Tokyo investor questions on yen, rates, geopolitics, and AI Report Interpretation

Overview

This first report in a series on issues raised by Japanese investors and corporate leaders presents Morgan Stanley’s house views on the yen, US rates, US-China relations, AI investment, and AI’s broader economic effects. Its central message is that near-term market dynamics remain driven by rate differentials, oil-linked Fed expectations, and policy deadlines, while AI offers visible productivity and return potential alongside growing medium-term financial and social risks.

Core views

On Japan and foreign exchange, Morgan Stanley argues that BoJ rate-hike expectations alone are unlikely to produce sustained yen appreciation. Its Japan Macro Strategy team puts USD/JPY fair value at around 167, even after the July 30 coordinated US-Japan FX intervention, and concludes that two to three BoJ hikes would not materially change the exchange rate. Recent market moves have inverted the usual relationship between the yen and JGB yields: the yen weakened even as JGB yields rose. The report attributes this to concern that the BoJ could remain behind the curve, causing rate markets to take cues from FX developments rather than FX responding to tightening expectations. The team expects BoJ policy rates to rise to 1.25% in September 2026 and 1.5% in January 2027, but says durable yen appreciation requires both lower US rates and a meaningfully more aggressive BoJ tightening cycle. Under lower US rates, USD/JPY could trend toward the low-to-mid 150s over the medium term. Conversely, a reacceleration in US inflation and a Fed turn toward tightening would widen US-Japan rate differentials and intensify depreciation pressure on the yen. For US rates and credit, the report rejects the view that rising investment-grade corporate issuance is the main cause of higher Treasury yields. Morgan Stanley’s US Rates Strategy team instead attributes the move principally to higher oil prices after the Middle East conflict and the market’s more hawkish reading of the Fed reaction function, transmitted through real yields rather than breakeven inflation. Since the conflict began, oil prices, Fed pricing, and 10-year Treasury yields have moved together; the operative market interpretation is that higher oil implies a more hawkish FOMC. Although excess investment-grade supply has widened credit spreads, dealer-inventory increases were too small to explain Treasury weakness, and corporate issuance has little explanatory power for the move in Treasury yields. AI-related corporate issuance was running at a record pace through July, up 29% year on year, but the report views this as a source of wider credit spreads rather than higher Treasury yields. Morgan Stanley maintains a 4.25% year-end target for the 10-year Treasury yield. A renewed Middle East escalation that lifts oil prices could raise real yields again, while an entrenched perception of an oil-linked Fed reaction function could sustain front-end volatility and credit-spread widening. On US-China relations, the report frames the relationship as “stabilization, not normalization.” It expects the September 24 summit to confirm managed stability rather than produce a grand bargain, because the structural disputes over semiconductors, rare earths, and export controls remain unresolved. The key event is November 10, when three policy measures are due to expire together: the Section 301 exclusion list, the suspension of the BIS Affiliates Rule, and the pause on China’s Wave 2 rare-earth export controls. The report notes that China’s April 2025 controls on seven heavy rare earths were never suspended and that US imports of yttrium, used in computer chips, are down 70% from January 2025 levels. Confirmation of Xi Jinping’s visit and softer signals on technology and trade restrictions could improve risk appetite for Chinese equities, but Morgan Stanley sees this as more likely to support a relief rally than a fundamental rerating. If the summit disappoints and the measures lapse without renewal, the report warns that the rare-earth and semiconductor equilibrium could break down and increase AI supply-chain costs. It also expects China’s AI-chip self-sufficiency ratio to reach 70% by 2030, reducing US export-control leverage over time. Morgan Stanley does not consider the AI capex cycle to be in overinvestment territory yet. Second-quarter 2026 earnings reinforced confidence in returns on investment and led to further upward capex revisions. Still, hyperscaler cash capex-to-revenue ratios, including finance leases, are expected to reach 36% in 2026, 44% in 2027, and 42% in 2028, materially above the dot-com peak of about 32%. The combined capex of the top five cloud players is forecast at $779bn in 2026, $1.23trn in 2027, and $1.396trn in 2028; the 2027 forecast was revised up another 20% versus July after second-quarter results. Morgan Stanley’s Internet analysts forecast 70% incremental margins and 30% ROIC for GPU leasing in 2026-27, supporting the view that return evidence is accumulating. However, the report stresses that capex deployment precedes monetization, leaving four hyperscalers’ free-cash-flow forecasts under downward revision through 2027 and widening their financing gap. AI-related credit issuance therefore needs to remain substantial, potentially increasing further, before investment cash flows catch up. The primary downside risk is disappointing AI returns if capability improvements or enterprise adoption lag; a sharp spending cut could hurt semiconductor and data-center equities. Power and grid constraints, permitting delays, supply-chain bottlenecks, and a wider capex-to-monetization lag could also slow deployment and cause AI credit spreads to widen sharply. On AI’s social and macroeconomic impact, the report finds evidence that AI is lifting output-led productivity, but argues that the distribution of gains will depend on diffusion speed and policy response. Industries with high AI exposure recorded faster output per worker growth than other industries in 2025, with high-exposure industries contributing 1.7 percentage points of the 2.4 percentage-point growth differential. The report attributes the gain mainly to faster output growth rather than labor displacement, which is broadly consistent with an optimistic outcome. Yet August 2026 data show labor disruption becoming increasingly visible: high-AI-exposed occupations are contributing up to 15 basis points to aggregate unemployment. In a baseline scenario where AI diffuses at roughly twice the speed of the internet era, the report expects faster output and productivity growth without a significant rise in unemployment or a recession. Across diffusion and feedback scenarios, however, peak unemployment-rate increases range from 0.2 percentage points to 4.1 percentage points. Rapid diffusion can remain manageable if task creation and wealth effects are sufficiently strong, but delayed fiscal and monetary responses could create a recession-type rise in unemployment. The report concludes that fiscal policy, automatic stabilizers, and retraining investment will be critical to whether AI produces broad prosperity or greater inequality, political backlash, and stronger redistribution such as wealth taxes or AI levies.

Analysis framework

The report answers five investor questions using Morgan Stanley’s cross-asset house views. It connects FX to US-Japan rate differentials and BoJ policy expectations; Treasury yields to oil, Fed pricing, and real yields; geopolitics to specific trade and export-control deadlines; AI investment to capex, revenue, ROIC, free cash flow, and credit financing; and AI’s social effects to productivity data and diffusion-policy scenarios.

Methodology notes

• MacroeconomicsInterest rate parity

  US-Japan rate differentials as a driver of USD/JPY

  The report argues that yen appreciation requires narrower US-Japan rate differentials, not simply additional BoJ hikes.

• Fixed Income and CreditThree-Factor Interest-Rate Decomposition

  Separating Treasury-yield moves into real yields and inflation expectations, while distinguishing Treasury rates from credit spreads

  Morgan Stanley attributes the rise in Treasury yields chiefly to real-rate and Fed-repricing dynamics, while treating heavy investment-grade issuance as a driver of wider credit spreads.

• Corporate Fundamentals and FinanceROIC–WACC spread

  Assessing AI capex through expected ROIC, incremental margins, free cash flow, and financing needs

  The report uses improving projected returns to support current capex, but highlights the cash-flow lag and financing gap as medium-term constraints.

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  Rare-earth controls and semiconductor restrictions affecting AI supply-chain costs

  The report links policy renewal or expiry to the availability and cost of inputs relevant to chips and the AI supply chain.

• Other

  Scenario analysis of AI diffusion speed, task creation, wealth effects, and policy response

  The report models a range of unemployment outcomes under differing diffusion and feedback assumptions to assess AI’s labor-market effects.

Asset mapping & comparison

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

• JPY / USDJPY

  The report links sustained yen appreciation to lower US rates and a narrower US-Japan rate differential.

  Strengths

  Could trend toward the low-to-mid 150s over the medium term if US rates decline.

  Weaknesses

  Two to three BoJ hikes alone are not expected to materially strengthen the yen.

  Comparison

  USD/JPY fair value is estimated around 167.

  Risks

  A US inflation reacceleration and renewed Fed tightening would widen rate differentials and increase yen-depreciation pressure.

• US Treasuries

  Higher yields are attributed primarily to oil-driven Fed repricing and real yields rather than investment-grade supply.

  Weaknesses

  Oil-linked hawkish Fed expectations can push real yields higher.

  Comparison

  Morgan Stanley maintains a 4.25% year-end target for the 10-year Treasury yield.

  Risks

  A renewed Middle East escalation could raise oil prices and Treasury yields further.

• AI-related investment-grade credit

  Heavy AI financing needs are linked to wider credit spreads and a continuing funding requirement.

  Strengths

  Improving ROI evidence supports ongoing investment.

  Weaknesses

  Free-cash-flow forecasts for four hyperscalers continue to be revised lower through 2027.

  Comparison

  AI-related issuance was up 29% year on year through July 2026.

  Risks

  A wider capex-to-monetization lag could cause AI credit spreads to widen sharply.

• Semiconductor and data-center equities

  They are exposed to hyperscaler AI spending and to potential rare-earth and export-control disruption.

  Strengths

  AI investment forecasts and return expectations remain strong in the near term.

  Weaknesses

  Their outlook depends on continued hyperscaler capex and successful monetization.

  Risks

  Disappointing AI capabilities or adoption could trigger spending cuts and a sharp selloff; supply-chain constraints could slow deployment.

• Chinese equities

  They could respond to signs of softer US-China technology and trade restrictions.

  Strengths

  A constructive September summit signal could lift risk appetite.

  Weaknesses

  The report does not see a summit outcome as resolving structural disputes.

  Comparison

  The report sees scope for a relief rally rather than a fundamental rerating.

  Risks

  Failure to renew key November 10 policy measures could disrupt rare-earth and semiconductor conditions.

Key data

• USD/JPY fair valueAround 167Morgan Stanley’s Japan Macro Strategy estimate.
• Expected BoJ policy rate1.25% in September 2026; 1.5% in January 2027The report’s expected tightening path.
• US 10-year Treasury year-end target4.25%Maintained by Morgan Stanley’s US Rates Strategy team.
• AI-related corporate bond issuance+29% YoY through July 2026Record pace; viewed as a credit-spread driver rather than a Treasury-yield driver.
• US imports of yttriumDown 70% from January 2025 levelsEvidence of the effect of China’s rare-earth controls.
• China AI-chip self-sufficiency70% by 2030Projected level cited as reducing US export-control leverage over time.
• Hyperscaler cash capex-to-revenue36% in 2026; 44% in 2027; 42% in 2028Including finance leases; above the dot-com peak of about 32%.
• Top-five cloud-player capex$779bn in 2026; $1.23trn in 2027; $1.396trn in 2028The 2027 estimate was revised 20% higher versus July.
• GPU leasing economics70% incremental margins; 30% ROICMorgan Stanley Internet analyst forecasts for 2026-27.
• AI-exposed industry productivity contribution1.7pp of a 2.4pp output-per-worker growth differential in 2025The report attributes the differential mainly to faster output growth.
• Potential peak unemployment-rate increase0.2pp to 4.1ppRange across AI diffusion and policy-feedback scenarios.

Impact & implications

Morgan Stanley’s views imply that the most important near-term cross-asset catalysts are US rate movements, oil-price developments, and the November 10 US-China policy deadline. For AI, current spending is supported by return evidence, but the cash-flow and credit-financing burden remains material until monetization catches up. The broader economic outcome depends on whether productivity gains diffuse widely enough and whether policy limits labor displacement and inequality.

Risks

• US inflation reacceleration and a Fed tightening pivot could widen US-Japan rate differentials and intensify yen depreciation.
• Renewed Middle East tensions could lift oil prices, push real Treasury yields higher, and prolong credit-spread widening.
• If the November 10 US-China policy measures expire without renewal, rare-earth and semiconductor conditions could deteriorate and raise AI supply-chain costs.
• AI model progress or enterprise adoption could disappoint, prompting hyperscalers to cut spending and pressuring semiconductor and data-center equities.
• Power and grid constraints, permitting delays, and supply-chain bottlenecks could delay AI deployment and widen AI credit spreads.
• Rapid AI diffusion without adequate task creation, wealth effects, or policy support could cause a recession-type rise in unemployment and greater inequality.

What to watch

• BoJ policy decisions expected in September 2026 and January 2027, alongside US-rate developments and the US-Japan rate differential.
• Oil prices, Fed policy pricing, real yields, and the persistence of the market’s oil-linked Fed interpretation.
• The September 24 US-China summit and the November 10 expiry or renewal of Section 301 exclusions, the BIS Affiliates Rule suspension, and rare-earth-control measures.
• Hyperscaler capex revisions, AI return evidence, free-cash-flow forecasts, and AI-related credit issuance.
• AI adoption, productivity data in highly exposed industries, labor-market disruption, and fiscal, monetary, and retraining responses.