Last week ended with one number on every desk: 79%.
That’s the implied probability of a January 2027 rate hike by Friday’s close — up from roughly 9% when the week began. The move wasn’t built on a single headline. It was built on a Fed that’s openly fracturing, an NVIDIA print that couldn’t lift the stock, and a string of regional data points that all leaned the same direction.
📈 The Market Backdrop
Under the surface, the rotation that started in early May kept working. Tech leadership stayed thin. Hedge funds kept trimming exposure. And the rally continues to lean on a handful of names doing most of the heavy lifting.
The cleanest tell was Wednesday night’s NVIDIA print — a textbook beat that failed to translate into a green tape Thursday. That doesn’t happen in early rallies. It happens in late ones.
🏦 Warsh Inherits a Divided Fed
Kevin Warsh was sworn in as Fed Chair on Friday. The committee he’s stepping into is no longer pulling in one direction.
Minutes from the late-April meeting, released Wednesday, showed a majority of officials signaling that hikes would become appropriate if inflation stayed sticky. Four members dissented — the most since 1992. That’s not a footnote. It tells you the room is starting to disagree on whether the next move is up or sideways.
Markets repriced immediately:
- January 2027 hike odds: ~9% → 79% in a single week
- The repricing happened before Warsh had a chance to say a word in public
- Every subsequent Fed speaker now matters more than usual
Warsh’s first real test is data, not rhetoric. PCE drops Friday, and his Fed will be reading the same number the market is.
🔥 NVIDIA Beat — And the Stock Fell Anyway
NVIDIA reported Q1 after the close on May 20. Every line item came in ahead:
- Revenue: $81.6B vs. $79.2B consensus, up 85% YoY
- EPS: $1.87 vs. $1.78
- Data Center revenue: a record $75.2B, up 92% YoY
The reaction was a sell. That’s now the familiar pattern after NVDA prints, and the interpretation matters: when a stock needs flawless results just to hold, the marginal buyer has already shown up. Positioning is full. Expectations are full. “Great” stops being enough.
This was the warning that ran in last week’s preview — Nvidia isn’t just a stock anymore, it’s a positioning gauge. And the gauge is reading crowded.
📊 This Week’s Key Data
Memorial Day shortens the week. Thursday and Friday carry almost everything that matters.
Tuesday, May 26
- 9:00 AM — S&P Case-Shiller Home Prices (March)
- 10:00 AM — Consumer Confidence (May) — forecast 92.0, prev 92.8
Wednesday, May 27
- Nothing scheduled. Earnings will run the tape.
Thursday, May 28
- 8:30 AM — Initial Jobless Claims — forecast 213K, prev 209K
- 8:30 AM — Durable Goods Orders (April) — forecast 3.3%, prev 0.8%
- 8:30 AM — Durables ex-Transportation
- 10:00 AM — New Home Sales (April) — forecast 665K, prev 682K
Friday, May 29 — the day
- 8:30 AM — GDP Q1 (Second Revision) — forecast 2.0%, in line with first read
- 8:30 AM — PCE (April) — forecast 0.5% MoM / 3.8% YoY
- 8:30 AM — Core PCE (April) — forecast 0.3% MoM / 3.3% YoY (prev 3.2%)
- 8:30 AM — Personal Income (0.4%) and Spending (0.5%)
- 9:45 AM — Chicago PMI — forecast 51.0, prev 49.2
Core PCE ticking up to 3.3% YoY would put the rate-hike narrative on much firmer ground. A miss in the other direction takes some of the air out of last week’s repricing — but doesn’t reverse it. The bar to walk that 79% number back is high.
📈 Earnings That Will Move the Tape
Wednesday after the close is the headliner. Three names. All AI-adjacent.
- Salesforce (CRM) — Wed AC. First enterprise software read on whether AI spend is actually showing up in customer commitments, or still stuck in pilot mode.
- Marvell (MRVL) — Wed AC. Custom AI silicon for hyperscalers. The “is the build-out still accelerating” question gets a direct answer here.
- Snowflake (SNOW) — Wed AC. Cloud data infrastructure. A read on whether enterprise AI projects are moving from POC to production.
Thursday is the consumer-and-infrastructure cleanup:
- Costco (COST) — Thu AC. Probably the cleanest single read on the U.S. consumer right now. Watch membership renewals and the discretionary-vs-staples mix.
- Dell (DELL) — Thu AC. AI server backlog. Coming off a big quarter, so the bar is high.
If CRM, MRVL, and SNOW all miss or guide down, the AI-infrastructure trade takes a real dent heading into June. If they hold, the narrative survives another quarter.
💼 Main Street Is Already Showing Cracks
It’s easy to underweight softer prints when the headline indexes are still near highs. Last week made it harder:
- Philly Fed collapsed
- Services PMI sat just above contraction
- Mortgage rates pushed back near a one-year high
This is the slowdown showing up in second-tier data first. Consumer Confidence on Tuesday and New Home Sales on Thursday will tell us whether it’s bleeding through to the consumer.
If those come in soft and PCE comes in hot, the word that gets reintroduced is stagflation. That’s the trade nobody wants to defend, and the one the market is least positioned for.
⚠️ Market Structure: The Signal Is in the Tape
You don’t need a macro framework to read what NVDA did. A blowout that doesn’t rally is positioning telling you the trade is full.
- Hedge funds are still de-grossing tech exposure
- Leadership has narrowed, not broadened
- Implied vol has stayed bid into the strength
Same setup that’s been building since early May, just further along. The tape can still hold. But the air is thinner up here, and individual disappointments hit harder than they should.
🌍 Bigger Picture: What’s Priced In
Base Case
- Core PCE prints in line at 3.3%
- Warsh communicates patience early
- Jan 2027 hike odds stay in the 60–80% range
- Indexes chop, tech wobbles
Downside Risk
- Core PCE prints hot (3.4%+)
- Hike odds break above 90%
- CRM / MRVL / SNOW disappoint
- Tech leadership cracks and breadth fails to step in
Upside Case
- PCE comes in cool
- Hike pricing fades back below 50%
- AI earnings reaffirm the build-out
- Rally broadens into laggard sectors
Bottom Line
→ Warsh inherits a divided Fed and a market that just repriced hawkish
→ NVIDIA’s failed rally is the clearest positioning tell in months
→ Friday’s PCE sets the tone for June
→ Wednesday’s AI earnings are the cleanest read on the trade everyone is in
→ The economy is cooling underneath, and the data is starting to show it
“You don’t need a recession to break a rally. You just need positioning that’s already maxed out — and a calendar that won’t give it a break.”