Brandon Leon

Week ahead · Macro · semis · jobs · quantum · June 29, 2026

Micron locked in $100 billion. The AI spenders got a cost problem.

Author Brandon Leon Posted 2026-06-29 (week of June 29; jobs Thursday, holiday Friday) Coverage Week ahead · Micron · the AI bifurcation · a first look at quantum

TL;DR: Micron's June 24 print was a blowout — a record $41.5B, an 84.9% gross margin, and a next-quarter guide of $50B that ran ~$7B past the Street's ~$43B consensus. The structural piece is what came with it: $100B of contracted, prepaid customer demand that lowers the cyclicality of the whole story. MU ripped 16% to a new high. Meanwhile the AI spenders — the megacap hyperscalers — sagged on cost worries, and the S&P spent June drifting lower. That split — the AI suppliers with signed, prepaid demand against the spenders still proving the payback — is the trade. Into a holiday-shortened week: quarter-end Tuesday, the June jobs report Thursday (the first labor read under a hawkish Warsh Fed), markets closed Friday. And a new theme we're starting to watch: quantum.
Where we are

The AI-infrastructure trade is bifurcating, and Micron drew the line. On June 24 it reported a record $41.5B quarter (gross margin 84.9%, EPS $25.11), then guided the September quarter to $50.0B — roughly $7B above the Street's ~$43B consensus — and disclosed 16 strategic customer agreements worth about $100B in minimum committed revenue, with ~$22B of cash already on deposit.1 The stock ripped ~16% to a new closing high near $1,213. That is not how a cyclical stock peaks; that is contracted, prepaid demand.

The same week, the AI spenders wobbled. Mega-cap tech sagged on worries about spiraling AI capital costs, money rotated out of the crowded megacaps, and the S&P 500 spent June down roughly 3%.2 So the market is starting to separate the AI suppliers — memory, the picks-and-shovels with order books — from the AI spenders who have to justify the bill. We have been long the supply side; this is the week the split became visible.

Underneath, the rates picture eased. May PCE landed roughly in line (headline 4.1%, the hottest since 2023, but core a steady +0.3% m/m), oil kept falling, and the 10-year eased to around 4.37%, below 4.40% for the first time in over a month — even with the Warsh Fed, whose June 17 debut flipped the dots to a 2026 hike, now leaning that way.3 The long end relaxed; the front end stayed on guard. That tension goes to the week's main event.

The week ahead is short and loaded. Tuesday is quarter- and half-end. Thursday July 2 brings the June jobs report — pulled forward a day because markets are closed Friday for Independence Day — and it is the first labor read under the new hawkish regime. A hot number is the one that pulls Warsh's hike forward. We frame it in Section 4. And in Section 5, a new idea: we begin looking at quantum.

1. Micron: the print that re-rated the cycle

We have written more about Micron than anything else this year — most recently the print-eve note that called the quarter priced and braced for a fade — so we will be precise about what changed. The headline beat was enormous — revenue $41.5B against a $33.5B guide and a ~$35B Street, gross margin 84.9% (a record), EPS $25.11 against ~$20.28 expected.1 Data-center revenue topped $25B in the quarter, a $100B+ annualized run-rate; HBM4 is ramping roughly twice as fast as HBM3E did, with over $1B already shipped. But the in-quarter beat was the consensus expectation. Two things actually moved the thesis.

First, the FQ4 guide: $50.0B ±$1.0B at ~86% gross margin and $31.00 EPS. Wall Street's consensus for the quarter sat near $43B in revenue and ~$25 in EPS; Micron's guide cleared both — about seven billion dollars of revenue upside, and a bottom line roughly a quarter above the bar. That is not momentum — it is a step-change in the earnings trajectory.

Second, and more important, the contracts. Sixteen strategic customer agreements now lock in roughly $100B of minimum committed revenue, covering about 20% of DRAM and a third of NAND volume, with ~$22B of customer cash already on deposit. And the structure matters: management described a collar — price ceilings near current levels, but a floor whose margins sit above the company's prior-cycle peak. Read that again. Even in a downturn, the contracted ~40% of revenue earns better than Micron's best-ever margins. That is the single most important sentence of the quarter, because it is what changes the multiple: a memory company with a contractual floor above prior peaks is structurally less cyclical than the company the market has always discounted.4

Chart 1 — Micron FQ3 actual, and the FQ4 guide vs. the Street's bar

A $41.5B quarter, then a $50B guide — ~$7B past the Street's ~$43B consensus.

FQ3 revenue ($B) FQ4 guide vs. Street bar $33.5 guide $35 Street $41.5 ACTUAL ~$43 Street bar $50.0 FQ4 GUIDE ~+$7B

Micron FQ3'26 (reported June 24): revenue $41.5B vs. the $33.5B company guide and a ~$35B Street estimate; gross margin 84.9%, EPS $25.11. FQ4 guide of $50.0B ±$1.0B sits roughly $7B above the ~$43B Street consensus. Sources: MU FQ3'26 press release / 8-K; CNBC; StockTitan; 24/7 Wall St.

Forward, the discipline is to not buy our own euphoria. Management was explicit that margins are near a cap (~86%) and that growth from here is volume, not price; the same contracts that put a floor under the downside put a ceiling on the upside torque. So the re-rate is real but bounded. Our through-cycle base case moved meaningfully higher off the blowout — the specific target stays in the standalone Micron note, where single-name numbers belong — but the uncapped memory torque, the part that reprices fastest if 2027 tightens further, now lives in the peers without price ceilings, not in chasing Micron at a new high. The cycle-end frame moved out, too: management sees no line of sight on supply catching demand, with only gradual improvement in 2028.4

There is a supply-side reason to think that floor holds. Memory capacity can only grow as fast as the industry can install leading-edge EUV lithography and the advanced 3D-stacking and packaging that high-bandwidth memory demands — and the lead times on both run for years, with the toolmakers' order books for the newest equipment already stretching toward the end of the decade. That is the mirror image of Micron's contracts: signed demand underneath, an equipment bottleneck capping how fast supply can answer it. It is the cleanest case that this cycle troughs higher than the last — and it is the same conclusion an independent outside read of the quarter reached this week, arriving from the supply side rather than the order book.8

We owe an honest grade. Going in — in last week's print-eve note — our modal read was a “solid-but-priced print that fades,” the Broadcom template where a beat into a stretched stock sells off. We came in deliberately cautious: roughly 60% of the position already booked into the run-up behind a hard stop, and a conservative through-cycle line held while the sell-side raced its targets higher. We were wrong on direction — the beat overwhelmed the in-price bar and the stock ripped instead of fading. But the framework did its job: the remaining ~40% caught the full +16%. Trimming into strength on the way up and keeping a runner behind a stop is exactly the design that lets you be wrong on the reaction and still right on the book. And the update is the honest part: the contractual floor is a new fact, so we are revising that conservative line up — in the open, the way we flagged the fade as a risk. We would rather be on the record and occasionally wrong than quietly right.

Simply: Micron crushed it — record sales, record margins, and a next-quarter forecast far above expectations. The big deal is $100B of locked-in, pre-paid contracts that make its business far less boom-and-bust than memory chips usually are. We expected the stock to sell the news; it soared instead. We got the reaction wrong but kept the right amount of stock to benefit. From here we don't chase it — the contracts cap how high it can run.

2. The split: suppliers vs. spenders

Zoom out and the week tells a cleaner story than any single name. The AI trade has run for two years as one undifferentiated thing — own the theme, own everything. That is ending. The market is now sorting AI into two books.

On one side, the suppliers: the memory makers, the foundries, the networking and power names that sell into the buildout and increasingly carry contracted order books. Micron's $100B of strategic agreements is the purest example — revenue that is signed, prepaid, and visible. On the other side, the spenders: the hyperscalers writing the checks, who now face the harder question of when $700B+ of annual capex turns into returns. That is the “AI cost” worry that pressured megacap tech this month, and it is why the S&P could be down ~3% on the month while Micron printed an all-time high.2 Schwab's mid-year outlook, out the same week, made the institutional version of the point: index earnings growth has become “increasingly dependent on a relatively small set of companies tied to AI capital spending,” and it calls that concentration the market's central risk.9 This is not a call that the spenders are in trouble — their capex is the suppliers' revenue, and a genuine cut would hit both. It is a call about where the visibility is.

When the cycle gets questioned, contracted supply reprices last and recovers first: it is revenue that is already signed, so it is the last to be renegotiated down and the first to snap back when demand turns. The actionable read follows: own the concentration one layer below the index — stay long the supply side with order-book visibility — and treat the megacap spenders as the part of the AI trade that now has to prove the return, with the late-July hyperscaler capex prints as the referee.

Simply: For two years, “AI stocks” moved together. Now they're splitting. The companies that sell the AI gear (like Micron) have signed contracts and clear demand. The giant tech companies buying the gear face questions about whether their massive spending will pay off. We'd rather own the sellers with locked-in orders than the buyers who still have to prove it.

3. The macro tape: a hawkish front end, a calmer long end

Two weeks on from Warsh's hawkish debut — the meeting where the dots flipped to an implied 2026 hike and the 2-year jumped 16 basis points, the move that killed the pre-FOMC call we graded last week — the rates market has settled into a split. The front end stayed on guard: the Fed has put a hike on the table and the 2-year is pricing it. But the long end relaxed. May PCE, released June 25, came in roughly as expected — headline 4.1% (the hottest since April 2023, still an energy story) but core a steady +0.3% on the month — and with oil continuing to fall, the 10-year eased to around 4.37% by week's end, below 4.40% for the first time in over a month.3

That shape — tight policy now, contained inflation expectations later — is a market that believes the Fed will act but does not believe inflation is spiraling. It is also why a hot jobs print Thursday matters so much: it is the one data point that could force the front end to pull the hike forward and drag the long end back up with it.

It is worth holding our own view against an outside one. Schwab's mid-year outlook, published the same week, watches the same band — it sees the 10-year holding roughly 4.0–4.5%, where it has traded since March — but it is more defensive on the long end than we are, citing term premium, fiscal supply, and oil as upside risks and favoring below-benchmark bond duration as a result. We read the long end as the calmer part of the curve right now; Schwab reads it as the more dangerous one. That disagreement is the tell: if our “calmer long end” call is wrong, the term-premium channel is exactly how it breaks — so we name it as Trip-wire #4 below. The part of Schwab's read we do share is the economy underneath: a roughly 3% second quarter resting on a consumer increasingly strained by soft real wages and higher energy — solid on the surface, fragile below it.9

Simply: The bond market thinks the Fed will keep rates high or even hike, but it doesn't think inflation is running away — so longer-term yields actually fell this week. The thing that could change that is a strong jobs report on Thursday. One caveat: some big institutional desks (Schwab among them) think those longer-term yields are more likely to head back up than we do — if they're right, it's the government-borrowing-and-term-premium story, and we're watching for it.

4. The week ahead: quarter-end, then the jobs print under a hawkish Fed

A four-day week with one number that matters. Monday and Tuesday close the quarter and the first half — expect month- and quarter-end rebalancing flows, which can exaggerate moves without meaning much. Then Thursday July 2 at 8:30 ET, the June Employment Situation — moved up from its usual Friday slot because U.S. markets are closed Friday July 3 for the Independence Day holiday.5 Thursday is also an early close on Wall Street ahead of the long weekend, so the report lands into thin, holiday liquidity — moves can overshoot.

This is the first labor read under the hawkish Warsh regime, and the framework is simple. The Fed has signaled it would rather hike than cut; the swing variable is whether the labor market gives it the excuse. A hot print — payrolls comfortably above 150K or average hourly earnings re-accelerating — pulls the hike forward in pricing, lifts the front end, and pressures the rate-sensitive parts of the tape. A soft print buys time and lets the “hawkish but patient” read hold. Our base case is a number in the middle that keeps the hike a December question, not a summer one — but in holiday liquidity, the tails are wider than usual.

Simply: Short week. The June jobs report comes Thursday instead of Friday (markets are closed Friday for July 4th). With the Fed now leaning toward raising rates, a strong jobs number would push markets to expect that hike sooner. A weak one buys time. Expect bigger-than-usual swings because trading is thin around the holiday.

5. New idea: we start looking at quantum

A standing feature from here: when a theme starts earning a place on the research bench, we say so early and in the open — thesis-in-formation, not a recommendation. This week's is quantum computing, and the trigger is policy.

On June 22, the administration signed two executive orders aimed squarely at the sector: one standing up a national initiative (QC-ADDS) to coordinate the Energy, Commerce, Defense, and intelligence agencies toward delivering a science-grade quantum computer to a DOE facility by 2028; the other a cryptographic mandate ordering federal agencies onto post-quantum cryptography by 2030–2031.6 That follows roughly $2B of CHIPS-Act quantum funding announced in May. The names responded violently: the pure-plays — IonQ (IONQ), Rigetti (RGTI), D-Wave (QBTS), Quantum Computing Inc. (QUBT) — are up 50%+ since late March, with IonQ posting 755% year-over-year revenue growth, D-Wave reporting bookings — new orders, not revenue — up nearly 2,000% even as its reported sales fell on a tough prior-year comp, and Rigetti signing a Commerce Department letter of intent for up to $100M.7

Now the discipline. Those growth rates are real and they are off a tiny base — these companies have minimal revenue, no profits, and valuations that price a commercial future that is years — possibly a decade or more — away. This is the opposite of the Micron setup: there are no contracted order books here, only option value on a frontier. So we treat it as exactly that — a speculative optionality sleeve, not a core position, sized like a venture bet, not a conviction holding. The way to play a policy-driven, pre-revenue theme without single-name blow-up risk is the basket: the quantum ETFs (QTUM, QTUP) spread the bet across the field and the larger-cap incumbents (the cloud and chip names with real quantum programs) so a single failed roadmap doesn't end the thesis. The overlap with the megacaps we are cautious on in Section 2 is deliberate, not a contradiction: there we won't pay up for hyperscaler capex on an uncertain payback; here those same balance sheets are ballast, the thing that keeps a basket of pre-revenue bets from living or dying on one company's roadmap.

What we are watching to decide whether this graduates from a watch-item to a position: (1) a credible path to useful error-corrected qubits, not just qubit counts; (2) commercial revenue that is recurring, not grant- and government-LOI-driven; (3) one of the large incumbents putting a quantum advantage into a shipping product. Until those, it is a frontier we respect and a position we keep small. We will build the thesis out in a standalone piece; consider this us planting the flag.

New idea · Quantum — thesis in formation

A policy-and-funding tailwind (June 22 executive orders, $2B CHIPS) is re-rating a pre-revenue sector. We treat it as a small, speculative optionality sleeve — played through the basket, not single names — until commercial revenue and error-correction milestones justify more.

Watch-list: the pure-plays IONQ, RGTI, QBTS, QUBT for the news flow; the ETFs QTUM / QTUP for sized exposure. Graduation triggers: useful error-corrected qubits, recurring commercial revenue, or an incumbent shipping a quantum advantage. Risk: extreme valuations, no profits, multi-year (possibly decade) commercialization — size it like a venture bet.

Simply: Quantum computing stocks are soaring after new government orders and funding. The growth numbers are huge but the companies are tiny, unprofitable, and years from real products. It's an exciting frontier, so we're starting to study it — but only as a small, high-risk bet spread across many names, not a big position in any one. We'll watch for real, paying customers and working error-corrected machines before doing more.

6. The book & trip-wires

Positioning into the short week: long the AI supply side with order-book visibility (memory and the contracted picks-and-shovels); underweight front-end duration under a hawkish Fed; neutral-to-cautious on the megacap AI spenders until the late-July capex prints; a small, basketed quantum sleeve as new optionality. Four things would change it.

Trip-wire #1 · A hot jobs print pulls the hike forward

June payrolls comfortably above 150K or AHE re-accelerating, Thursday July 2.

The first labor read under a Fed that would rather hike. A hot number lifts the front end and pressures rate-sensitive equities into thin holiday liquidity. Response: add to the front-end-duration short; trim the rate-sensitive longs; the supply-side semis with contracted demand are the relative hold.

Trip-wire #2 · Hyperscaler capex confirmation (late July)

MSFT / GOOGL / META / AMZN raise — or guide cautiously on — 2026 CapEx at the late-July prints.

This is the referee for the supplier-vs-spender split. A raise validates the supply-side order books (and Micron's $100B of SCAs); a cautious tone is the first real crack in the demand the suppliers have contracted. Response: a raise confirms the supply-side long — stay the course, no de-risking; a cautious tone is the cue to trim the suppliers into it, because their contracted visibility would then be borrowing against demand the buyers are signaling they will slow.

Trip-wire #3 · Quantum graduates — or the froth breaks

A real error-correction / commercial-revenue milestone (upside) — or a funding-driven blow-off that unwinds (downside).

New watch-item. The sector is policy-fueled and pre-revenue; either a genuine technical/commercial proof point graduates it to a real position, or the speculative move exhausts and the basket gives it back. Response: the sleeve stays venture-sized either way; we add only on a graduation trigger, not on momentum.

Trip-wire #4 · The long end breaks higher

The 10-year yield pushes back above ~4.5% on term premium, fiscal supply, or oil — not on strong growth.

Where we and Schwab disagree. We read the long end as the calmer part of the curve; Schwab is more defensive there. A 10-year that breaks ~4.5% for reasons other than a hot economy says the term-premium channel won, re-prices long-duration assets, and undercuts the “long end relaxed” half of our macro call. Response: if it breaks on supply and term premium rather than data, the front-end-duration short is no longer enough — we cut overall duration toward the below-benchmark stance Schwab is already running.

Simply: We're holding AI “seller” stocks with locked-in orders, leaning against short-term bonds, staying cautious on the big AI spenders until they report spending plans in late July, and keeping a tiny quantum bet. The jobs report Thursday and big-tech earnings in late July are the next two things that could change the plan — along with any sharp jump in long-term interest rates.
Get the next note by email →   All writing

Sources & footnotes

  1. Micron (MU) fiscal Q3 2026, reported June 24, 2026: revenue $41.46B (record; vs. $33.5B company guide and ~$35.25B Street estimate); gross margin 84.9% (record); non-GAAP EPS $25.11 (vs. ~$20.28 estimated). Data-center revenue above $25B in the quarter; HBM4 ramping ~2× the HBM3E pace, >$1B shipped. FQ4'26 guide: revenue $50.0B ±$1.0B (vs. ~$43.2B Street consensus), gross margin ~86%, non-GAAP EPS $31.00 ±$1.00 (vs. ~$25 Street). Stock rose ~15% in extended trading, ~+16% to a new closing high near $1,213.56 (intraday high ~$1,255) by June 25, from a ~$1,049 prior close. Sources: MU FQ3'26 press release / 8-K (SEC EDGAR, CIK 0000723125); CNBC (June 24, 2026); StockTitan; 24/7 Wall St. ↩
  2. Mega-cap technology weakness and rotation in June 2026: the S&P 500 was down roughly 3% on the month with three sessions left, as tech sagged on concerns over rising AI capital costs and investors rotated into non-tech sectors (advancers outnumbering decliners). Sources: T. Rowe Price global markets weekly update; CNBC market coverage (June 2026). ↩
  3. May 2026 PCE (BEA, released June 25, 2026): headline PCE +0.4% M/M, ~4.1% annual rate (highest since April 2023); core PCE +0.3% M/M, unchanged from April. Treasury yields fell across maturities into week's end as oil declined and PCE matched expectations; the 10-year, near 4.40% on June 25, slipped to ~4.37% by June 26 — below 4.40% for the first time in over a month. Sources: BEA Personal Income and Outlays release; CNBC / Trading Economics (Treasury yields, June 25–26, 2026); T. Rowe Price weekly update. ↩
  4. Micron strategic customer agreements and outlook (FQ3'26 call, June 24, 2026): 16 strategic customer agreements covering ~20% of DRAM and ~one-third of NAND volume; ~$100B minimum committed revenue (remaining performance obligations); ~$22B of customer cash deposits (~$18B received). Management characterized the largest agreements as carrying price ceilings near the current (calendar Q2) market price and floor prices through the term, with floor-price profitability described as “higher than peak margins at any time in the past.” Gross margin near a structural cap (~86%); growth from here characterized as volume- rather than price-driven. Supply: no line of sight on when supply catches demand, with only gradual improvement expected in 2028. Sources: MU FQ3'26 earnings call transcript / prepared remarks; verify against the 8-K on SEC EDGAR. ↩
  5. June 2026 Employment Situation (nonfarm payrolls) scheduled for release Thursday, July 2, 2026 at 8:30 a.m. ET — moved up from the usual first-Friday slot because U.S. equity and bond markets are closed Friday, July 3, 2026, for the observed Independence Day holiday (July 4 falls on a Saturday). Source: BLS schedule of releases for the Employment Situation. ↩
  6. Executive orders signed June 22, 2026 targeting quantum: one establishing the Quantum Computer for Application Development and Discovery Science (QC-ADDS) initiative, coordinating the Departments of Energy, Defense, and Commerce and the intelligence community toward delivering a science-enabling quantum computer to a DOE facility by 2028; a second directing federal agencies to transition high-value assets to post-quantum cryptography by 2030 (key establishment) and 2031 (digital signatures). Follows ~$2B in CHIPS-and-Science-Act quantum funding announced in May 2026. Sources: Trefis; U.S. News & World Report; coverage of the June 22, 2026 executive orders. ↩
  7. Quantum pure-play performance (as of late June 2026): IonQ (IONQ) reported 755% year-over-year revenue growth; D-Wave (QBTS) reported quarterly bookings (new orders) of $33.4M, up ~1,994% — though reported Q1'26 revenue was ~$2.9M, down ~81% YoY against a one-time large-system sale a year earlier, a distinction that itself underscores the pre-revenue nature of the group; Rigetti (RGTI) signed a Department of Commerce letter of intent for a proposed award of up to $100M over three years under the CHIPS Act, shares +44.3% over the prior month. The sector is up 50%+ since late March. Figures are off very small revenue bases; the companies are pre-profit with extreme valuations. Sources: Trefis; Motley Fool; Fast Company; U.S. News (June 2026). Re-verify current levels before reuse. ↩
  8. Supply-side framing corroborated by an independent outside analysis this week: Mark Meldrum, “Market Outlook: June 28, 2026” (video commentary). Meldrum argues memory capacity “can only grow as fast as ASML can make EUV equipment,” with a parallel back-end bottleneck in 3D stacking / hybrid bonding for high-bandwidth memory, concluding that “cheap memory is over” and that demand is “structurally higher and more diversified.” We treat his specific equipment figures as estimates and keep our text to the qualitative mechanism; the lead-time-as-supply-gate point is consistent with Micron management's own “no line of sight on supply catching demand” guidance (see note 4). Source: M. Meldrum, YouTube, June 28, 2026. ↩
  9. Charles Schwab, “2026 Mid-Year Market Outlook” (Schwab Center for Financial Research) — Trader Talks webcast dated June 25, 2026 and the accompanying written outlook. Equities/economy (Liz Ann Sonders, Kevin Gordon): “the earnings backdrop is the strongest it has been in years,” but with concentration as the central risk — global earnings growth “increasingly dependent on a relatively small set of companies tied to AI capital spending”; second-quarter GDP tracking ~3%, supported by consumer spending and business investment but pressured by negative real wage growth, weak savings, and higher energy; inflation sticky, Fed expected to stay patient. Fixed income (Kathy Jones / Collin Martin): the 10-year Treasury expected to hold ~4.0–4.5% “with risks to the upside”; favor below-benchmark duration (“now is not the time to favor long-duration investments”); opportunities cited in investment-grade corporates, high yield, and preferreds. Note: the written outlook predates the June 17 FOMC, so on Fed policy specifically we defer to the post-FOMC hawkish read. Sources: schwab.com/learn 2026 mid-year outlook (U.S. stocks & economy; taxable fixed income); Schwab Center for Financial Research press release, June 2026. ↩

Nothing on this page is investment advice. Forward-looking statements are scenarios, not promises. See disclaimer.