Research — 2026-06-27 AM

Top of mind

The week just closed with the AI-financing story moving from a thesis to a tape event. The Nasdaq fell a fifth straight session Friday and lost 4.6% on the week, its worst stretch in months, while the report that triggered it hardened: the New York Times now says OpenAI is weighing a delay of its listing into 2027 rather than price below $1T, with Sam Altman calling any cut to the trillion-dollar figure a "nonstarter" T3. The company filed its S-1 confidentially on June 8 and is targeting roughly $600B of compute through 2030 against revenue that has not hit its own marks T3. This is exactly the sequence the 2026-06-26-ai-buildout-financing-turn-v1 dossier opened on Friday predicts: in a debt-and-issuance-funded capacity cycle, the financing turns before the demand does, and it cracks first in the primary market. A delayed IPO held back for a valuation the tape will not currently pay is a primary-market refusal — the cleanest early form of that turn. SoftBank, the largest expected outside holder, fell as much as 13% on the report T3.

The discipline here is to keep the two scares separate, as the house view already does. The demand-air-pocket version stays rebutted — Micron's accelerating guide and the memory complex's structural tightness are not in question. What is being priced is the supply-curve and capital-cycle leg: whether the builders can keep funding the next round, and whether the multiples paid for the build over-extrapolate its duration. Friday's tape priced that as rotation, not rupture — the Dow still finished the week up 0.6%, the S&P off only about 2%, the VIX at 18.89, defensives bid. America rotated; it did not panic. That distinction is the whole content of the equity-cycle position right now, and Friday confirmed it again.

The week ahead is the first real test of whether the rotation stays orderly. It is a holiday-shortened, data-heavy week — the June jobs report lands Thursday, a day early, with US markets closed Friday for the observed Independence Day T3. A hot payroll print into a Fed that has already ratified a hawkish reaction function is the macro hinge.

Market context

US cash markets closed Saturday and Sunday; figures below are Friday June 26 settled / weekly. Next session Monday June 29.

S&P 500: 7,354.02, −0.05% Friday; ~−2% on the week T3
Nasdaq Composite: 25,297.62, −0.24% Friday — fifth straight down day; −4.6% on the week T3
Dow: 51,876.11, −0.09% Friday; +0.6% on the week (the week's relative winner on defensive rotation) T3
10Y yield: ~4.4% T3
VIX: 18.89, +1.4% T3
DXY: 101.27, near a one-year high T3
Oil: extended its decline despite a cargo-ship strike off Oman in the Strait of Hormuz — Brent held below ~$75 T3; specific WTI/gold settles not confirmable from a primary source at writing [T3 only; directional]
Gold: closed its worst week in months (~−5%) on hawkish-Fed signaling, having broken below $4,000 for the first time since November T3

Business & corporates

OpenAI / the AI-financing channel — The IPO-delay report is the load-bearing corporate item of the week, not for OpenAI's own economics but for what a held-back listing signals about the funding layer beneath the entire build. Altman is holding for $1T against a last private mark reported near $730–850B; Kalshi traders put roughly a 59% chance of an IPO being announced by March 2027, and the broader read is that unmet revenue targets plus $600B of compute commitments through 2030 are what is forcing the wait T3. A company that chooses not to raise public equity at the price on offer is the primary market declining to fund the build at the multiple the build needs. That is the financing turn the dossier is built to track, and it registered this week in the most visible place such turns always show first.
General Mills (fiscal Q4, Wednesday July 1) — The most thesis-relevant print of the week for AlphaSteve, because it is a direct read-through to Conagra, the closest watchlist name. The Street looks for roughly $0.80 EPS on $4.6B revenue T3. The tell is the FY27 guide, not the quarter. Packaged-food volumes have been soft and the staples cohort has traded on defensive rotation rather than fundamentals; a weak General Mills guide that drags the group could pull Conagra toward its $11.50 trigger, which sits only about 9% below Friday's level CAG. Watch the guide and the cohort reaction, not the headline beat or miss.
Nike and Constellation Brands (Tuesday June 30) — Both test the expectations-not-sectors discriminator in consumer names. Nike is expected to print about $0.13 EPS on $10.85B revenue, both down year over year T3; Constellation Brands reports fiscal Q1 2027 with roughly $3.21 EPS on $2.4B expected T3. Neither is a portfolio or watchlist name, but the reactions add cross-sectional evidence on whether the market is still selling confirm-rather-than-accelerate guides in consumer cyclicals the way it has in software.

Geopolitics & macro

June jobs report (Thursday July 2, a day early) — The week's macro hinge and the next live test of the rate-path position. Consensus is about 172k T3. The context that makes this load-bearing: the disinflation-substance leg of the kit's rate view already lost its pre-registered rematch when May core PCE printed 3.4% on June 25, above both consensus and the Fed's own 3.3% June dot T1. The Fed has ratified higher-for-longer as its own central case via the June dots (2026 median funds rate 3.8%, nine of eighteen members projecting a hike) T1. A payroll print at or above consensus reinforces the hawkish base directly; only a clear downside miss reopens any cut conversation against the dots. The release moving up a day for the holiday compresses the week's risk into Thursday.
Iran / Strait of Hormuz — The IRGC again declared the strait closed over the weekend, citing continued Israeli strikes in southern Lebanon as a breach of the memorandum, with Iranian media saying Tehran is suspending tanker traffic and weighing whether to pull out of the deal T3. This is the residual (c) Lebanon-seam firing again — the exact mechanism the 2026-06-26-PM and the Iran-flanks-decoupling theme point at — but it is the recurring factional rhetoric, not physical rupture: Iran has declared the strait closed repeatedly through this crisis, and Friday's tanker strike off Oman was met with oil falling, not rising, the cleanest possible look-through. The 60-day roadmap agreed in Switzerland on June 22 (working groups on nuclear, sanctions, and dispute resolution; a Hormuz de-confliction line; IAEA inspectors invited back) remains the governing framework T3. Branch (b), framework-with-friction, stays the live description. Monday's oil open is the first observable.
Warsh in Portugal (Wednesday July 1) — The new Fed chair speaks about Europe at 9:30 ET T3. A speech on Europe is unlikely to move the US rate path, but any unscripted line on the domestic reaction function lands into a tape that has already priced his June hawkish lean. Worth reading for tone, not expecting a signal — Warsh has said the Fed dropped forward guidance.

Technology & sectors

AI infrastructure — the supply-curve lean on the tape — Friday extended the cohort de-rating that has run all week, and the character matters: a five-day Nasdaq slide concentrated in chips and AI-adjacent names, bought in defensives, on no change to the underlying supply facts (HBM still sold out through 2026, Micron's guide still accelerating). That is positioning and multiple-duration repricing, not a demand break — the variant the 2026-06-05-ai-infrastructure-capacity-dossier-v1 names. The mispriced variable is the supply curve and the financing that funds it, not the demand line. SK Hynix's roughly $29.65B Nasdaq ADR, filed last week into the worst tape of the month and slated near July 10, is the textbook late-cycle issuance marker — capital raised at the top of the cycle, per the Baker-Wurgler issuance base rate T3. The ADR debut is now a forward observable in its own right.

Day ahead

Weekend — no US session today (Saturday) or tomorrow. Week ahead, US:

Mon Jun 29 — no major scheduled releases T3
Tue Jun 30 — Nike fiscal Q4 earnings; Constellation Brands fiscal Q1 2027 earnings
Wed Jul 1 — ISM Manufacturing PMI (June); General Mills fiscal Q4 earnings (~8:00 CT); Fed Chair Warsh speaks on Europe (Portugal, 9:30 ET)
Thu Jul 2 — June Employment Situation (released a day early; consensus ~172k); ISM Services PMI (holiday-shifted); weekly jobless claims
Fri Jul 3 — US markets closed (observed Independence Day)

Themes emerging

The dominant pattern across the last week of notes is the convergence of two standing positions onto one tape. The AI infrastructure capacity dossier (asset side) and the AI buildout financing dossier (liability side) are now being priced as a single story: the market is repricing the duration of the build and the durability of its funding at the same time, and the OpenAI IPO delay is the event that fused them. This is the second consecutive note in which the financing channel, not the demand channel, is doing the work — the dossier's core sequencing claim is on the tape early. A second, quieter theme is the divergence between how stress travels across regions: the prior week's Asian deleveraging (circuit breakers, SoftBank down double digits) did not cross into a US breadth break; the US response has been orderly defensive rotation. That deleveraging-versus-rotation split is worth watching as a cross-regional stress gauge — if a US session ever converts the rotation into an indiscriminate sell-everything tape, that is the regime change the equity-cycle position is waiting for, and the one that would finally bring defensives like Conagra into trigger range. Neither theme is new enough this run to spin out; both are tracked.

Implications for AlphaSteve

No top-down stance shift. The posture holds: work specific names where margin of safety is real, hold cash where nothing meets the threshold, and do not read the AI-led de-rating as either a buying signal or a market-wide break yet. The portfolio stays 100% cash by construction; nothing on the watchlist has triggered. The week's value to the kit is in two forward tests — the General Mills guide as a read-through to Conagra, and the jobs print as the rate-path hinge — neither of which calls for action today.

Active thesis PLTR: no read-through; price far out of band (~$113 area vs. $60 trigger). No action.
Watchlist CAG: General Mills July 1 print is a direct cohort read-through; a weak FY27 guide that drags staples is the most plausible near-term path toward the $11.50 trigger (~9% below Friday). Flag for the Wednesday scan.
Watchlist MP: no fresh evidence; price (~$55) above $50 central / $42 trigger. Carries.
Sector view: AI-cohort de-rating confirms the supply-curve / capital-cycle lean; no new sector promotion. The bearish read is on the multiple and the financing, not the demand.
Base rate: no update; the Baker-Wurgler issuance base rate is being actively exhibited by SK Hynix's ADR and the OpenAI IPO hold, not revised.
Daily scan: add the General Mills-to-Conagra read-through (Wed) and the SK Hynix ADR debut (~Jul 10) as forward observables; the deleveraging-versus-rotation discriminator stays a standing cross-regional gauge.

House view reconciliation

AI infrastructure capacity — confirms / extends. The five-day Nasdaq slide on no supply-side news, plus SK Hynix's ADR into a weak tape, extends the supply-curve / capital-cycle lean (per the AI infrastructure capacity position, last reviewed 2026-05-27 long-form, and the changes log through 2026-06-26 PM). Structural-demand concession stands. No band change.
AI buildout financing (dossier v1) — confirms. The OpenAI IPO-delay hardening to a NYT story with Altman's $1T "nonstarter" framing is a second-session confirmation of the financing-turns-first sequencing claim opened 2026-06-26. No weight or band change; the dossier is too young for one.
Equity-market cycle — confirms. Friday's orderly rotation (Dow +0.6% week vs. Nasdaq −4.6%; VIX 18.89; defensives bid) is the de-rating-as-rotation form the position settled on at 2026-06-26 PM. Confirmation in its mildest form; no band change.
US rate path — no new evidence; carries; Thursday is the next arbiter. The disinflation-substance leg already lost its May PCE rematch (2026-06-25 PM). The June jobs print is the next test of the hawkish base. No change this run.
Iran / Hormuz — confirms branch (b) / look-through; held at (a) ~40% / (b) ~52% / (c) ~8%. The weekend IRGC re-closure over Lebanon is the residual (c) Lebanon-seam firing again, consistent with the Iran-flanks-decoupling theme, but it is recurring factional rhetoric the look-through discipline neutralizes; the Oman tanker strike met with falling oil is the second-day look-through pass. No weight change. Monday's oil strip is the observable.
Software / SaaS, USD, rare-earth cohort, power equipment — no relevant new evidence; all carry.
Earnings cycle character — no view change; the week's consumer/staples prints (Nike, Constellation, General Mills) will test the expectations-not-sectors discriminator. Logged as forward observables, not evidence yet.

House view changes this run

No weight changes. No confidence-band changes. Confirming developments logged only.
AI infrastructure capacity / AI buildout financing: confirming developments logged (Nasdaq fifth straight down day and −4.6% week; OpenAI IPO-delay hardening to NYT with Altman $1T "nonstarter" and ~$600B compute-through-2030 framing; SoftBank −13%; SK Hynix ADR into a weak tape).
Equity-market cycle: confirming development logged (orderly defensive rotation; Dow +0.6% week vs. Nasdaq −4.6%; VIX 18.89).
Iran / Hormuz: held at (a) ~40% / (b) ~52% / (c) ~8%; weekend IRGC re-closure over Lebanon logged as a confirming Iran-flanks-decoupling / residual-(c)-seam data point, not a trigger.
Scan additions: General Mills-to-Conagra read-through (Jul 1) and SK Hynix ADR debut (~Jul 10) as forward observables.