AI Boom Continues but Interest Rate Risks Loom; Recommend Increasing Allocation to Japanese Financial and Chemical Stocks

Overview

This research report focuses on the investment rotation between AI and non-AI sectors in the Japanese stock market. J.P. Morgan points out that although AI semiconductor stocks have recently dominated market gains and fundamentals remain strong, macro-level expectations for rising long-term interest rates in the US and Japan are changing market risk appetite. Therefore, the firm advises investors to prepare for investment in non-AI stocks while maintaining AI exposure, balancing portfolio risk through an 'AI + Financials' barbell strategy, and capturing structural opportunities in the chemical, construction, and successfully transformed IT service sectors.

Core views

Fundamentals in the AI sector remain solid, but caution is needed regarding the turning point in macro liquidity. The report argues that increased global investment in AI data centers and security will continue to support Japan's AI semiconductor sector. By breaking down stock performance, current AI semiconductor P/E ratios have fallen back to reasonable levels, with recent gains driven mainly by earnings (EPS) growth in sub-sectors like memory and cables, rather than valuation bubbles. However, red flags are appearing in macro funding conditions: non-financial institutions have extremely high equity allocations and low cash reserves, coupled with very low equity risk premiums. Once subjected to external shocks, this could easily trigger capital flows from stocks to bonds or lead to concentrated selling of AI stocks. Under expectations of rising interest rates, the allocation value of non-AI sectors becomes prominent. The J.P. Morgan team has raised its forecast for the US 10-year Treasury yield at the end of 2026 from 4.55% to 4.75%, and Japan's July budget request may trigger a rise in domestic long-term interest rates. In this context, the firm recommends adopting a 'barbell strategy' in the non-AI space: on one hand, overweight the financial sector to directly benefit from potentially accelerated rate hikes in the second half of the year; on the other hand, focus on the chemical sector, where profit margins are expected to expand significantly in April-June due to falling crude oil prices from highs and high naphtha prices. Additionally, developments in the Middle East (such as expectations of a US-Iran ceasefire) may provide support for the construction sector. Industry ratings are diverging, with IT companies previously feared to be 'disrupted by AI' seeing a turnaround. At the industry level, the firm has downgraded trading companies (due to insufficient optimization of capital policies and declining valuation attractiveness) and cosmetics (due to long-term low ROE and risk of capital outflows), while maintaining a neutral stance on the automotive sector. Notably, IT service and software companies, which were previously under pressure due to fears of being 'replaced by AI', are beginning to outperform the broader market. Taking Recruit as an example, it has successfully demonstrated to the market that AI matching technology can drive business transformation and expects a 20% growth in ARPJ (Average Revenue Per Job). The explosion in demand for Sovereign AI and Security AI is becoming a new long-term growth logic for these enterprises.

Analysis framework

The report adopts an analytical approach combining a top-down macro liquidity framework with a bottom-up fundamental breakdown. First, the firm precisely decomposes the stock performance of the AI sector into two dimensions: EPS (earnings) and P/E (valuation), thereby disproving the 'AI bubble theory' and confirming the earnings-driven nature of the current rally. Subsequently, it introduces a cross-asset allocation perspective by monitoring the position levels, cash ratios, and equity risk premium (ERP) of non-financial institutions to keenly capture potential liquidity reversal risks. In industry comparison, the report maps forward EPS growth rates against stock performance in two dimensions to identify sectors with 'expectation gaps' where earnings are improving but stock prices are lagging (such as chemicals). Combining this with interest rate sensitivity (such as the rate-hike benefit logic for the financial sector), it ultimately derives a balanced 'barbell strategy' offering both offensive and defensive capabilities.

Methodology notes

• Industry/Sector Analysis FrameworkUpstream-Midstream-Downstream Transmission in Industry Chain

  Diffusion Effect of AI Industry Chain Demand

  The report points out that AI investment is not isolated but transmits along the industry chain: starting from upstream segments like AI data centers and memory, gradually diffusing to electronic components such as MLCC (Multi-Layer Ceramic Capacitors), and eventually extending to mid-to-downstream application ends like Physical AI and Security AI. This transmission logic helps investors grasp the leading sub-sectors at different stages.

• Industry/Sector Analysis FrameworkVolume-Price Split

  Decomposition of Stock Price Drivers (Separating EPS and P/E)

  The firm splits the gains of tech stocks into two indicators: earnings expectations (EPS) and valuation multiples (P/E). Comparison reveals that the semiconductor sector is typically 'earnings-driven', while hardware sectors like MLCC are 'valuation-expansion-driven'. This decomposition helps judge the solidity of stock price increases and subsequent risks.

• Quantitative/Factor/Portfolio TheoryCapital Flow/Chip Analysis

  Cross-Asset Positioning and Liquidity Shock Risk

  The report assesses market crowdedness by observing the equity allocation ratio, cash holdings, and equity risk premium (ERP) of non-financial institutional investors. When equity positions are extremely high and cash is extremely low, the market's buffer capacity against sudden shocks is very weak, easily triggering stampede-style selling. This serves as an important contrarian indicator for judging top-of-market risks.

Asset mapping & comparison

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

• Recruit Holdings

  Beneficiary of Turnaround for Those Disrupted by AI

  Strengths

  Successfully demonstrated to the market that its business can coexist with AI, driving business transformation using AI matching technology, with expected ARPJ growth of 20%.

  Comparison

  Stock price has risen 1.5 times since April, significantly outperforming the TOPIX index over the same period.

• Murata Manufacturing

  Core Supplier of AI Hardware, Beneficiary of Valuation Expansion

  Strengths

  Holds a majority market share in the MLCC field for advanced AI servers.

  Weaknesses

  Current P/E ratio is higher than that of the semiconductor and equipment sectors, indicating signs of overheating in valuation.

  Comparison

  Stock price has risen 2.1 times since April, outperforming the average level of the technology sector.

  Risks

  If expectations for margin expansion from price hikes fail to translate into actual EPS upgrades, it may trigger a valuation correction.

• Sony Group, Fujitsu, NEC

  IT/Electronics Giants Successfully Coping with AI Impact

  Strengths

  Alleviated excessive market concerns about AI disruption by releasing new medium-term plans, demonstrating business resilience.

  Comparison

  After experiencing a 10% decline, stock prices have all achieved a rebound of approximately 20%.

Key data

• Contribution Rate of AI Semiconductor Stocks to Japanese Stock Market Gains60%Cumulative contribution since April 2026
• Year-End Forecast for US 10-Year Treasury Yield4.75%Raised from previous 4.55%
• 12-Month Forward EPS Forecast for Non-AI Sector+20%Upward revision magnitude since early 2026
• Stock Price Increase of MLCC Manufacturers Since April2.1x and 3.6xFor Murata and Taiyo Yuden respectively
• Recruit Expected ARPJ Growth Rate20%Benefiting from transformation to AI matching business

Impact & implications

The report hints that market style is shifting from a single bet on AI towards diversified defense. For investors, this means that while core AI assets (such as semiconductors and equipment) still hold value, portfolios must increase exposure to assets sensitive to macro interest rates (such as financial stocks) as a hedge. Meanwhile, the spillover effect of market capital is seeking new valleys; traditional IT service enterprises that can prove their business models can coexist with or even benefit from AI are experiencing a significant window for valuation repair.

Risks

• Non-financial institutional investors have excessively high equity positions and lack cash; macro shocks could easily trigger concentrated selling of core assets such as AI semiconductors.
• Equity Risk Premium (ERP) is at a low level; unexpected rises in long-term interest rates in the US and Japan could trigger a systemic transfer of funds from stocks to bonds.
• High-valuation hardware sectors (such as MLCC) face the risk of valuation compression if profit margin expansion expectations are not realized in financial reports.
• Some domestic demand sectors like cosmetics have long-term low ROE; amidst continuous capital flow towards AI, there is a risk of further downward adjustment in P/E and P/B ratios.

What to watch

• Whether the Japanese government's budget request released in July includes clear investment plans for Security AI and Physical AI.
• Profit margin expansion of electronic component manufacturers such as MLCC during the development of new generation products, and whether this is truly reflected in EPS expectations.
• The catalytic effect of evolving geopolitical situations in the Middle East (such as progress in US-Iran ceasefire agreements) on Japan's construction and energy resource sectors.
• The actual impact of US inflation data and the Federal Reserve's policy path on the 10-year Treasury yield.