A weekly read of how the evidence shifted — all 10 theses, one digest per ISO week. Evidence language only; the picks vs benchmark are the receipts.
| thesis | start | end | shift |
|---|---|---|---|
| Connectivity Is the New Bottleneck new | supports | strongly supports | strengthened |
| Physical AI new | supports | supports | nudged up |
| Debasement new | mixed | mixed | nudged down |
| Profit for Size new | supports | strongly supports | nudged up |
Held steady: app layer, integration gap, memory, musk premium, power, seat software
NOW has the cleanest mechanism exposure today: AI ACV already >$1B, +24% rev growth and 33.7% FCF margin versus INTU +10.4% and SAP +6%, with 74.8% gross margins; still -54.6% off highs, so growth plus workflow lock-in beats INTU’s cheaper 10x fwd P/E and SAP’s scale.
BTC-USD is the purest scarce-asset expression of the debasement mechanism, trading -49.3% vs 52w high with 30d outperformance vs QQQ of +5.1%; it beats MSTR (mNAV 0.82x discount but equity/operational overlay) and STRC (preferred at 86.75, capped upside).
ACN is the purest integrator play, posting record $21.1B bookings with a sharp gen-AI jump, yet still at fwd P/E 10 and -51.9% vs 52w high with +18% 30d vs QQQ; that beats PLTR's richer 59x multiple and IBM's negative relative momentum and near-flat growth.
MLI offers the cleanest compounding exposure with 25.5% revenue growth, 11.3% FCF margin and a 15.7% rise in adjusted operating income, still at 15x fwd P/E and 10% below its 52w high—superior to CROX's -1.7% revenue decline despite the 9x multiple, and to SYF's 7x multiple which lacks comparable high-margin FCF quality.
| thesis | pick | WTD | bench | alpha |
|---|---|---|---|---|
| The Musk Premium short | $TSLA | +18.7% | -0.7% | +19.4% |
| Memory Is the Real Shortage | $MU | +10.4% | +2.6% | +7.8% |
| Power Is the Next Bottleneck | $CEG | +9.7% | +2.3% | +7.4% |
| Connectivity Is the New Bottleneck | $MRVL | +5.5% | +2.6% | +2.9% |
| AI Eats Seat-Based Software short | $MNDY | +2.8% | -4.7% | +7.5% |
| App Layer Eats the Model Layer | $INTU | +0.4% | -0.7% | +1.1% |
| Profit for Size | $SYF | -1.2% | -0.2% | -1.0% |
| Debasement | $MSTR | -1.5% | -0.7% | -0.8% |
| The Integration Gap | $PLTR | -6.6% | -0.7% | -5.9% |
| Physical AI | $TSLA | -18.7% | -0.7% | -17.9% |
Entry ref = last close before Mon; exit ref = last close on/before the week's end. Short picks invert: positive = the thesis advanced. Equal-weight, no transaction costs.
Headlines show current cloud backlog at ~€22.9B, up 26% and reversing two prior weak quarters, which supports the cloud ERP / system-of-record adoption leg of the expectation and the thesis that proprietary enterprise workflows remain valuable. However, EPS missed consensus, the stock fell, cloud-transition speed disappointed some investors, and operating-profit outlook was cut on M&A dilution, while coverage is thin-to-absent on Business AI/Joule attach, AI bookings/ARR, S/4HANA net-new logos, and model-partnership commentary—so AI monetization proof is missing and the print does not cleanly
To support the thesis that HBM-driven memory shortage is forcing sustained capex into etch/deposition, LRCX must show accelerating memory-segment systems revenue and robust bookings/guidance tied to HBM stack intensity, confirming tool demand remains the binding bottleneck rather than easing. A miss on memory revenue, flattish or declining memory bookings, or cautious capex commentary would undercut the mechanism by signaling HBM capacity additions or AI demand are no longer outrunning supply, reasserting the old cycle.
To support the debasement thesis, the print must confirm MSTR is still accumulating BTC as a leveraged scarce-asset wrapper without stressing the capital structure—i.e., runway comfortably above 12 months and no interruption to preferred dividends—so the corporate vehicle remains viable through the stress-test phase. Any disclosure of runway compressing toward <12 months, STRC/preferred pressure, or a pause in BTC buys would undercut or kill the thesis by breaking the accumulation mechanism.
To support the thesis that CROX is a mispriced durable profit machine, the print must show operating margins holding near/above 20% and FCF margin remaining solidly positive (~10%+) even with flat-to-down revenue, proving cash keeps compounding at a low multiple rather than fading with the fad narrative. It would undercut or kill the thesis if margins compress materially or FCF deteriorates, confirming the cheap fwd P/E is a value trap as profitability erodes instead of re-rating on durable cash generation.
Whether Kimi K3 open weights on July 27 actually close the gap to latest closed frontier on independent benchmarks within the 3-month double-down window.
Whether any hyperscaler or optical supplier quantifies 1.6T/CPO order conversion or shipment dollars that would test the double-down path while the names remain depressed.
Whether spot BTC ETF flows flip back positive tomorrow and if STRC can hold above the mid-80s.
Watch whether hyperscaler/lab commentary quantifies thousands of new deployment-engineer hires to trigger the double-down criteria.
SK Hynix Q2 earnings on July 29 for confirmation of HBM sold-out status, contract pricing, and margin trajectory.
Whether TSLA auto gross margin and FCF show any stabilization commentary or if DXYZ's discount to NAV widens further on the SpaceX mark-down.
Any verified Optimus unit output or supplier/chip resolution details that would reverse the current zero-production stall.
Any concrete multi-GW hyperscaler PPA at premium pricing or capacity-auction clearing prices that would confirm escalating scarcity.
Whether SYF and MLI begin to re-rate on the back of these beats, or if the next cheap-quality peer prints show margin erosion that would validate the value-trap kill path.
Any seat-count, NRR, or AI-pricing commentary in the next Monday.com/HubSpot prints that shows whether agent pivots are monetizing or cannibalizing seats.
computed 2026-07-25 · #2026-W30
One email each week: which theses gained or lost evidence, what tripped a flag, how the picks did against their benchmark. No pitches, unsubscribe in one click.