Stock-market dispatch — 2026-07-27

Headless run (Mon). Research output, not advice — brain never trades. Marks are live 07-27 intraday twelvedata quotes (markets open); 47/47 symbols priced, 0 429s (api.twelvedata.com reachable; SSNLF OTC excluded → stale). Supabase sync + push are performed by the wrapper, not this run.

Top of mind

SLB's Q2 call turned the oilfield→data-center pivot from a story into a print, and the tape agreed (+11%). Data Center Solutions grew +80% YoY / +33% sequential with the run-rate target raised to >$2B annualized by end-2027 on new (plural) hyperscaler customers — clearing slb-oilfield-pivot-to-dc-infrastructure's own graduation bar and moving Steps 2–3 partial→confirmed; the chain graduates low-med→medium and enters the feed. It's a differentiated instrument for AI-DC exposure (an energy-services vehicle, not a tenth semis name) even as the cluster stays over-weight. The day's other signal was breadth, not AI: today's ingest seeded three thin verticals — healthcare (two new chains: biosecure-cdmo-scarcity-to-western-cdmo-pricing-power→TMO and managed-care-mcr-inflection-to-payer-rerate→UNH/ELV, the latter confirmed first-party by UNH's 8-quarter-low MCR), industrials ex-AI (HON beat-and-raise, order-led, 8 "high-growth verticals"), and materials/financials (two prospect-chain hypotheses, PPTA + CRBG). Meanwhile AI-infra/semis sold off hard (INTC −7.9%, MU −7.0%, ARM −8.1%, MP −7.5%, WST −7.7%) while NOW ripped +7.4% two weeks after its beat-and-raise.

Strongest-conviction buys

Ranked shortlist (4c). Prices 07-27 intraday, twelvedata.

⚠ LIN (high, 0.78) — $512.28 (+1.2%), through target → watchlist. LLY (med-high 0.60) — $1196 (+0.9%), near 52w hi → priced-for-perfection, watchlist.

Watchlist

Undervalued candidates (today)

• agentic-ai-seat-erosion-to-saas-rerate — NOW $98.78, still −53% off high, base $140–150; the +7.4% today is the market starting to close a gap that persisted two weeks post-beat.
• hbm-cowos-as-binding-bottleneck — MU $920.95, −7.0% into a broad semis selloff on a thesis whose supply-constraint leg keeps getting first-party corroboration.

New chains to investigate (hypothesis-stage)

• higher-for-longer-to-life-annuity-spread-rerate — CRBG (EQH/JXN/MET). Prospect-chains: the Warsh higher-for-longer regime widens long-duration life/annuity general-account spread — the rate-regime beneficiary the book only routes through short-duration P&C float + brokerage NII. To graduate: a life/annuity name's spread-income disclosure confirming the long-end lift (steps 3–4 ⚠ unverified). → /explore-chain.
• china-antimony-controls-to-domestic-producer-rerate — PPTA. China's antimony export controls survived the Nov-2026 REE truce (11-company whitelist) → strategic premium to the sole US antimony resource, a beneficiary the REE-only stack never names. To graduate: a US-gov offtake/price-support signal or PPTA production milestone (steps 3–4 ⚠ unverified). → /explore-chain.

New theses (now active)

• slb-oilfield-pivot-to-dc-infrastructure — SLB, medium (0.52), graduated on its own Q2 call (DC Solutions +80% YoY, >$2B run-rate by end-2027). Emitted to the feed.
• managed-care-mcr-inflection-to-payer-rerate — UNH/ELV, medium (0.50). UNH Q2 MCR 86.7% (8-qtr low) + FY guide $19.50–20; ELV FY ≥$27. Cost-trend scare normalizing; much of the re-rate already realized (falsifier: Medicaid still trough — CNC/MOH).
• biosecure-cdmo-scarcity-to-western-cdmo-pricing-power — TMO, low (0.42). BIOSECURE (law) forces re-sourcing off Chinese CDMOs into a genuinely capacity-scarce Western market; TMO the cleanest listed beneficiary, exposure leg still open.

Updated theses (existing active)

• cuda-moat-erosion-to-nvda-rerate / cuda-moat-erosion-at-inference — SemiAnalysis "Can AMD break the CUDA moat?" (independent written corroboration) upgrades AMD to "great chance" — but inference-specific and conditional on two unresolved risks (Helios rack ramp; unstable internal GPU clusters), and the same source frames disaggregated-inference as Nvidia's emerging moat ("pie growing for everyone"). AMD added as a conditional long-side name; NVDA conviction held low (0.32) — corroborates share-shift-at-inference, not "NVDA does poorly." as_of→07-27.
• ai-gas-demand-to-appalachian-producer-price-floor — EQT: basis-tightening leg confirmed & strengthened (record M-2 winter strips), but the realized producer-price-floor leg re-tagged FORWARD (2027+) — EQT Q2'26 realized prices fell / EPS miss; part of the basis strength is mechanically inverse to a soft Henry Hub, not pure demand-pull. Held medium (page already framed the floor forward). as_of→07-27. Calibration filed.

Contradicted / weakened

• No active thesis falsified. 2 calibrations filed (both refinements, no downgrades): (1) 4e gate fired — kazatomprom-supply-cut-to-western-uranium-premium (med-high) × UUUU at −12.6% sector-excess, but NXE (+5.8%) / DNN (+9.7%) legs of the same mechanism beat sector → vehicle-selection miss, not thesis miss (relative-position-mismeasured); thesis held med-high. (2) Gas producer-price-floor refined to a 2027+ forward event (endpoints-evidenced-middle-assumed).

Open questions worth a human's eye

• UUUU vs NXE/DNN: the uranium chain is working, but the highest-beta vehicle lagged. Does the developer-tier (UUUU) systematically under-deliver theme torque, or is this one drawdown regime?
• SLB & UNH both ran on the print — the two freshest confirmed chains are already re-rated. Is the edge gone (react-to-news) or is there a durable second leg (SLB DC run-rate compounding; UNH MCR holding)?
• NOW +7.4% — is the two-week-late move the start of the gap-closing the thesis predicted, or a dead-cat in a broad SaaS/semis unwind?

What I looked at

• Run status: headless; steps 0.5–7 in-turn; price fetch paced 8/min — 47/47 priced, 0 429s. Step 4d (weekly ledger) not due (last mark Fri 07-24). Full per-concept bull/base/bear narrative carried forward (headless budget); the load-bearing marks live in the signal feed + tables above.
• Breadth (2a): ai-infrastructure 48% ⚠ over → net-new steered to thin verticals: healthcare (BIOSECURE→TMO, MCR→UNH/ELV), materials (PPTA antimony), financials/rate-regime (CRBG life-annuity). SLB pivot emitted is ai-infra-cluster but a differentiated instrument (graduated, not net-new-drafted).
• Ingestion veins (0.5–0.7): podcast 4 (Biotech Hangout 190, All-In open-source, Capital Allocators/NYU [zero-novelty], Moonshots — diarized AssemblyAI, ~$1.08); earnings 2 (HON Q2 beat-and-raise, SLB Q2 — DC pivot); feeds 1 (SemiAnalysis AMD/CUDA, partial paywall). FabKnowledge/Apricitas/Construction-Physics nothing new/on-topic.
• Steps 1–2b: 2 autoresearch (gas producer-floor gap → floor leg forward/2027+; healthcare bucket → BIOSECURE/MCR/GLP-1 chains). prospect-chains: 2 auto-drafted (CRBG, PPTA). A circulating "ClinicalInvestor July-2026 PDUFA calendar" was found unreliable (bad dates) and built on by nothing.
• Sources: 9 promoted + 9 ingested → 6 created (2 mechanisms [biosecure-cdmo, managed-care-mcr]; 4 entities [TMO, ELV, NVO, DYN] + 1 person); ~11 updated (hon/slb/amd/unh/lilly/eqt entities; cuda concepts; biosecure question; slb/cuda/gas mechanisms); AMD now carried as conditional long; 4 zero-novelty (NYU endowment, All-In, Moonshots, HON-entity-only).
• Live feedback (4e): 23 rows, 2 flags < −10% sector (PWR −16.9% medium → no gate; UUUU −12.6% medium-high → gate FIRED, 1st time). Best: CF/CBAM +11.0%, DNN +9.7% sector.
• Signal feed (7): full active/armed set — 57 rows, validated 57/57 clean (dry-run; Supabase sync deferred to wrapper). +4 net-new rows (SLB, TMO, UNH, ELV). Prices re-marked 07-27; as_of re-dated only for movers (EQT gas-floor, NVDA cuda-moat, + the 4 new); others carried forward.