Goldman Sachs: The Markets
2026-05-18 · Hosted by — · Goldman Sachs
Executive Summary
Goldman Sachs trader Sean Tatasia joined host Chris Hussey to discuss what he calls a “tale of two markets” — AI-related US equities (hyperscalers, semiconductors, data centers) powering indices to record highs while the “real economy” (consumer, cyclicals, homebuilders) quietly suffers from the oil-driven stagflationary shock. Key metrics: S&P EPS growth is tracking +17% year-over-year for Q1 (sixth consecutive quarter of double-digit growth), forward 12-month EPS growth estimates have been revised up to +13%, and the S&P multiple has actually compressed from 22x to 21x year-to-date — meaning earnings are doing the “heavy lifting.” The next three focuses are the 30-year yield at 5%+ (a psychological threshold), Nvidia earnings next week, and US-China meeting outcomes.
Key Stories & Changes
1. The “Tale of Two Markets” Framework
AI/tech indices: S&P up 8% YTD; NASDAQ up 26% YTD; Semis up ~70% from March 30th lows
Real economy (cyclicals, consumer): Quietly suffering from oil shock; the broadening-out trade from early 2026 has decisively reversed
Hyperscalers committed to spend $755 billion in CapEx in 2026 — up 38% year-over-year — driving semiconductor earnings beats consistently
Q1 EPS tracking: +17% year-over-year (6th consecutive quarter of double-digit growth); forward 12-month EPS estimates revised up to +13%
Despite S&P up 8% YTD, the multiple has compressed from 22x to 21x — earnings are doing all the heavy lifting, not multiple expansion
2. Stagflation Shock and the Cyclical Collapse
Early 2026 thesis: Above-trend growth + below-trend inflation + Fed cutting rates = opportunity in homebuilders, consumers, industrials, cyclicals
What actually happened: Oil price spike from Middle East conflict drove a growth shock and inflation shock simultaneously
Clients who positioned for cyclical equities (homebuilders, consumers — benefiting from tax refunds from the “one big beautiful bill” and Supreme Court tariff ruling) were instead forced to rotate back into AI/hyperscalers
AI stocks viewed as “inelastic demand” — able to withstand macro pressures that consumers cannot
Hot CPI and PPI this week; market now pricing in potential Fed hike rather than cuts — “not a good setup for the real economy”
3. Is It a Bubble? Goldman’s Analytical Response
Tatasia’s definition of a bubble: A group of stocks goes up on “air” and when it pops, returns to starting levels
His answer: Not a bubble — hyperscaler CapEx at $755B is real; semiconductor earnings are genuinely beating because the underlying demand exists
Goldman’s GS Risk Appetite Indicator: At 5-year highs
Goldman prime book semiconductor exposure: Making new record highs every single day
Key concern: Leverage in levered ETF products — 2x and 3x semiconductor ETFs are “short gamma” products; when the underlier goes up, they buy on rebalance; when it goes down, they sell — a 3% decline can turn into 10% because of this forced de-leveraging
4. AI Hardware Evolution: The Next Trade
Phases of the AI trade (evolution over 3+ years):
Goldman baskets team’s next trade: Liquid cooling for data centers
Traditional air cooling uses almost as much electricity as the AI compute itself
Liquid cooling can reduce cooling power consumption by up to 10x
Theme is up ~30% YTD and expected to accelerate as hyperscalers adopt the technology
5. What to Watch Next
30-year bond yields at 5%+: Described as a “huge psychological level” for equity investors; speaks to “tough position with rising inflation into potentially weaker growth”
Nvidia earnings next week: Largest company in the world reporting; will show CapEx direction and next steps of AI trade
US-China meeting outcomes: China equities have been “notable underperformer”; investors watching whether technology space opens up and Chinese AI becomes “investable”
Trends Identified
1. AI Hardware Trade Now Entering the Infrastructure Layer
The progression of the AI trade from semiconductors to hyperscalers to data centers to memory to fiber optics to liquid cooling represents a systematic expansion of which companies benefit from the AI buildout. Goldman’s baskets team sees liquid cooling as the next wave — a theme that is still early (30% YTD) compared to the more mature semiconductor and fiber optic trades. This framework of “AI hardware store shovels” is proving highly useful for identifying the next trade.
2. Stagflation Is the Real Economy’s Problem, Not the Index’s
The most important structural insight from the episode is that the S&P 500’s AI/TMT concentration insulates the index from stagflationary shocks that genuinely hurt the consumer, homebuilder, and cyclical economy. Clients who positioned for broad economic strength (based on tax cuts, Supreme Court tariff ruling, potential Fed cuts) have been wrong, but the index hasn’t punished them — because the index doesn’t represent the real economy anymore.
3. Leverage in AI Trade Is the Hidden Risk
The proliferation of 2x and 3x leveraged semiconductor ETFs has created a dangerous amplification mechanism. These products are structurally “short gamma” — they must buy on up days and sell on down days to maintain constant leverage. This means that any negative fundamental catalyst that might normally cause a 3% decline can cascade into a 10% decline due to forced de-leveraging. This is the key mechanical risk that investors are underappreciating. —-
Sentiment Analysis
Overall Market Sentiment: Constructively Bullish With Significant Leverage Risk
Sean Tatasia’s tone was fundamentally constructive on the AI trade but highly alert to the leverage-driven volatility risk embedded in the current market structure.
Risk Factors Highlighted
30-year yield at 5%: Psychological threshold already breached; speaks to structural capital constraints as AI buildout competes with government borrowing for a limited capital supply
Leveraged ETF de-leveraging cascade: 2x/3x semiconductor ETFs create forced selling that amplifies any fundamental negative surprise; “3% can turn into 10% very quickly”
GS Risk Appetite Indicator at 5-year highs: Signals excessive optimism; historically correlated with subsequent volatility
Stagflation impacting real economy while masking in index: The index’s AI concentration hides genuine economic stress; if AI spending ever slows, the exposed consumer/cyclical weakness would be fully revealed
Nvidia earnings execution risk: As the largest company, any negative guidance revision could disproportionately impact sentiment and trigger the leveraged ETF cascade
US-China relationship: AI investability question: Chinese tech equities are the “notable underperformer”; if the US-China meeting doesn’t open a path for Chinese AI companies to be considered investable by Western investors, a significant recovery catalyst is deferred
This episode was covered in today’s The Market Signal — 2026-05-18, a cross-source synthesis of multiple podcast reports.