Brandon Leon

Week ahead · Macro · jobs · memory · quantum · July 6, 2026

We said suppliers reprice last. The market sold them first.

Author Brandon Leon Posted 2026-07-06 (week of July 6; first Warsh-Fed minutes Wednesday, SK hynix lists Friday) Coverage Week ahead · the jobs print · the memory unwind · the Warsh minutes

TL;DR: The June jobs report came in soft and strange — +57K against a ~113–115K consensus, with the unemployment rate falling to 4.2% only because half a million people left the labor force. A September hike came off the table in pricing and commentary, and the Dow closed at a record after a 595-point jobs-day gain. But the week's real event was Meta: word that it will sell access to AI compute it isn't fully using cratered the suppliers we're long — Micron round-tripped its entire post-earnings rip while every megacap spender finished the week green. Our supplier-vs-spender call got inverted in week one, and we grade it below. The long end of the curve led yields higher all week, so our “calmer long end” call is leaking too. This week: the first minutes of the Warsh Fed Wednesday, $119B of Treasury duration on auction, and SK hynix's Nasdaq debut Friday — the event that tells us whether the memory unwind was flows or something worse.
Where we are

The most useful thing we can publish this week is a grade sheet. Eight days ago we wrote that when the AI cycle gets questioned, contracted supply reprices last and recovers first. In week one the market did the opposite: it sold the suppliers first and hardest — Micron −13.9% on the week, SanDisk −16.5%, the SOX −4.4% — while every megacap AI spender finished green and the Dow closed at a record.4 The trigger was not a capex cut. It was Meta reportedly moving to sell access to AI compute capacity it doesn't fully use — a spender turning supplier — landing on the most crowded trade of the first half, with Korean retail leverage unwinding behind it and SK hynix bringing a ~$29B listing to Nasdaq this Friday.5

The macro data broke the other way. June payrolls printed +57,000 against a ~113–115K consensus, with April and May revised down a combined 74K — and the 4.2% unemployment rate is an asterisk, not a strength, because it fell only on people exiting the labor force.1 Under the old regime that print would have markets pricing cuts. Under Warsh's hike-biased Fed it meant relief: a September hike came off the table, October stayed merely “potential,” and the Dow rallied 595 points to a record close on the news.2

The curve, however, kept arguing with us. The long end led yields higher all week — the 10-year rose from 4.38% to ~4.49%, tagging our 4.50% trip-wire intraday Wednesday, and the 30-year closed near 4.97% — even as the front end rallied on the soft jobs print.6 That bear-steepener is the exact shape of Schwab's scenario, not ours. The wire hasn't tripped on a closing basis. We grade it honestly in Section 1 anyway.

The week ahead is dense. Wednesday at 2 PM ET brings the first FOMC minutes of the Warsh Fed — the inside look at how close June actually came to a hike. Tuesday through Thursday the Treasury auctions $119B of 3s, 10s, and 30s into a long end already on edge. And Friday, SK hynix targets its Nasdaq debut — the memory event of the summer, and a direct read on Micron. We frame all three in Section 6, and the refreshed trip-wires in Section 7. Everything below is scenario work with named trip-wires, not point forecasts.

The 250th · A founder's ledger Portrait of Robert Morris, financier of the American Revolution

This weekend the republic turned 250, so here is a founding story for the week we just had. Robert Morris was reputed the richest man in revolutionary America and one of only two people — Roger Sherman was the other — to sign the Declaration of Independence, the Articles of Confederation, and the Constitution. He was also the Revolution's financier. When the Continental treasury was empty, which was constantly, Morris floated the war on his personal credit — his hand-signed “Morris notes” circulated when the government's own paper wouldn't — and he helped fund the final march to Yorktown out of his own ledger. Washington offered him the first Treasury secretaryship; he declined and recommended a young Alexander Hamilton instead.

Then the best credit in America destroyed himself the ordinary way: leverage. Morris borrowed to amass some six million acres of frontier land, the credit crunch of 1797 left him unable to roll his paper, and the man who financed independence spent three and a half years in a Philadelphia debtors' prison — where Washington, loyal to the end, came to dine with him. Congress passed the federal Bankruptcy Act of 1800 in part to get him out; he died broke and nearly forgotten in 1806. The lesson is older than the country it helped build, and it is this week's lesson too: no balance sheet is strong enough to save a leveraged, crowded position when the flows turn. Happy 250th — the ledger keeps the receipts.

Robert Morris (1734–1806). Public-domain portrait, via Wikimedia Commons.

1. The scoreboard: three calls, honestly graded

We put three falsifiable positions on the record last week. Here is how each did — scored the way we'd score anyone else.

Grade 1: The jobs base case — half right, wrong path

We wrote: “our base case is a number in the middle that keeps the hike a December question, not a summer one.” The consequence landed — the hike moved out, not in. But the path was wrong: +57K is not a middle number, it's a soft one, the smallest gain since February, and a break in the spring's streak of upside surprises.1 Our named trip-wire was for a hot print pulling the hike forward; the print that arrived was the opposite tail. Half credit, and the composition is the part we take least comfort in — more on that in Section 2.

Grade 2: “Suppliers reprice last” — wrong in week one

The miss · owned in full

We said contracted supply reprices last and recovers first. In the first week of the trade, the suppliers fell double digits while the spenders rallied.

Micron −13.9% on the week to $975.56 — a full round trip of the post-earnings rip, 19.6% below the June 25 record close. SanDisk −16.5%, Western Digital −8.1%, the SOX −4.4% with an 11% two-day collapse. Meanwhile MSFT, GOOGL, META, AMZN, AAPL — the spenders we were “neutral-to-cautious” on — all finished the week higher.4 On price, the split trade lost money in week one. That's the grade, and there's no dressing it up.

Now the dissection, because the why matters for what we do next. What we actually claimed was a floor under earnings: take-or-pay contracts, prepaid deposits, floor margins. That floor held — memory contract prices are still rising (TrendForce's fresh 3Q26 read has contract prices up another double-digit percentage quarter-over-quarter across both DRAM and NAND), and not one contract changed.5 What week one tested was a floor under the stock — and there isn't one, because the first half's leaderboard had made long-memory the most crowded position in the market: SanDisk was the single best stock in the S&P 500 in H1 at +858%, Micron was up over 300%, and the SOX had just printed its best quarter on record.3 Crowded trades don't reprice on fundamentals; they reprice on flows — and the Meta headline was the catalyst that flipped them. The reusable lesson, filed next to June's Warsh lesson: a contractual floor under earnings is not a floor under the stock when positioning is the marginal seller. The thesis survives; the tactics get humbler. The single-name numbers — including the base-case re-rate and the pre-committed stop — live in the standalone unwind note we published Thursday, where single-name numbers belong.

Grade 3: “Calmer long end” — leaking, not yet broken

Last week we named our disagreement with Schwab — we read the long end as the calmer part of the curve; they were defensive on it — and we set the trip-wire at a 10-year close above ~4.50% driven by term premium or supply, not a clean growth surprise — it carries this week as Trip-wire #2 below, upgraded to warm. The week's tape: 10-year from 4.38% to ~4.49% (+10–11bp, an intraday tag of the wire on Wednesday), 30-year from 4.87% to ~4.97%, while the 2-year rose just 7bp and then rallied on jobs day.6 The long end led. That is a term-premium-and-supply shape — hot JOLTS Tuesday, Warsh's “prices are too high” at Sintra Wednesday, and $119B of duration supply looming.7 By the letter of the wire: not tripped, no close above 4.50%. By the spirit: the week traded Schwab's scenario, not ours, and we say so. If this week's auctions take 4.50% out on a close, we execute the stated response — cut overall duration to below-benchmark — without relitigating it.

Simply: We made three calls last week. The jobs call got the consequence right (no summer rate hike) but expected a boring number and got a weak one. The “own the AI suppliers” call lost money in its first week — the stocks we liked fell hard while the ones we were cautious on rallied — and we explain why without excusing it. And long-term interest rates rose the way Schwab warned they might, not the way we expected. One half-right, one wrong, one leaking. We show the work below.

2. The jobs print: slack water, with an asterisk

The headline was a clean miss: +57,000 jobs in June against a consensus around 113–115K, the smallest gain in four months. April and May were revised down a combined 74K (April 179K→148K, May 172K→129K), which retroactively cools the spring “hot streak” that had the market bracing for hikes. Wages behaved — average hourly earnings +0.3% on the month, +3.5% on the year, both in line.1

The asterisk is the unemployment rate. It fell to 4.2% — a one-year low, after four straight months at 4.3% — and that is the wrong-reason kind of decline: labor-force participation dropped 0.3 points to 61.5%, the lowest since March 2021, and household-survey employment fell by roughly half a million. Fewer people working, fewer people looking. Indeed's economists called the market “slack water” — an unmoving tide, low hiring and low firing at once — and that's the honest read: not a collapse (jobless claims sit at 215K, still historically low), but a labor market losing motion.1

Chart 1 — The June jobs print, and the wrong-reason unemployment decline

Payrolls missed by half. The unemployment rate “improved” — because half a million people left the labor force.

Nonfarm payrolls (thousands of jobs) 129 May (revised) ~113 June consensus 57 JUNE ACTUAL Apr + May revised −74K combined The 4.2% asterisk Unemployment 4.3% 4.2% May → June Participation 61.8% 61.5% May → June Lowest participation since March 2021; household employment −507K

June 2026 Employment Situation (BLS, released July 2): nonfarm payrolls +57K vs. a ~113–115K consensus; April and May revised down a combined 74K. Unemployment fell 4.3%→4.2% while participation fell 61.8%→61.5% (lowest since March 2021) and household-survey employment dropped ~507K — a rate decline driven by labor-force exit, not hiring. Sources: BLS; CNBC; Yahoo Finance; Indeed Hiring Lab.

The translation under a hike-biased Fed is what matters for the tape. Under the old regime, +57K with falling participation reads dovish and markets price cuts. Under Warsh's Fed — the one that flipped the dots to an implied 2026 hike in June — the soft print bought relief: July hike odds fell to ~20%, September came off the table in the coverage, and October is “potential,” not priced.6 Bad news was good news, and the Dow's 595-point record close on jobs day is what that looks like. But hold the two halves of the week together and the mix gets less comfortable: a chair saying “prices are too high” on Wednesday, and a labor market going slack on Thursday.7 Softening employment plus a Fed that can't cut into 4%-handle inflation is a stagflation-adjacent mix we have to respect even though it is not our base case — it's the scenario where both our equity longs and everyone's duration get hit. We're naming it, because last month taught us not to fade the uncomfortable branch.

Simply: Hiring slowed to a crawl in June, and the “better” unemployment rate is an illusion — it only fell because half a million people stopped looking for work. The silver lining: a weak report means the Fed's threatened rate hike moves further away, which is why stocks actually rallied. The worry: a slowing job market plus inflation the Fed calls too high is a nasty combination if both keep going.

3. The Meta shock: our split thesis meets week one

Here is the sequence that inverted our trade. On Wednesday July 1, reports landed that Meta is building a cloud business to sell and lease its excess AI compute. Semis were routed within hours — Micron −10.6%, SanDisk down double digits — and the selling went global overnight: Korea's Kospi fell 7.9% Thursday, with SK hynix −14.6% and Samsung −9.1%, in what Seoul coverage simply called the Meta shock. Thursday added two more weights: Apple reportedly exploring sourcing memory from China's CXMT and YMTC for devices sold in China, and Michael Burry — already short NVDA, the SOXX, and equipment names as of Tuesday — adding a Micron short.4 Micron finished the week at $975.56: the entire post-earnings rip, repossessed, plus 7% below where it traded the night before the best quarter in memory history.5

Why a Meta cloud business hits memory harder than Meta. Our whole framework said the spenders' capex is the suppliers' revenue, so visibility favors the suppliers holding signed contracts. The Meta story attacks the premise from a direction we didn't draw: if a hyperscaler can resell its surplus compute, then overbuilt capacity doesn't show up as canceled orders years from now — it shows up as competing supply today, in the cloud market, at marginal prices — even while today's take-or-pay contracts remain fully in force. It's the growth beyond the contracted floor that resold capacity competes with. A spender becoming a supplier collapses the neat line our Section 2 drew last week. It also reframes the capex number itself: some part of that $700B+ was building inventory of compute, not consuming it. The market repriced the entire upstream chain on that single reframe — which is exactly what you'd expect when the upstream chain is also the most crowded trade of the half.

What survives, and what we're watching. Three things. First, the demand floor under memory earnings is intact: contract prices are still rising into 3Q, the take-or-pay agreements are unchanged, and nothing in the Meta story cancels a single committed bit — the standalone note walks why the unwind is a flows story with a real but bounded fundamental kernel.5 Second, the SK hynix listing Friday cuts both ways: it removes Micron's scarcity premium as the only US-listed memory proxy, but a ~$29B raise earmarked for new fabs and EUV tools is also the supply-side discipline question made flesh — the EUV bottleneck we wrote about last week is about to be tested with fresh capital.5 Third — and this is the one we'd flag even if nothing else had happened — NVIDIA announced compute partnerships built on revenue-sharing and credit support the same Wednesday: financing your customers' purchases of your own product. Vendor financing at the top of a capex cycle is one of the classic late-cycle tells — it was a key feature of telecom in 1999. One announcement is not a verdict, and NVIDIA's balance sheet is not Lucent's. But it goes on the watch-list in bold.4

Simply: Meta signaled it has more AI computing power than it needs and will rent the extra out. That one headline made investors question the whole AI supply chain — if the big buyers overbought, the sellers' future orders look shakier — and the selling hit memory-chip stocks hardest because they'd risen the most. The chipmakers' signed contracts and rising prices are all still in place; what broke was the crowded positioning, not the business. Separately, Nvidia started helping customers finance purchases of its own chips — a pattern that marked the top of past tech cycles — so we're watching that closely.

4. The tape: a record Dow, a rotation, and a steepener

Step back from the AI complex and the week was strong — deceptively so. The S&P rose 1.8% to 7,483, the Nasdaq 2.1%, and the Dow 2.0% to a record close of 52,900 — powered not by tech but by the rotation out of it: on jobs day, utilities, staples, materials, and financials led while tech and consumer discretionary were the only sectors down. Monday's open set the tone — Alphabet joined the Dow, the Supreme Court left Fed governor Lisa Cook in her seat, and the U.S. and Iran announced a deal to halt attacks — and the market absorbed what JPMorgan estimated at ~$165B of quarter-end pension rebalancing out of equities without blinking.2 The bookends printed at the half: the S&P finished H1 up 9.6% (Q2 alone +14.9%, the best quarter since 2020), the Nasdaq up 12.8%, and the small-cap Russell — quietly — up 21.9%, the best of the majors.3

Underneath, the rates story ran opposite to the equity calm, and we covered the shape in Section 1: long end up 10–11bp with the 30-year knocking on 5%, front end up just 7 and rallying on jobs day. Two more pieces complete the macro picture. Oil kept validating the peace — WTI settled at $68.69, below the pre-war range we've been anchoring to, with OPEC+ adding its fourth straight quota hike and Hormuz transits recovering.7 And gold told you what the soft jobs print really meant: after its worst quarter in 13 years — a casualty of the Warsh hawkish repricing — it ripped back above $4,100 within an hour of the payrolls miss. The dollar had its biggest weekly drop in about three months. The market did not read Thursday as “economy fine, hike delayed.” It read it as “the Fed's hike case just got complicated.”7

Simply: The broad market had a great week — the Dow hit an all-time high — but the winners were boring sectors like utilities and banks, not tech. The first half closes with big gains everywhere. Under the surface: long-term interest rates crept up all week (the uncomfortable kind of rise, driven by supply and inflation worry rather than growth), oil stayed cheap as the Iran peace holds, and gold surged after the weak jobs number — a sign investors are hedging, not celebrating.

5. Quantum check-in: the froth is leaking, not breaking

Two weeks into our watch-item, the tape is doing roughly what our trip-wire's downside branch sketched — slowly. The pure-plays popped Monday on South Korea's ~$129B national R&D plan naming quantum a strategic field, then gave it all back over three straight down days: IONQ −0.4% on the week, RGTI −2.3%, QBTS −1.0%, QUBT −1.4%, and the QTUM basket −0.6%. That caps a June in which the pure-plays fell 19–26% while Quantinuum — the profitless-but-institutional newcomer that listed June 4 — rose ~35% from its debut: the speculative bid is rotating within quantum, from the retail names to the credentialed one.8

The week's two SEC filings are the sector in miniature. D-Wave won an NSF grant — real, governmental, and $1.57M, a rounding error against a ~$7B market cap. And Quantum Computing Inc. nearly doubled its authorized share count, 260M to 460M — the dilution machinery we flagged as the sector's funding model, now in an 8-K.8 Nothing here changes the sleeve: venture-sized, basket-first, no adds on momentum. The next real catalyst is dated: the QC-ADDS executive order's 90-day technical specifications land around September 20, when we find out what the government actually intends to buy.

Simply: Quantum stocks drifted lower again — the hype is deflating gradually, not popping. One company got a government grant worth a tiny fraction of its value; another authorized itself to nearly double its share count (which waters down existing owners). Both tell you what this sector runs on. We keep our small, spread-out bet and add nothing.

6. The week ahead: Warsh's minutes, $119B of duration, and the SK hynix debut

A full five-day week, and three events carry it. Wednesday July 8 at 2 PM ET: the FOMC minutes from June 16–17 — the first minutes of the Warsh Fed. The statement was 130 gutted words; the minutes are where we learn how the room actually argued. What we're reading for: whether a June hike was seriously discussed (nine officials already project one by year-end), how broad the support for the 3.8% dot really was, any detail on Warsh's task forces — especially the balance-sheet one, where an aggressive-QT lean would hit the long end directly — and how the committee squares “prices are too high” with a labor market that has since gone slack.9

Around the minutes, the Treasury sells $119B of exactly the paper the market spent last week selling: $58B of 3-years Tuesday, $39B of 10-years Wednesday, $22B of 30-years Thursday. With the 10-year at 4.49% and the 30-year at 4.97%, the auction tails are the week's cleanest test of whether Trip-wire #2 below fires — term premium showing up as weak demand for duration.9 The data calendar is lighter: ISM Services Monday, jobless claims Thursday, with Fed governor Waller on a policy panel in Rome Monday. Earnings trickle in ahead of next week's flood — Levi's Wednesday night, Pepsi Thursday morning, Delta Friday morning.

Friday is the one we've circled: SK hynix targets its Nasdaq ADR debut (SKHY), raising up to ~$29B — one of the largest memory listings on record, proceeds earmarked for new Korean fab capacity and EUV tools. For Micron it's a double event: the scarcity premium of being the only US-listed memory name ends, and the capital that funds the industry's next capacity round gets priced in public. How SKHY trades — and how memory trades around it — is the single best test of whether last week's unwind was positioning (our read) or the start of the cycle question.5 One more housekeeping item: SpaceX joins the Nasdaq-100 Tuesday under the fast-entry rule — index flows arriving three weeks after our initiation flagged exactly this kind of mechanical bid. And the horizon: June CPI lands Tuesday July 14, the same morning the banks open Q2 earnings season, with the hyperscaler capex prints — the referee of our whole framework — starting July 22.9

Simply: Three things matter this week. Wednesday, we see the meeting notes from the new Fed chair's first meeting — the details of how serious they are about raising rates. All week, the government auctions $119 billion of bonds into a market already nervous about long-term rates. And Friday, Korea's SK hynix — Micron's biggest rival — starts trading on Nasdaq in a ~$29B listing that will show us whether last week's chip selloff was temporary positioning or a deeper worry.

7. The book & trip-wires

Positioning into the week: cash-heavy by design into the late-July capex prints — the deployment plan hasn't changed, and neither has the discipline of not deploying into an unwind. The Micron runner sits behind its pre-committed stop, by rule and not by discretion; the supply-side thesis is intact on demand and humbler on flows. Duration stays low at both ends — the front end can reprice on any hot number under a hike-biased Fed, and the long end is one bad auction from our wire. The quantum sleeve is unchanged. Four things would change it.

Trip-wire #1 · The Warsh minutes read hawkish

Wednesday's minutes show a June hike was seriously argued, or the balance-sheet task force leans toward accelerated QT.

The statement was 130 words; the minutes are the regime's first full text. Evidence the committee was closer to hiking than the market believes — or QT acceleration on the table — repricies the front end and hands the long end a second supply problem. Response: we don't fade incentive-aligned hawkishness (June's lesson, in the frame on the wall); defensive into the July 14 CPI, no duration adds, no dip-buying the rate-sensitive longs.

Trip-wire #2 · The long end breaks — carried, and now warm

The 10-year closes above ~4.50% or the 30-year takes out 5.00% on weak auction demand rather than hot data.

Carried from last week — this is the level we set when we publicly disagreed with Schwab's long-end caution — and upgraded from watch to warm: the 10-year tagged the level intraday Wednesday and closed the week at ~4.49%, with $119B of supply landing Tuesday–Thursday. A close through the wire on tailing auctions means the term-premium channel won and our “calmer long end” call is dead by its own rule. Response: cut overall duration to below-benchmark — the response we pre-committed when we named the disagreement with Schwab — and trim the longest-duration equity exposure.

Trip-wire #3 · The SK hynix debut and the memory tape

SKHY lists Friday July 10. Either memory stabilizes through the listing with 3Q contract pricing intact — or the unwind finds a second leg.

If the listing clears, the flow overhang lifts, and Micron holds its line with contract prices still rising, last week reads as positioning — the runner survives and the thesis gets its first clean test back. If the tape breaks through the pre-committed stop instead, we're out by rule — no averaging down, no discretion — and the full accounting is already written in the standalone note. Response: the stop decides, not us.

Trip-wire #4 · Monetization spreads — the spender/supplier line blurs again

A second hyperscaler moves to resell compute, or NVIDIA-style vendor financing expands beyond this week's announcements.

Meta reselling compute turned a spender into a supplier; NVIDIA financing its customers' purchases is the classic late-cycle structure. One of each is a data point; two of either is a pattern. Response: re-underwrite our premise that supplier order books — hyperscaler contracts with limited cancellation risk — anchor the trade, before the late-July capex prints (Alphabet July 22, Microsoft and Meta July 29, Amazon July 30) rather than after — and treat any new vendor-financing disclosure as a cycle-quality downgrade, not a growth headline.

Simply: We're keeping plenty of cash, holding our remaining Micron behind a firm pre-set exit price, staying light on bonds at both ends, and leaving the quantum bet tiny. The four things that would change our plan: the Fed's meeting notes revealing they nearly raised rates in June; long-term rates breaking above the line we've drawn; how the chip market behaves around SK hynix's Friday debut; and any sign that more tech giants are renting out spare computing power or financing their own customers.
Get the next note by email →   All writing

Sources & footnotes

  1. June 2026 Employment Situation (BLS, released Thursday July 2, 2026, 8:30 AM ET): nonfarm payrolls +57,000 — smallest gain in four months — vs. a consensus of ~115K (Dow Jones, per CNBC) / ~113K (Bloomberg, per Yahoo Finance). Unemployment rate 4.2% (from 4.3%, where it had held four months; lowest in a year), driven by labor-force exit: participation −0.3pp to 61.5% (lowest since March 2021), household-survey employment −~507K. Average hourly earnings +0.3% m/m, +3.5% y/y, both in line. Revisions: April 179K→148K (−31K), May 172K→129K (−43K), combined −74K. Sector detail: professional/business services +36K, social assistance +25K, healthcare +22K, leisure & hospitality −61K (seasonal giveback), government +8K. “Slack water / unmoving tide”: Laura Ullrich, Indeed Hiring Lab. Sources: BLS Employment Situation; CNBC; Yahoo Finance; Indeed Hiring Lab (July 2, 2026). ↩
  2. July 2, 2026 session and week (4-day week; markets closed Friday July 3 for observed Independence Day — equities traded a full session Thursday on light volume, bond market closed 2 PM ET per SIFMA): S&P 500 7,483.24 (~flat Thursday; +1.76% on the week), Nasdaq Composite 25,832.67 (−0.80% Thursday; +2.11% week), Dow 52,900.07 (+594.83 / +1.14% Thursday — record close; +1.97% week), Russell 2000 2,996.11 (−0.5% week, only major index down). Thursday sector rotation: utilities +2.21%, staples +2.03%, materials +1.94%, financials +1.53%; tech and consumer discretionary the only decliners. Monday drivers: Alphabet added to the Dow; Supreme Court declined to remove Fed governor Lisa Cook; U.S.–Iran deal to halt attacks. Quarter-end: JPMorgan-estimated ~$165B of global pension/sovereign rebalancing out of equities into bonds, concentrated June 29–30, absorbed. Tesla −7.49% Thursday to $393.45 despite record Q2 deliveries (480,126). Sources: CNBC, AP, TheStreet market wraps; Cboe; JPMorgan estimate as reported. ↩
  3. Quarter- and half-end statistics (June 30, 2026): Q2 2026 S&P 500 +14.87%, Nasdaq +21.41% — best quarter since 2020 for both; H1 2026: S&P +9.55%, Nasdaq +12.79%, Dow +8.85%, Russell 2000 +21.86%. Philadelphia Semiconductor Index (SOX): ~+88% in Q2 (best quarter on record), ~+101% H1. H1 S&P 500 leaders: SanDisk +858% (#1), Micron +300%+, Intel +278%. June was the S&P's first losing month since March. Sources: index closes corroborated against FRED; CNBC quarter-end coverage. ↩
  4. The “Meta shock” and the AI-complex week: Meta reported to be building a cloud business to sell/lease excess AI compute (Wednesday July 1) — semis routed same day (Micron −10.6%, SanDisk ~−11%, SOX two-day collapse ~11%); Kospi −7.89% Thursday July 2 to 7,648.09, SK hynix −14.57%, Samsung Electronics −9.06% (“Meta shock,” Seoul coverage). Apple reportedly exploring sourcing memory from CXMT and YMTC for devices sold in China (AAPL rallied to $308.63). Megacap AI spenders (MSFT, GOOGL, META, AMZN, AAPL) all higher on the week; NVDA −2.6% over the two rout days. NVIDIA announced compute partnerships with revenue-sharing and credit-support structures July 1 (Sharon AI up to 40,000 GB300s; Firmus up to 170,000 GPUs / 360MW; NVIDIA newsroom). Michael Burry disclosed shorts on NVDA, TSLA, SOXX, AMAT, CAT (June 30) and added an MU short (July 2). Sources: press coverage of the Meta report; NVIDIA blog (July 1, 2026); Seoul Economic Daily and Korean market coverage; verified daily closes. ↩
  5. Memory week (closes verified to the cent): MU $975.56 July 2 (−13.85% on the week; −19.6% below the June 25 record close of $1,213.56; ~7% below the pre-earnings close); SanDisk $1,745.00 (−16.5% week; −25.3% from its $2,335.00 June 25 record); Western Digital $539.00 (−8.1%); SOX −4.37% on the week (13,203.57→12,626.22). TrendForce 3Q26 update (July 3), as carried in trade-press summaries: contract prices still rising — conventional DRAM ~+13–18% QoQ, NAND ~+10–15% QoQ (ranges per the trade-press carries; we have not pulled the underlying report). SK hynix ADR listing (Nasdaq: SKHY) targeting Friday July 10, raising up to ~$29B (₩45.45T board ceiling), proceeds earmarked for Yongin Cluster Phase 1, Cheongju P&T7, and EUV tools. Single-name analysis, the base-case re-rate, and the pre-committed stop: see the July 2 unwind note. Sources: verified exchange closes; TrendForce; listing coverage. ↩
  6. Rates week (daily closes, CNBC/H.15-corroborated): 2-year 4.07% (6/26)→4.10→4.14→4.17→4.14% (7/2, CNBC 4.137%) — +7bp on the week, rallying ~2bp on jobs day; 10-year 4.38%→4.38→4.44 (JOLTS)→4.48 (Warsh Sintra)→~4.485–4.49% (+10–11bp; intraday tag of ~4.50 Wednesday; no close above 4.50%); 30-year 4.87%→4.97% (intraday 4.973%). Curve: 2s10s ~4bp steeper. Fed pricing post-jobs: July 28–29 hike odds ~19.8% (from ~28.9% Wednesday); September hike taken off the table in coverage; October a “potential” hike, not fully priced (CME FedWatch as carried: ~41.8% odds of one 25bp hike in 2026, ~21.7% no change — single-feed figures, treat as soft). Week's other data: JOLTS 7.59M openings vs. 7.3M consensus (the hawkish surprise); ADP +98K vs. ~110K; ISM Manufacturing 53.3 (prices paid 73.0, −9.1pts); initial claims 215K; May factory orders −1.3% (ex-transportation +1.9%). Sources: CNBC rates coverage; FRED H.15-corroborated closes; BLS JOLTS; ADP; ISM; Census. ↩
  7. Warsh, geopolitics, commodities: Fed Chair Kevin Warsh at the ECB's Sintra forum, Wednesday July 1 (panel with Lagarde and Macklem): “we've all looked around, and we've seen that prices are too high”; declined to signal the July 29 decision; on political pressure: “We've been an independent central bank for a very long time… you're going to see no changes on that.” Treasury yields rose on the remarks. U.S.–Iran: deal to halt attacks announced Monday June 29; indirect talks in Doha through the week. WTI: $69.23 (6/26, first close below $70 since Feb 27)→$68.69 (7/2), −0.8% on the week (Wednesday low $68.58); OPEC+ fourth consecutive quota increase (+188K bpd for July). Gold: Q2 2026 −~14%, worst quarter in 13 years, then +2% on the week — back above $4,100/oz within an hour of the payrolls miss ($4,170 spot Friday, highest since June 23). Dollar (DXY): −0.58% on the week (~101), biggest weekly drop in ~3 months; euro $1.1442, two-week high. Sources: CNBC Sintra and metals coverage; EIA; OPEC; FX coverage. ↩
  8. Quantum week (June 29–July 2, closes verified): IONQ $49.12 (−0.39% week), RGTI $17.94 (−2.29%), QBTS $22.53 (−1.01%), QUBT $9.05 (−1.42%); QTUM ETF $155.01 (−0.62%). Monday pop on South Korea's Sixth Basic Plan (>₩200T / ~$129B national R&D through 2030, quantum a named strategic field) fully unwound over three down days. Filings: D-Wave 8-K (June 30) — selected for a $1,566,250 NSF National Quantum Virtual Laboratory grant; Quantum Computing Inc. 8-K (June 30) — authorized shares raised 260M→460M effective June 29; no new equity offerings filed by the four pure-plays in-week (SEC EDGAR). June calendar month: pure-plays −19% to −26% (IONQ −26.1%, RGTI −24.4%, QBTS −20.4%, QUBT −18.9%) vs. Quantinuum (QNT, listed June 4) +35% from its debut close. Next dated catalyst: QC-ADDS 90-day technical specifications, ~September 20, 2026 (DOE launched the program June 23). Sources: SEC EDGAR 8-Ks; verified closes; DOE; Korean government plan coverage. ↩
  9. Week of July 6–10, 2026: FOMC minutes (June 16–17 meeting) Wednesday July 8, 2:00 PM ET (Federal Reserve calendar) — the first minutes of the Warsh Fed; June meeting context: rate held at 3.50–3.75%, 2026 median dot 3.8% (implying a hike), nine officials projecting at least one 2026 hike, Warsh declined to submit a dot, statement cut to ~130 words. Treasury auctions: $58B 3-year (Tue July 7), $39B 10-year reopening (Wed July 8), $22B 30-year reopening (Thu July 9) — all settle July 15 (TreasuryDirect). Data: ISM Services Monday 10 AM (May: 54.5); May trade balance Tuesday; wholesale inventories and consumer credit Wednesday; initial claims Thursday. Fed speakers: Gov. Waller, policy panel, ECB research conference, Rome, Monday (Fed Board calendar). Earnings: Levi Strauss Wed AMC; PepsiCo Thu BMO (consensus ~$2.19); Delta Fri BMO (revenue consensus ~$17.5–17.7B by provider). SK hynix ADR (SKHY) Nasdaq debut targeted Friday July 10. SpaceX (SPCX) joins the Nasdaq-100 effective Tuesday July 7 (15-day fast-entry rule) — see our initiation. OPEC+ seven-country producer group (Saudi Arabia, Russia, Iraq, Kuwait, Kazakhstan, Algeria, Oman) met Sunday July 5 on August quotas. Horizon: June CPI Tuesday July 14, 8:30 AM ET — the same morning big banks open Q2 earnings; hyperscaler capex prints from July 22 (GOOGL), July 29 (MSFT, META), July 30 (AMZN); the 10% global Section 122 tariff expires July 24 absent Congressional extension. Sources: federalreserve.gov; TreasuryDirect; BLS; Census; company notices; Nasdaq; OPEC. ↩

Nothing on this page is investment advice. We work in scenarios and trip-wires, not price targets — everything above is for thought and process, not for trading. Forward-looking statements are scenarios, not promises. See disclaimer.