Compare campaigns
Compare campaigns
Read the claims and horizons together. Shared instruments alone do not imply a conflict.
← All campaignsAI spending: the supplier earnings reversal case
Counter-case to persistent optical and memory gains: by March 2027, AI spending reductions and weaker supplier orders reverse the pricing or margin gains those campaigns expect.
- Why now
- Financing demands and delivery obligations warrant an explicit counter-case. Current supplier and cloud growth challenge immediate-collapse claims; no broad spending-cut conclusion is established.
- In scope
- Verified spending revisions, cancellations and supplier pricing or margins · Delivery economics and financing as possible transmission channels
- Out of scope
- Inferring spending cuts from stock dispersion or GAAP losses alone · Treating a short-term pullback as the end of long-term AI demand · Issuance calendars without a filed, measurable event
- Expected horizon
- Through March 2027, matching the next two reporting cycles of the optical and memory cases.
- What supports it
- A disclosed spending reduction is followed by corroborating order deterioration and pricing or margin reversals in optical or memory suppliers
- What would break it
- Through March 2027, customers maintain spending and both core supplier groups sustain orders and margins without the proposed reversal
Next review 2026-10-30 · Earlier on a verified capital-spending guide-down or a customer cancellation disclosed in a filing or on an earnings call; earlier also on a new Oracle equity or convertible authorisation, since the $20 billion at-the-market program is exhausted and a replacement filing would confirm the funding gap is structural rather than a timing effect. Otherwise review at the October hyperscaler reporting cycle, beginning with Microsoft's scheduled October 28 report (Microsoft investor-relations calendar, 2:30 PM PT), and test the financing channel — issuance terms, prepayment recurrence and credit marks — separately from the order channel. Aggregator-reported capital-spending figures do not satisfy the disclosure test; a filing or transcript does.
Research ideas
All 0 ideas →Issued calls · returns compare the symbol with SPY after 30 and 90 days. Positive excess supports a BUY; negative excess supports TRIM or SELL. WAIT and research ideas have no score. These are call outcomes, not a campaign return.
A pullback that holds above the 200-day average ($286.14 at the Oct 9 close) while RSI cools below 55, then a settled close above $300.40, the October 6 close on the Google agreement
Invalidation: A settled close below the 200-day average ($286.14 at the Oct 9 close), or a third-quarter report (expected around Nov 9, provider date) that does not confirm the Google agreement's terms
reclaim the 20-day (~$256) with hyperscaler spending language intact on the Jul 29–30 prints
Invalidation: capex language softens and the July shelf breaks — then the contracted-electrons bid is repricing too
held the AI unwind at −5%; reclaim the 20-day (~$1,085) and print a higher low
Invalidation: loses the July low — the complement bid was narrower than the month suggested
first higher low after a clean Jul 29–30 capex read — the −18% month is a beta markdown, and the entry is print-gated, not chart-gated
Invalidation: lower lows after capex language holds firm — then the fuel-cycle derate is regime, not beta
Friday printed a reversal bar (low $804, closed $848.95); a close above that bar's high ($903.93) confirms it — the trend-hold entry itself still needs the base and a 20-day reclaim ($1,023)
Invalidation: a close below Friday's low ($804) voids the reversal; the knife is still falling
the general never broke (−3.7% in 30 days vs the ETF's −15.7%); a reclaim of its 50-day ($209.81) is the complex-repair signal — the leader turns first
Invalidation: a close below the $186.48 volume shelf means the general finally fell and the whole complex read resets bearish
the complex switch: first close back above the 50-day ($597.33) after a higher low, confirmed through the 20-day ($610.93); until then every bounce is a bounce inside a downtrend
Invalidation: a close below Friday's low ($536.81) means the bounce attempt failed and the next leg down is open
the complex's only accumulation-shaped bar: +5.7% on 1.35× volume, closed near the high after touching $700.39; above Friday's high ($808.65) with the July 28 print ahead, the bounce has legs
Invalidation: a close below $700.39 means Friday was a dead-cat, not a turn
reclaim the pre-registered $270 line and hold it — the trigger the thesis flag broke this week, now the recovery tell
Invalidation: lower lows continue below $270 after the prints — then the derate is regime, not scare
gated on the prints, not the chart — a storage/memory print (STX Jul 28, WDC Jul 29, SNDK Aug 5) showing DRAM/NAND pricing intact, plus a first higher low
Invalidation: a print that shows pricing momentum actually rolling — then the derate was early, not wrong
the R1 playbook leg — hold the 20-day retest (~$202) through the July 28–29 storage prints, then a higher low; the efficiency-scare component tends to mean-revert if the fundamentals hold
Invalidation: the retest fails and the pullback becomes a breakdown below the July shelf, or hyperscaler capex language actually softens on the July 29–30 prints
the quality expression of the rotation — already back above all three moving averages with RSI under 60 and half a billion of operating cash flow; buyable at structure, add on strength over the 20-day (~$338 now support)
Invalidation: a finish under the 50-day (~$320) puts it back in the broken cohort; under the one-year VWAP (~$310) the rotation call was early
the rotation's confirming leader — but RSI 78 and 3% off its high is chasing; reopen on the first orderly pullback that holds the 50-day
Invalidation: n/a — not owned; the pass is a timing pass, not a thesis pass
the earnings proof-point of the rotation (beat + raised subscription guide) but still in a downtrend, 43% off its high — watch for a base; this is a lane signal, not yet an entry
Invalidation: the bounce fades below the June low with the lane's 7-day breadth rolling back under 50%
AI memory: pricing discipline through the supply ramp
AI memory demand and disciplined capacity additions will sustain DRAM/HBM and NAND supplier pricing and economics through the next two reporting cycles; track each product separately.
- Why now
- The latest reviewed producer disclosures describe pricing gains and customer agreements, alongside planned capacity expansion. The question is how long pricing survives the supply response.
- In scope
- DRAM/HBM and NAND pricing, producer supply, inventories and customer commitments · Product-specific evidence; one product cannot validate another
- Out of scope
- Nearline HDDs and their density cycle · Treating a transient labor catalyst as structural demand · Unverified contractual pricing floors
- Expected horizon
- 6–18 months; evaluate the common spending counter-case through March 2027.
- What supports it
- Pricing and demand remain firm as planned capacity enters service · Customer inventory and commitment evidence supports actual consumption
- What would break it
- Any two are verified: accelerated greenfield supply, a negative contract-price quarter, customer inventory build, or major HBM order reductions · By March 2027, spending reductions produce broad memory pricing and margin deterioration
Next review 2026-10-30 · Read SK hynix (provider-estimated 2026-10-27) and SanDisk (provider-estimated 2026-10-29) results for product pricing, HBM orders and capacity. Micron's fiscal 2026 10-K (filed 2026-10-09) was read in the 2026-10-16 review: it adds commitment evidence and new risk language on customer inventories built under long-term agreements, and verifies no failure leg. Earlier on a negative contract-price report, a customer order reduction, or an industry greenfield or equipment acceleration (the greenfield clause fired on secondary reports only, Nanya's Fab 5A addition and SK Group's expansion directive of 2026-10-11 and 2026-10-12, and counts once an issuer release or named wire confirms it). Micron's fiscal Q1 2027 margin versus ~86.25% and its fiscal Q2 guide are the named tripwire.
Research ideas
All 0 ideas →Issued calls · returns compare the symbol with SPY after 30 and 90 days. Positive excess supports a BUY; negative excess supports TRIM or SELL. WAIT and research ideas have no score. These are call outcomes, not a campaign return.
reclaims and holds the 20-day ($888.26) with RSI back above 50
Invalidation: RSI falls back under 40 while price stays below the 20-day
price back above the 20-day ($264.25) with the 30-day loss narrowing
Invalidation: RSI below 40
no entry while price sits 13.0% below the 20-day ($1,393.06) and 28.2% below the 50-day
Invalidation: a base built back above the 20-day on rising volume forces a re-look
holds the 20-day ($834.21) and turns RSI back above 50
Invalidation: either drive maker signals added unit or greenfield capacity, or a close below the 200-day ($533.00)
not before the Aug 5 call's allocation and gross-margin language is public and read; then a reclaim of the 20-day ($511.47)
Invalidation: any unit- or greenfield-capacity addition signalled by either drive maker — the supply-discipline pillar fails
reclaim the 20-day (~$256) with hyperscaler spending language intact on the Jul 29–30 prints
Invalidation: capex language softens and the July shelf breaks — then the contracted-electrons bid is repricing too
held the AI unwind at −5%; reclaim the 20-day (~$1,085) and print a higher low
Invalidation: loses the July low — the complement bid was narrower than the month suggested
first higher low after a clean Jul 29–30 capex read — the −18% month is a beta markdown, and the entry is print-gated, not chart-gated
Invalidation: lower lows after capex language holds firm — then the fuel-cycle derate is regime, not beta
Friday printed a reversal bar (low $804, closed $848.95); a close above that bar's high ($903.93) confirms it — the trend-hold entry itself still needs the base and a 20-day reclaim ($1,023)
Invalidation: a close below Friday's low ($804) voids the reversal; the knife is still falling
the general never broke (−3.7% in 30 days vs the ETF's −15.7%); a reclaim of its 50-day ($209.81) is the complex-repair signal — the leader turns first
Invalidation: a close below the $186.48 volume shelf means the general finally fell and the whole complex read resets bearish
the complex switch: first close back above the 50-day ($597.33) after a higher low, confirmed through the 20-day ($610.93); until then every bounce is a bounce inside a downtrend
Invalidation: a close below Friday's low ($536.81) means the bounce attempt failed and the next leg down is open
the complex's only accumulation-shaped bar: +5.7% on 1.35× volume, closed near the high after touching $700.39; above Friday's high ($808.65) with the July 28 print ahead, the bounce has legs
Invalidation: a close below $700.39 means Friday was a dead-cat, not a turn
reclaim the pre-registered $270 line and hold it — the trigger the thesis flag broke this week, now the recovery tell
Invalidation: lower lows continue below $270 after the prints — then the derate is regime, not scare
gated on the prints, not the chart — a storage/memory print (STX Jul 28, WDC Jul 29, SNDK Aug 5) showing DRAM/NAND pricing intact, plus a first higher low
Invalidation: a print that shows pricing momentum actually rolling — then the derate was early, not wrong
the R1 playbook leg — hold the 20-day retest (~$202) through the July 28–29 storage prints, then a higher low; the efficiency-scare component tends to mean-revert if the fundamentals hold
Invalidation: the retest fails and the pullback becomes a breakdown below the July shelf, or hyperscaler capex language actually softens on the July 29–30 prints
first clean post-debut base — not the open (debut ~2026-07-10)
Invalidation: DRAM ASP momentum turns negative, HBM4E timeline slips, or the debut air-pocket breaks down without basing
own the up-cycle with cycle sizing — record DRAM/HBM pricing power, HBM booked through 2027; size so a 20-30pt margin reversion is survivable
Invalidation: DRAM ASP momentum turns negative — reduce first, ask questions later; also watch the SKHY ADR closing the proxy-premium gap
only on evidence the +69% NAND operating margin is a baseline, not the peak — two more prints holding pricing
Invalidation: NAND ASP momentum rolls — the −111%→+69% swing is a cycle-top signature, not a new run-rate
own the cleanest un-hyped inflection — Seagate's company-wide GAAP operating margin compounding 23%→32% over four consecutive quarters (fiscal Q4 2025 to Q3 2026) on AI nearline-storage demand; Seagate reports no separate nearline margin
Invalidation: Seagate's quarterly GAAP operating margin (operating income ÷ revenue, as filed) fails to rise for two consecutive quarters, or DRAM ASP momentum turns negative (the cycle tell)
same nearline thesis as STX — margins 10%→36% on the same AI storage driver, no hype premium
Invalidation: nearline margin progression stalls or reverses for two consecutive quarters, or DRAM ASP momentum turns negative (the cycle tell)
Shared evidence
9 sources linked to both campaigns. Each interpretation belongs to its own claim.