Stock-market dispatch — 2026-07-31
Headless run (Fri). Research output, not advice — brain never trades. Marks are live 07-31 intraday
twelvedataquotes; 47/47 symbols priced, 0 429s. Weekly ledger (4d) done on live-digest basis (hypothetical re-mark deferred, not fabricated). Supabase upsert + push are performed by the wrapper, not this run.
Top of mind
The semis rout reversed hard, and it reversed exactly where the FCF-ROI refinement said it should — on the asset-heavy consumption leg. After yesterday's flush (MU −9.9%), today was a violent snapback: MU +18.4% ($874.66), TSM +7.6%, INTC +11.3%, CBRS +19.9%, ARM +7.4%, DELL +9.5%. The catalyst is the load-bearing refinement of ai-roi-reckoning: Amazon is FCF-negative (−$7.6B TTM; capex $169B > OCF $161B) and spending ≥$200B, yet rose ~8–9% because AWS reaccelerated to +37% (op income +64%, ~39% margin) (2026-07-31-autoresearch-amzn-aapl-q2-2026-capex-reaction-roi-discriminator). So the 07-30 "punish the FCF-negative raiser" binary is too coarse — the real discriminator is visible ROI on the capex, not the sign of FCF (META was punished for a return-less raise; AMZN was forgiven because the return showed in the cloud line; Apple fell >6% but on guidance/supply, not capex — rule doesn't apply). AWS reaccelerating is the demand that fills the HBM order book — direct confirmation of hbm-cowos-as-binding-bottleneck, which is why MU ripped. 1 calibration logged (proxy-mistaken-for-cause: FCF-sign was a proxy; ROI-visibility is the cause). Separately, PWR +17.3% on a Q2 blowout (below).
Strongest-conviction buys
Ranked shortlist (4c). Prices 07-31 intraday, twelvedata.
| # | Ticker | Causal chain (mechanism) | Conviction | Valuation vs base | Fundamentals (one line) | Next catalyst |
|---|---|---|---|---|---|---|
| 1 | MU | hbm-cowos-as-binding-bottleneck | High (0.80) | Yes — $874.66 (+18.4%, −30% off hi) | ~6× fwd, HBM booked thru CY2027; AWS +37% reaccel = visible demand confirming the consumption leg the FCF-discrimination rewards | HBM4 ramp / SK Hynix reads |
| 2 | PWR | pwr-transformer-moat-to-eps-doubling | Med-high (0.73, ↑ from med) | Marginal — $657.98 (+17.3%, −17% off hi) | Q2 blowout ($4.24 vs $3.29; rev $9.56B); record $53B backlog, 95% of big projects "still in engineering"; HICO breaker deal widens the critical-path moat; FY EPS guide $16.45–16.95 hits the "2030 doubling" early | Q3 backlog conversion |
| 3 | CCJ | ai-capex-to-power-and-materials-cascade + kazatomprom-supply-cut-to-western-uranium-premium | Med-high (0.73/0.60) | Yes — $88.23 (+4.3%, −35% off hi) | Uranium premium intact; DNN/NXE legs beat sector (+9.7%/+6.6%); Q2 reported today 07-31 (transcript pending → tomorrow's ingest) | Q2 call read |
| 4 | TSM | hbm-cowos-as-binding-bottleneck | Med-high (0.70) | At base — $403.31 (+7.6%, −16% off hi) | 18–19× fwd; CoWoS booked thru 2027; +custom-ASIC broadening | N2 ramp |
| 5 | NOW | agentic-ai-seat-erosion-to-saas-rerate | Med-high (0.62) | Yes — $110.07 (−4.9%, −45% off hi) | Q2 beat+raise held; base $140–150 (~+30%); today's pullback re-opens the value; FCF-positive | Q3 print; ACV conversion |
⚠ LIN (0.78) — $508.64 (flat), near target → watchlist. LLY (0.60) — $1154.97 (−4.6%) → watchlist. MET/PRU (higher-for-longer) both near 52w highs → largely priced, watchlist (below).
Watchlist
| Ticker | Thesis | Conviction | Price | Undervalued vs base? | Next catalyst |
|---|---|---|---|---|---|
| MU | hbm-cowos-as-binding-bottleneck | High | $874.66 (+18.4%) | Yes | HBM4 ramp |
| PWR | pwr-transformer-moat-to-eps-doubling | Med-high | $657.98 (+17.3%) | Marginal | Q3 backlog conversion |
| CCJ | kazatomprom-supply-cut-to-western-uranium-premium | Med-high | $88.23 | Yes | Q2 today |
| TSM | hbm-cowos-as-binding-bottleneck | Med-high | $403.31 (+7.6%) | At base | N2 ramp |
| NOW | agentic-ai-seat-erosion-to-saas-rerate | Med-high | $110.07 (−4.9%) | Yes | Q3 print |
| MET | higher-for-longer-to-life-annuity-spread-rerate | hypothesis | $97.07 (52w hi) | No (priced) | Q2 NII early Aug |
| PRU | higher-for-longer-to-life-annuity-spread-rerate | hypothesis | $122.77 (52w hi) | No (priced) | Q2 NII early Aug |
Undervalued candidates (today)
- agentic-ai-seat-erosion-to-saas-rerate — NOW $110.07, −4.9% and −45% off high while the rest of the tape ripped — the seat-erosion re-rate value re-opens; FCF-positive, the trait the FCF-ROI rule rewards.
- hbm-cowos-as-binding-bottleneck — MU +18.4% is a re-rate toward base, not away — the AWS-reaccel demand confirmation is the thesis, not exhaustion; still −30% off high.
New chains to investigate (hypothesis-stage)
- solid-state-battery-bom-shift-to-lithium-metal — QS / ALB (thin materials/critical-minerals vertical, breadth-2a steer). Construction Physics: the >$4B solid-state-battery wave re-weights the battery BOM (pure lithium-metal anode, solid sulfide/halide electrolyte → more lithium metal, less graphite/separator). ⚠ Source names only foreign/private players (CATL/BYD/LG/Samsung); the US tradeable (QS) exposure is the gap, readiness 4/9, manufacturing the binding barrier. Priority low. To graduate: a QS dated commercial-volume milestone with a named OEM offtake.
New theses (now active)
- None. A big corroboration/confirmation day (MU/PWR/AMZN); net-new steered to a thin-vertical hypothesis + a thin-financials question strengthening (breadth 2a: ai-infra 47% ⚠ over).
Updated theses (existing active)
- ai-roi-reckoning — the FCF rule refines to a visible-ROI discriminator. AMZN FCF-negative + huge capex but +8–9% on AWS +37% reaccel; META punished for a return-less raise; Apple control case.
as_of→07-31. - pwr-transformer-moat-to-eps-doubling — conviction medium→medium-high on the Q2 blowout. Record $53B backlog with "95% not in backlog — still in engineering" (pipeline ahead of the record book); HICO high-voltage-breaker deal internalizes a second critical-path component; FY EPS guide $16.45–16.95 already at the "double by 2030" level. Step 3 partial→confirmed.
as_of→07-31. - higher-for-longer-to-life-annuity-spread-rerate — first external corroboration with named tradeables (MET, PRU). Bianco (Macro Voices): the bond market, not the Fed, disciplines inflation → higher-long rates (10Y 4.677%, term premium 0.73%). Step 3 open→toward partial; gap remains a disclosed life-insurer spread (MET/PRU Q2, early Aug). Both near 52w highs → largely priced.
Contradicted / weakened
- None today. 4d/4e gate: 23 live rows; 2 flags <−10% sector — TJX −12.1% (medium → no gate) and PWR −10.4% (med-high, but the mark is pre-Q2-beat; today's +17.3% beat confirms the thesis → gate acknowledged, not fired). 0 auto-calibrations (8th consecutive mark). Deep returns (MP −36%, CCJ −22%, UUUU −23%) are critical-minerals beta, yet the cluster's sector-excess is +0.8% — the picks beat a de-rated theme.
Open questions worth a human's eye
- Does CCJ's Q2 call (07-31) confirm the uranium-premium leg first-party? Transcript wasn't posted by run time → tomorrow's earnings-ingest. The DNN/NXE legs already beat sector; CCJ is the liquid anchor.
- Is MU's +18.4% the start of the asset-heavy leg decoupling from the AI-ROI skepticism, now that AWS reaccel gave the visible-ROI proof? The whole MU thesis rides on the consumption leg being rewarded while FCF-negative raisers are faded.
- Do MET/PRU Q2 calls (early Aug) show a widening disclosed net investment spread? That's the single datapoint that graduates the life/annuity chain from hypothesis to active — but both already sit at 52w highs.
What I looked at
- Run status: headless; steps 0.5–7 in-turn; price fetch paced 8/min — 47/47 priced, 0 429s. Weekly ledger (4d) on live-digest basis (hypothetical re-mark deferred). Per-concept bull/base/bear narratives carried forward (headless budget); load-bearing marks in the feed + tables.
- Breadth (2a): ai-infrastructure 47% ⚠ over (43 distinct chains) → net-new steered to thin materials (solid-state BOM shift) and thin financials (life/annuity spread). Bucket sourcing came via the ingestion veins + one financials/rate-regime autoresearch.
- Ingestion veins (0.5–0.7): podcast 1 (Macro Voices #543 Bianco — diarized AssemblyAI, $0.19); earnings 1 (PWR Q2; CCJ/LIN report today but transcripts not yet posted → next run); feeds 1 (Construction Physics solid-state battery; SemiAnalysis nothing new, FK paywalled).
- Steps 1–2b: 2 autoresearch (AMZN/AAPL FCF-ROI discriminator; bond-market-disciplines-inflation → MET/PRU). prospect-chains: no new auto-draft (wiki well-connected; today's net-new mapped onto existing pages).
- Sources: 5 promoted + 5 ingested → 4 created (amzn, met-lifeco, pru, solid-state-battery-bom-shift-to-lithium-metal) + ~5 updated (ai-roi-reckoning[refined], pwr-transformer-moat[conviction↑], higher-for-longer-spread, pwr, +quotes bianco/earl-austin). 0 new mechanisms (1 strengthened: PWR step 3→confirmed).
- Live feedback (4e): 24 records; 2 flags <−10% sector, 0 auto-calibrations (TJX medium; PWR med-high but pre-beat/stale). Best live sector-excess: AMZN +11.1% (the live twin of today's visible-ROI read), DNN +9.7%, CF/CBAM +7.4%, NXE +6.6%.
- Signal feed (7): full active/armed set (see exports/signals.json). Prices re-marked 07-31;
as_ofre-dated only for rows whose research moved today (ai-roi-reckoning, pwr-transformer-moat).