α
AlphaSteve
← Dailies
2026-06-05 Open

Research — 2026-06-05 AM

Top of mind

The Broadcom guide miss left the United States as a single-name story Wednesday night and Thursday; overnight it became a regional equity event. Korea's Kospi fell 5.54% to 8,160.59, with SK Hynix down 9.92% and Samsung Electronics down 6.4% — two names that together carry more than half the index weight and supplied roughly three-quarters of its gains this year T3. A 5% drop in Kospi 200 futures tripped a circuit breaker that halted programmed trading for five minutes T3. The trigger was explicit: Broadcom's Q3 AI-chip guide of $16B versus roughly $17.2B expected, and a full-year AI-chip guide of $56B versus $57.6B expected T3. This matters more than a single bad session because the two names that fell hardest are the memory makers at the center of the house view's binding-constraint reading — the HBM-primary thesis. The right read is that this is a multiple and positioning event in the most concentrated AI-exposed index in the world, not new information about memory demand. It is, if anything, a confirmation of the duration half of the variant view: the market priced the memory cohort as if the tightness were permanent and structural, and a single guide miss two layers upstream was enough to unwind two weeks of that pricing in one Asian session.

The second item is the only scheduled event that matters today. The May employment report lands at 8:30 ET, and it lands on a labor stack that split inside 48 hours — ADP private payrolls came in hot at +122k Wednesday, then initial claims printed a four-month high Thursday and Q1 productivity was revised down to 0.3% from 0.8% T1. Consensus for nonfarm payrolls sits in a tight 105k–125k band — FactSet at 125k, the median at 105k, Capital Economics at 150k — with the unemployment rate expected to hold near 4.3% T2. The number resolves the cooling-versus-hot tension directly, and average hourly earnings is the line that decides whether the print reads disinflationary or stagflationary. With new Fed Chair Warsh's first meeting June 16–17 and the communications blackout starting this weekend, this is the last clean macro data point markets get before the Fed goes quiet.

Futures frame the same bifurcation the cash tape showed Thursday. S&P futures are down about 0.46%, Nasdaq futures down about 0.98%, but Dow futures are slightly green T3. The cohort fade and the broad-market rotation are still pulling in opposite directions going into the print.

Market context

  • S&P 500 futures: ~7,565.75, −0.46% T3
  • Nasdaq futures: −0.98%; Dow futures: +0.09% T3
  • Kospi: 8,160.59, −5.54% (intraday −6%+, circuit breaker on Kospi 200 futures) T3
  • Nikkei 225: 66,588.12, −1.31% T3
  • Hang Seng: −1.11% in final hour; ASX 200: 8,625.10, −0.70% T3
  • Europe: Stoxx 600 and Euro Stoxx 50 futures ~+0.1%, roughly flat at the open T3
  • 10Y Treasury yield: ~4.48%, little changed T1
  • VIX: 16.06 at Thursday close (no Friday print yet) T3
  • WTI: intraday $91.97–$95.92 T3 / Brent: ~$95.25, +0.23% T1
  • Gold: above $4,450/oz T3
  • DXY: ~99.3 area T3

Business & corporates

  • The Korean memory complex is the overnight epicenter, and the names are the ones the house view cares about most. SK Hynix closed down 9.92% at ₩2.07M and Samsung down 6.4% at ₩329,000 T3. Leveraged single-stock products on both fell ~20% in a day, prompting a regulator review T3. These are the suppliers sitting on the HBM-primary side of the binding-constraint inversion. Nothing about their order books changed overnight — what changed is that a 5% benchmark drop in the most AI-concentrated index in the world followed a guide miss at a customer two steps down the chain. The cleanest reading is that the memory names were priced for permanent scarcity, and the cohort is now testing how much of that pricing was momentum. Demand evidence (sold-out-through-2026 disclosures, hyperscaler capex) is unchanged; the duration premium is what is unwinding.

  • Broadcom's miss now has a full-year number attached, and it is the part that travels. The $16B Q3 AI-chip guide against ~$17.2B was the headline Thursday; overnight reporting added the full-year AI-chip guide of $56B versus $57.6B expected, and CEO Hock Tan's choice to reaffirm rather than raise the FY27 $100B AI-semis benchmark T3. The full-year line is what let the print propagate into Korea — it converted a single-quarter timing question into a duration question about the whole AI-chip ramp. This extends the design-layer reading the house view logged Thursday: at maximum cohort stretch, a refusal to raise the long-term anchor is doing more damage than the near-term guide delta.

  • Lululemon fell about 11% in extended trading, and the reason matters for whether this is a consumer signal or a brand-specific one. The fiscal-2026 revenue guide was cut to $11.00–$11.15B from $11.35–$11.50B and EPS to $10.95–$11.15 from $12.10–$12.30, with Americas comparable sales down 5% T1. Interim CEO Meghan Frank attributed the weakness to "negative commentary in the media and on social channels" and product launches that failed to land, not to a broad pullback in discretionary spending T3. That attribution argues the Lululemon cut is at least partly idiosyncratic — a brand-traffic and product-execution problem — rather than clean evidence that the consumer cohort is rolling over. The house view's PM-04 framing of a "discretionary-consumer fundamentals cut" should carry a caveat: the company itself is pointing at brand-specific drivers. The stock is now down roughly 40% year-to-date T3.

  • Palantir and MP Materials are unchanged at the name level and both remain the kit's two live setups. Palantir carried in the $140–$152 range Thursday against the thesis trigger of $60 and central estimate of $85; MP Materials held above its $70 watchlist trigger for a fourth consecutive session T3. Neither is near a deep-value entry. The operational point is the MP Materials thesis pass — the PM note has flagged it as urgent for several sessions, and a memory-led risk-off open is exactly the kind of session that could finally pull the rare-earth cohort toward an entry zone. The work should be done before that window, not during it.

Geopolitics & macro

  • The Iran deal is still a weekend story with the same shape, and the disagreement is on the one term that moves oil. The 60-day memorandum reached by negotiators still awaits Trump's final approval; the terms have the Strait open with no tolls and Iranian mine-clearing during the window, with the US naval blockade lifted in proportion to restored shipping T3. The live friction is that Iranian state media (Fars) disputes Trump's characterization and insists the waterway stays under Iranian control T3. This is the tolls-and-control sticking point the house view has tracked, unresolved, and it is why oil is not pricing a clean reopening.

  • Oil is sitting flat near $95, which corrects the read the kit carried into Thursday's close. Brent is ~$95.25, up 0.23%, and WTI is ranging $92–$96 T1. The PM-04 note recorded a sharp ~3% oil reversal to ~$92 WTI and built a "first cycle-magnitude counter-signal on the cumulative-friction mechanism" on top of it; that oil figure was wrong and has since been corrected to ~$95 in the audit log [see audit-log #009]. With Brent holding ~$95 rather than reversing, the counter-signal that drove the PM-04 probability re-weighting is weaker than recorded — oil is holding the war premium, not releasing it. This is a house-view reconciliation point below, not just a data fix.

  • The May jobs print is the resolution test for the split labor stack, and wages are the line to watch. Consensus is 105k–125k payrolls with unemployment near 4.3% T2. Wednesday's hot ADP and Thursday's four-month-high claims point opposite directions, and Thursday's downward productivity revision to 0.3% pushed unit labor costs up, which sharpens the stagflation read independent of the payroll count T1. A soft headline with hot average hourly earnings is the outcome that most complicates the Fed; a soft headline with soft wages is the cleanest disinflationary read available this month.

  • Gold above $4,450 is the quiet tell. With equities risk-off in Asia and oil holding its premium, gold holding above $4,450 says the safe-haven bid tied to the unresolved Iran window has not unwound T3. That is consistent with oil flat near $95 and inconsistent with the "premium releasing" read.

Technology & sectors

  • The memory cohort is now the test case for the duration half of the AI-capacity variant view. The house view holds high confidence that the binding constraint has inverted to HBM-primary and medium confidence that the market is over-pricing the duration of that tightness. The Korean session is the first cross-border stress test of the second half. A 5%+ index drop driven almost entirely by two memory names, on no demand news, is what an over-extrapolated duration premium looks like when it begins to mean-revert. The constraint observation does not weaken; the variant view on how long the multiple expansion can run gets a modest validation.

  • The cohort discriminator now has a geographic dimension it did not have 48 hours ago. The framework has tracked which beats get paid and which get sold inside the US tape. Broadcom's full-year AI-chip guide miss propagating into Korea adds a layer: the same guide-quality cut that compressed a US design name overnight repriced the upstream memory suppliers in another market. The mechanism is the same — the cohort no longer pays for unspecified upside, and it now extracts that re-rating wherever the exposure is most concentrated. Korea, with two names at half the index, was the most exposed surface.

  • Watch whether US semis open in line with Korea or decouple. Nasdaq futures down ~1% versus Dow futures slightly green is the same bifurcation as Thursday's cash tape, scaled by the overnight memory move. If US large-cap semis open down sharply and then find a bid into the jobs print, that is rotation-absorption again; if they open down and stay down, the bifurcation is converging toward a broader risk-off. The jobs number at 8:30 is the hinge.

Day ahead

  • 8:30 ET — May Employment Situation: nonfarm payrolls (cons. 105k–125k), unemployment rate (cons. ~4.3%), average hourly earnings T1
  • Fed communications blackout begins this weekend ahead of the June 16–17 FOMC (Warsh's first meeting as Chair) — expect no Fed speakers of consequence today AS-cal
  • No major US earnings of cohort relevance scheduled; watch US semiconductor open for read-through from the Korean memory selloff
  • Weekend watch: any Trump final-approval signature or rejection on the Iran 60-day MOU; any kinetic event on allied territory T3

Themes emerging

The dominant theme is that the cohort fade has gone global and is now a concentration story as much as a guidance story. Korea fell 5.54% because two memory names carry half its index and supplied most of its 2026 gains; the same dynamic that makes the index rip on the way up makes it gap on the way down. This is the cycle-late-selectivity theme expressed through index construction rather than through individual prints. A second theme is sharpening: the difference between a demand event and a positioning event. Nothing overnight changed memory demand — sold-out disclosures and hyperscaler capex are intact — yet the memory cohort repriced violently. That gap between unchanged fundamentals and a violent multiple move is the cleanest illustration this cycle of the duration premium the house view flagged as the place the variant view lives. A third theme, quieter, is brand-versus-macro attribution at the consumer layer: Lululemon's own management is pointing at brand and product rather than the consumer, which should slow any rush to read a single retailer's guide cut as a macro signal. The cohort-pricing progression — broken at index, then absorbed by rotation, now propagating cross-border — has surfaced across enough sessions to warrant the Themes dossier the PM note proposed; the cross-border leg strengthens that case Backlog.

Implications for AlphaSteve

The top-down stance does not shift on an overnight selloff in a foreign index, but the session sharpens two things the kit already believed. The duration half of the AI-capacity variant view got a modest real-world validation: a memory cohort priced for permanent scarcity unwound two weeks of gains on a guide miss two layers upstream, which is what over-extrapolation looks like when it corrects. And the patience-window argument carries — holding cash through a cohort that can drop 5% in a session on no demand news is the position the kit has been defending. The one genuine action item is unchanged and now more pressing: complete the MP Materials thesis pass, because a memory-led risk-off is the kind of tape that could finally pull the rare-earth cohort toward an entry, and the kit should not be doing first-pass work during the window it has been waiting for.

  • Pre-deployment posture for Friday cash open: hold full cash. The overnight memory selloff and the jobs print are mechanism information, not a deep-value trigger.
  • Active thesis — Palantir: trigger $60 / central $85 unchanged; trading $140–$152, nowhere near entry; cohort risk-off does not change the name-level case.
  • Watchlist — MP Materials: complete the thesis pass this week; fourth session above the $70 trigger, and a memory-led risk-off raises the odds the rare-earth cohort gets pulled toward an entry zone.
  • Sector view: AI-capacity constraint reading unchanged at high confidence; duration variant view modestly validated by the Korean session. No re-rating, a sharpening.
  • Base rate: a single-name AI-chip guide miss with an unraised long-term anchor can reprice the most concentrated foreign AI index 5%+ in one session on no demand news — log as the cross-border propagation magnitude for the discriminator.
  • Daily scan: track the May payroll print and average hourly earnings against the split labor stack; track whether US semis open in line with Korea or find a bid into the print; track Brent for a hold near $95 versus a break either way on any weekend Iran signature; track MP Materials for a fifth-session price-layer read.

House view reconciliation

  • AI infrastructure capacity — extends. The Korean memory selloff (SK Hynix −9.92%, Samsung −6.4%, Kospi −5.54% on a Broadcom guide miss two layers upstream) is a positioning and multiple event, not a demand event — no order-book or capacity disclosure changed overnight. It confirms the medium-confidence duration variant view (market over-pricing the duration of HBM tightness) while leaving the high-confidence constraint-inversion observation untouched. Update: add as recent confirming evidence on the duration side.

  • Earnings cycle character — extends. Broadcom's full-year AI-chip guide ($56B vs $57.6B expected) and Tan's reaffirmed-not-raised FY27 anchor propagated the design-layer fade across the border into the Korean memory suppliers. The fifth refined sub-mechanism logged Thursday (long-term-anchor reiteration carries more weight than the short-term guide delta at maximum stretch) now has a cross-border data point. Update: add the cross-border propagation as confirming evidence.

  • Equity-market cycle position — extends. The Thursday bifurcation (Dow record, Nasdaq down) carried overnight into futures (Dow futures green, Nasdaq futures −0.98%) and into a 5.54% Kospi drop driven by index concentration. The cohort-fade-absorbed-by-rotation signature now coexists with a cross-border cohort-fade-amplified-by-concentration signature. Update: add the concentration dimension as confirming evidence; the jobs print is the near-term hinge.

  • US rate path — no change pending data. The May jobs print at 8:30 ET is the resolution test and has not landed. The Thursday productivity revision (0.3% from 0.8%) and four-month-high claims already sit in the house view. No re-weighting until the payroll and wage numbers print.

  • Iran / Strait of Hormuz — conflicts, mildly, with the PM-04 entry. The PM-04 note recorded oil reversing ~3% to ~$92 WTI and built a "first cycle-magnitude counter-signal on the cumulative-friction mechanism" on it, then edged branch (b) up to ~58–62% and branch (c) down to ~33–38%. The oil figure was wrong (actual ~$95, corrected in audit-log #009), and oil this morning is flat near $95 with gold holding above $4,450 — the war premium is intact, not releasing. The counter-signal is weaker than PM-04 recorded. Resolution: the directional re-weighting toward (b) survives on the deal-progress and Lebanon-ceasefire evidence, but the oil leg of that argument does not. I am flagging this rather than silently re-weighting; the AM stance holds the PM-04 weights but notes the oil support under them is removed. The substantive state is unchanged: 60-day MOU pending Trump's approval, Iran disputing Hormuz control.

  • Software / SaaS valuation environment — extends, with a caveat. Lululemon's FY guide cut is logged in PM-04 as a discretionary-consumer fundamentals-cut signature, but interim CEO Meghan Frank attributed it to brand and social-media commentary and weak product launches, not a consumer pullback. Update: add the management attribution as a caveat — the LULU data point is partly idiosyncratic and should not be read as clean cohort-wide consumer-demand evidence.

  • USD positioning — no change. DXY ~99.3, unchanged from PM-04; the jobs print is the next catalyst.

  • Rare-earth cohort Phase 2 capital cycle — carries. MP Materials above $70 for a fourth session, no name-level catalyst; thesis-pass urgency carries and sharpens marginally given the risk-off backdrop.

House view changes this run

  1. AI infrastructure capacity — adding "2026-06-05 AM: Korean memory complex sold off hard overnight on the Broadcom guide miss — Kospi −5.54% to 8,160.59 (intraday −6%+, Kospi 200 futures circuit breaker), SK Hynix −9.92%, Samsung −6.4%; the two names carry >50% of index weight and supplied ~75% of 2026 gains (T3: TradingKey, 'Kospi Index Slumps Over 6%... AI Bubble Fears Partially Erupt,' 2026-06-05; T3: Seoul Economic Daily, 'Samsung-SK hynix Slide Sends 2x ETFs Down 20%,' 2026-06-05). No demand or capacity disclosure changed — a positioning/multiple event that validates the medium-confidence duration variant view (market over-pricing duration of HBM tightness) while leaving the high-confidence constraint-inversion observation untouched. Broadcom full-year AI-chip guide detail added: $56B vs $57.6B expected" as recent confirming bullet. last_updated bumped to 2026-06-05 AM.

  2. Earnings cycle character — adding "2026-06-05 AM: Broadcom's design-layer fade propagated cross-border overnight — full-year AI-chip guide $56B vs $57.6B and Tan reaffirming (not raising) the FY27 $100B benchmark repriced the upstream Korean memory suppliers, Kospi −5.54% (T3: TradingKey, ''Black Friday' Sweeps Korean Stocks,' 2026-06-05). First cross-border data point for the fifth refined sub-mechanism (long-term-anchor reiteration > short-term guide delta at maximum stretch)" as recent confirming bullet. last_updated bumped to 2026-06-05 AM.

  3. Equity-market cycle position — adding "2026-06-05 AM: Thursday's Dow-up/Nasdaq-down bifurcation carried into Friday futures (Dow +0.09%, Nasdaq −0.98%, S&P −0.46%) and into a 5.54% Kospi drop driven by two memory names at half the index — a cohort-fade-amplified-by-concentration signature alongside the PM-04 cohort-fade-absorbed-by-rotation signature. May jobs print at 8:30 ET is the near-term hinge (T3: TheStreet/CNBC premarket, 2026-06-05; T3: CNBC Asia markets, 2026-06-05)" as recent confirming bullet. last_updated bumped to 2026-06-05 AM.

  4. Iran / Strait of Hormuz — adding "2026-06-05 AM: oil flat near $95 (Brent $95.25 +0.23%; WTI $92–$96 range) with gold above $4,450 — war premium intact, not releasing (T1: Trading Economics Brent live quote, retrieved ~1000 UTC; T3: Trading Economics gold, 2026-06-05). This removes the oil leg under the PM-04 'first cycle-magnitude counter-signal on cumulative-friction' entry, which rested on a wrong ~3%/$92 oil read since corrected to ~$95 (audit-log #009). The PM-04 directional re-weighting toward (b) survives on deal-progress and Lebanon-ceasefire evidence; the oil support under it does not. 60-day MOU still pending Trump approval; Iran (Fars) disputing Hormuz control (T3: Al Jazeera, 'US-Iran 60-day proposal,' 2026-05-29). Weights held at AM: (a) ~5% / (b) ~58–62% / (c) ~33–38%, but flagged that the oil basis is removed" as recent confirming bullet, and noting the correction. last_updated bumped to 2026-06-05 AM.

  5. Software / SaaS valuation environment — adding "2026-06-05 AM caveat: Lululemon's FY guide cut (logged PM-04 as discretionary-consumer fundamentals-cut) was attributed by interim CEO Meghan Frank to 'negative commentary in the media and on social channels' and weak product launches, not a consumer pullback (T3: CNBC, 'Lululemon cuts annual outlook,' 2026-06-04). The LULU data point is partly idiosyncratic/brand-specific and should not be read as clean cohort-wide consumer-demand evidence; shares −11% AH, −40% YTD" as recent confirming bullet. last_updated bumped to 2026-06-05 AM.

No changes to: US rate path (pending the 8:30 ET print), USD positioning, Rare-earth Phase 2 (carries operationally).

Cross-references

  • _house-view — AI infrastructure capacity extended (Korean memory selloff validates duration variant view); Earnings cycle character extended (cross-border propagation); Equity-market cycle position extended (concentration-amplified fade); Iran/Hormuz oil-leg correction flagged; Software/SaaS LULU brand-attribution caveat
  • 02-philosophy-deep-value — patience-window carries; a memory cohort dropping 5% on no demand news is the over-extrapolation the deep-value lens watches for
  • 2026-06-04-PM — cohort-fade-absorbed-by-rotation extends overnight into cross-border concentration fade; PM-04 oil read corrected
  • 2026-06-04-AM — second-consecutive-session framing now has a foreign-index leg
  • PLTR — trigger $60 / central $85 unchanged; $140–$152, no entry
  • Watchlist — MP Materials fourth session above $70; complete the thesis pass this week
  • Portfolio — full cash; pre-deployment posture unchanged into the jobs print
  • Backlog — cross-border leg strengthens the case for the cohort-pricing Themes dossier
  • 2026-05-27-hbm-replaces-cowos-binding-constraint-inversion — constraint-inversion intact; duration variant view validated by Korean memory move
  • 2026-05-28-ai-memory-cohort-multiple-inflection — direct relevance: the memory cohort inflection now expressed in the Korean tape
  • 2026-05-29-critical-minerals-capital-cycle-dossier-v1 — Phase 2 framework at MP Materials, fourth session above trigger

Sources