Goldman Sachs Cross-Asset Commentary: Focus on Useful Output per Watt, CPI, Earnings Season, and Reverse Dispersion Opportunities

Overview

This is a cross-asset market commentary written by personnel from Goldman Sachs FICC and Equities, rather than a formal Global Investment Research report. It covers Asian momentum stocks, energy geopolitical risks, AI competition and inference economics, hyperscaler capital expenditure, U.S. CPI, Q2 earnings season, rates, and options hedging. The core title, “Useful Output per Watt,” points to a shift in AI investment from simply expanding capital expenditure toward measuring efficiency by cost per unit, energy consumption per unit, and actual commercial output.

Core views

The core views are: first, geopolitical conflict in the Gulf will continue to disrupt oil prices, but policy constraints make the base case closer to a tense compromise than sustained loss of control, with Brent potentially remaining in the $75–85 per barrel range; second, intensifying AI model competition and migration toward open-source and small models will depress API and token prices, without necessarily generating sufficient demand elasticity, while valuations of hardware beneficiaries may also be affected by hyperscalers’ capital discipline; third, if U.S. CPI does not materially exceed expectations, it is more likely to support risk assets, although fiscal issuance and global duration supply continue to steepen the curve; fourth, Q2 earnings growth is strong but already reflected in high expectations, shifting market focus to second-half guidance; fifth, when implied correlation is extremely low, index options and related reverse dispersion trades offer attractive hedging potential.

Analysis framework

The report uses a top-down cross-asset macro and market-positioning framework, combining geopolitics, oil-product supply and demand, CPI event risk, revisions to corporate earnings expectations, changes in AI industry business models, and options-implied correlation to assess short-term catalysts for risk assets and longer-term capital-efficiency themes.

Methodology notes

• Cross-asset strategyMacro event-risk analysis

  CPI, oil prices, and earnings guidance jointly determine short-term risk-asset pricing

  The author treats U.S. CPI, energy geopolitical risks, and Q2 earnings season as the key short-term events and assesses their impact on rates, the dollar, equities, and demand for options hedges.

• Industry analysisAI unit economics framework

  Useful output per watt/useful output per dollar

  The author emphasizes that AI investment should not be assessed solely by capital expenditure or compute scale, but by how much genuinely usable output can be generated per unit of energy and investment.

• Derivatives strategyImplied correlation and dispersion

  Reverse dispersion and index-option hedging

  When implied correlation is near historical lows, options at the index level offer relatively greater appeal and can hedge systemic repricing risks triggered by oil prices, CPI, or earnings guidance.

Asset mapping & comparison

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

• AI hardware and semiconductors

  Driven by AI capital expenditure and inference demand, but valuations depend on hyperscaler capital discipline

  Strengths

  If lower token prices unlock demand, hardware could still benefit from growth in inference volumes.

  Weaknesses

  Migration of enterprise workloads to small models, local models, and CPUs/edge devices could weaken the demand elasticity for centralized high-end compute.

  Comparison

  Compared with the earlier simple trade of buying hardware beneficiaries, hardware stocks are now more highly correlated with AI spenders.

  Risks

  Token-price deflation, slower capital expenditure, and shareholder demands for greater capital efficiency.

• Hyperscalers and neo-clouds

  At the center of pressure from AI inference-price deflation and the shift toward decentralized inference

  Strengths

  They still possess frontier models, cloud infrastructure, and an enterprise customer base.

  Weaknesses

  Huge capital expenditure must demonstrate returns; if inference becomes decentralized, the neo-cloud model faces structural challenges.

  Comparison

  Frontier models are more like high-end inference engines, while everyday production inference may shift toward specialized, localized models.

  Risks

  Falling API prices, insufficient demand elasticity, and capital-discipline pressure.

• Energy and oil & gas

  Affected by Gulf geopolitical conflict, passage through Hormuz, and refined-product tightness

  Strengths

  Geopolitical risk premiums and tight refined-product markets support energy prices.

  Weaknesses

  U.S. policy objectives favor avoiding persistently triple-digit oil prices.

  Comparison

  The author believes refined-product tightness may matter more for rates than the Brent spot price itself.

  Risks

  Attacks on regional energy infrastructure and simultaneous disruption at Hormuz and Bab el-Mandeb.

• U.S. equities

  Influenced jointly by CPI, rates, and Q2 earnings guidance

  Strengths

  Q2 EPS growth expectations are strong, with upward earnings revisions.

  Weaknesses

  High expectations may already be reflected in prices, shifting market focus to second-half guidance.

  Comparison

  The financial sector has already risen significantly ahead of earnings, so the marginal impact of an earnings beat may be limited.

  Risks

  An upside CPI surprise, conservative guidance, and continued increases in long-end rates.

• Rates and the dollar

  CPI event risk affects front-end rates and the dollar

  Strengths

  If CPI does not produce a significant upside surprise, front-end receiving positions and a weaker dollar could benefit.

  Weaknesses

  Global duration supply and fiscal issuance remain structural headwinds.

  Comparison

  Forward guidance can flatten the curve, but fiscal supply drives it steeper.

  Risks

  A significant CPI upside surprise and a continued drift higher in long-end yields.

• Index options/reverse dispersion

  Low implied correlation makes index optionality more valuable for hedging

  Strengths

  Can hedge systemic repricing risks triggered by oil prices, CPI, and earnings guidance.

  Weaknesses

  If low volatility persists and event risks do not materialize, hedging costs may weigh on returns.

  Comparison

  The author considers reverse dispersion to appear attractive.

  Risks

  Continued low correlation and unrealized volatility.

Key data

• Brent base-case range$75–85 per barrelThe author's base case assumes Iran exercises de facto control over transportation, the United States accepts the operational reality, and traffic gradually resumes.
• Brent upside scenarioAbove $100 per barrelThis would require more direct attacks on regional energy infrastructure or simultaneous disruption at Hormuz and Bab el-Mandeb.
• S&P 500 Q2 EPS growth consensusApproximately 23–24% year over yearThe author describes this as one of the strongest quarters in several years.
• In-quarter earnings expectation revisionsUp approximately 3%This is contrary to the usual seasonal pattern of downward revisions.
• Corporate guidanceMore than 50% of companies guided above consensusSignificantly above the historical average.
• TSMC Q2 salesUp 36% year over yearBroadly in line with, to slightly above, market consensus.
• Performance of South Korean memory stocksDown 5–6%Affected by the weekend's escalation in Gulf geopolitics and weakness in the Asian AI chain.
• Performance of South Korean electrification-related stocksDown approximately 7%Reflecting pressure on high-momentum and AI-related sectors in Asia.

Impact & implications

The portfolio implication is that the AI trade may shift from “buy hardware and beneficiaries of AI spending” toward stricter assessment of capital efficiency, output per unit of energy, and beneficiaries with genuine cash flow. If hyperscalers come under pressure from shareholders demanding capital discipline, hardware beneficiaries may face valuation pressure even if earnings estimates are revised upward. At the macro level, if CPI does not materially rise, this would be favorable in the short term for risk assets and front-end receiving positions, while long-end rates remain constrained by fiscal supply. In energy, an uncontrolled oil-price surge is not the base case, but tight refined-product markets are more important for inflation and rates.

Risks

• Escalation of Gulf geopolitical conflict spreading to regional energy infrastructure.
• Simultaneous transportation disruption at Hormuz and Bab el-Mandeb.
• A significant upside surprise in U.S. CPI, driving rate-hike expectations and long-end yields higher.
• Strong Q2 earnings but conservative second-half guidance, leading to market repricing.
• Falling AI token prices without sufficient demand elasticity, weakening the AI hardware and cloud capital-expenditure narrative.
• Hyperscaler shareholders demanding capital discipline, suppressing valuation multiples across the AI value chain.
• Continued global fiscal issuance and duration supply driving the curve steeper.

What to watch

• Whether U.S. CPI is significantly above expectations.
• Whether Brent remains at $75–85 per barrel or breaks higher toward the above-$100-per-barrel risk scenario.
• Transportation conditions at Hormuz and Bab el-Mandeb and the security of regional energy infrastructure.
• Second-half guidance from financial companies and S&P 500 companies during Q2 earnings season.
• The impact of Meta, xAI, open-source models, and small-model deployment on API and token prices.
• Hyperscaler capital-expenditure guidance, shareholder returns, and statements on capital discipline.
• Implied correlation, index-option pricing, and reverse dispersion opportunities.