buy and on the Conviction List, with focus on improving consumables utilization.
• Micron target priceUS$900 (raised from US$400)Maintained at Neutral, but CY26/CY27 revenue and non-GAAP EPS forecasts were significantly raised due to conventional DRAM pricing and tight supply-demand.
• Oil and gas capexExpected to grow 11% in 2027Driven by accelerating deepwater FIDs, AI- and digitalization-enabled exploration, and global shale development.

Impact & implications

The portfolio implication is that the AI theme has expanded from investments in single models or applications to multiple industry chains including hardware components, AI server materials, cloud and data centers, DRAM, power, and energy capital expenditure. The report does not recommend indiscriminately chasing AI concepts, but instead emphasizes selection based on earnings forecast upgrades, improvements in ROE/CROCI, valuation positioning, material cost pressure, and cash flow sustainability. Within China assets, internet giants may see support from a bottoming EPS downgrade cycle and valuation multiples in 2H26, but model price competition, capex/FCF pressure, and macro demand remain important constraints.

Risks

• If AI infrastructure capital expenditure slows, it could weaken the earnings upgrade thesis for Japanese technology components, cloud, data centers, AI server materials, and DRAM.
• Intensifying price competition in Chinese AI models could compress profitability for cloud and model companies and affect hyperscaler free cash flow.
• Fluctuations in material costs, copper prices, coal prices, and other commodity prices could alter profit elasticity for basic materials and hardware companies.
• China internet and financial sectors still face uncertainty around macro demand, regulation, the pace of valuation recovery, and asset quality.
• Micron and the DRAM industry face risks from supply release, price declines, and cycle reversal.
• AI materials companies such as Panasonic and JX Advanced Metals face risks in capacity expansion, product qualification, and demand realization.
• The Fisher & Paykel view depends on assumptions of US tariff exclusion and improved consumables utilization; if these do not hold, earnings expectations will be affected.
• Energy Top Projects face risks from FID delays, lower oil and gas prices, cost overruns, and policy constraints.

What to watch

• Whether China's daily AI tokens can progress toward the 350 trillion target, and whether cloud service price increases can continue.
• Whether Alibaba's EPS downgrade cycle has bottomed, and whether Tencent's valuation multiple can find support in 2H26.
• Realization of AI orders, material costs, and FY26-FY28 profit upgrades for Japanese technology component companies.
• Changes in JX Advanced Metals' AI server material volumes, pricing, and copper price assumptions.
• Panasonic's AI-related capacity expansion, next-generation AI data center products, and progress in portfolio adjustment.
• Price changes and earnings forecast revisions for coal, copper, steel, cement, gold, and paper in China.
• Incremental gains for Chinese brokers in Hong Kong market ROE, AI company IPO underwriting, and co-investment income.
• Changes in Bank of Ningbo's net interest margin, asset quality risk resolution, NPL coverage ratio, and payout ratio.
• Micron's 3Q results, DRAM price trends, and the degree of supply-demand tightness in CY27.
• Whether deepwater FIDs, global shale investment, Big Oils project returns, and 2027 oil and gas capex growth can materialize.