in10x//theses
receipts upd 2026-07-24 19:42 CEST
EVIDENCE, TRACKED DAILY//UPDATED WEEKDAYS 07:00 CET

10 THESES

Ten investment theses I actually hold, with the kill criteria written down upfront so I can't quietly move the goalposts. Every weekday my pipeline pulls the prices, fundamentals and news, and an AI scores the evidence — it publishes whether it agrees with me or not. You decide what you believe.

// the stack rank — where the evidence points right now

my honest ordering: today's evidence strength first, and each pick's edge over its benchmark since it was named. it will be wrong sometimes — that's what the receipts are for.

// the brain — how the theses connect

T1 ↔ T3 · reinforce: same driver, two exits: models commoditize — value escapes the labs into products (T1) or into the people who deploy them (T3)T1 ↔ T2 · reinforce: one capability wave, two surfaces: intelligence lands in software apps and in robotsT1 ↔ T4 · reinforce: shared driver: if AI usage keeps scaling, inference scales — and data movement is the bottleneck that gets paidT2 ↔ T6 · tension: $TSLA sits on both sides: if Optimus ships, the premium is earned (T2); if it doesn't, the premium compresses (T6)T6 ↔ T7 · reinforce: one factor, long and short: price gravitates toward cash generation — short the narrative (T6), long the cash machines (T7)T5 ↔ T7 · reinforce: both bet on repricing toward the real: scarce assets (T5) and durable cash flows (T7) over paper promisesT4 ↔ T8 · reinforce: one buildout, two bottlenecks: the data has to move (T4) and the racks have to be powered (T8)T1 ↔ T8 · reinforce: demand flows downhill: if AI usage keeps compounding, inference eats electronsT9 ↔ T4 · reinforce: adjacent bottlenecks in the same buildout: bandwidth between chips (T4) and the memory beside them (T9)T9 ↔ T8 · reinforce: the buildout's two scarcities: electrons (T8) and bits (T9) — both priced by shortage nowT1 ↔ T10 · tension: same driver, opposite ends: T1 longs applications with proprietary data, T10 shorts the seat-priced ones without a moat — if AI value flows to apps indiscriminately, T10 dies; if it discriminates, both winT1 App Layer Eats the Model Layer — today: strongly supportsT1app_layerT2 Physical AI — today: supportsT2physical_aiT3 The Integration Gap — today: strongly supportsT3integration_gapT4 Connectivity Is the New Bottleneck — today: strongly supportsT4connectivityT5 Debasement — today: mixedT5debasementT6 The Musk Premium — today: strongly supportsT6musk_premium · shortT7 Profit for Size — today: strongly supportsT7profit_for_sizeT8 Power Is the Next Bottleneck — today: supportsT8powerT9 Memory Is the Real Shortage — today: strongly supportsT9memoryT10 AI Eats Seat-Based Software — today: supportsT10seat_software · short
10 theses · 11 synapses · node color = today's evidence · size = strength of today's read · the map grows as theses are added
T1T3 reinforce same driver, two exits: models commoditize — value escapes the labs into products (T1) or into the people who deploy them (T3)
T1T2 reinforce one capability wave, two surfaces: intelligence lands in software apps and in robots
T1T4 reinforce shared driver: if AI usage keeps scaling, inference scales — and data movement is the bottleneck that gets paid
T2T6 tension $TSLA sits on both sides: if Optimus ships, the premium is earned (T2); if it doesn't, the premium compresses (T6)
T6T7 reinforce one factor, long and short: price gravitates toward cash generation — short the narrative (T6), long the cash machines (T7)
T5T7 reinforce both bet on repricing toward the real: scarce assets (T5) and durable cash flows (T7) over paper promises
T4T8 reinforce one buildout, two bottlenecks: the data has to move (T4) and the racks have to be powered (T8)
T1T8 reinforce demand flows downhill: if AI usage keeps compounding, inference eats electrons
T9T4 reinforce adjacent bottlenecks in the same buildout: bandwidth between chips (T4) and the memory beside them (T9)
T9T8 reinforce the buildout's two scarcities: electrons (T8) and bits (T9) — both priced by shortage now
T1T10 tension same driver, opposite ends: T1 longs applications with proprietary data, T10 shorts the seat-priced ones without a moat — if AI value flows to apps indiscriminately, T10 dies; if it discriminates, both win
// receipts → every pick and earnings call, timestamped · scored against the market · nothing edited after the fact

// theses — tap a card for the companies, criteria + evidence

T1 // app_layer
App Layer Eats the Model Layer

Intelligence becomes free; value flows to applications with proprietary data and workflows, not model labs

today's evidence: strongly supports pick: $NOW
NOW +6.9% · INTU +14.8% · SAP +7.2% 30d vs QQQ
→ watch: 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.
T2 // physical_ai
Physical AI

VLA models turn robots into software platforms; public markets underprice it because leaders are private

today's evidence: supports pick: $TSLA
TSLA -14.7% · KOID -6.7% · SYM +5.7% 30d vs QQQ
→ watch: Any verified Optimus unit output or supplier/chip resolution details that would reverse the current zero-production stall.
T3 // integration_gap
The Integration Gap

Enterprise AI is hard; value flows to integrators and AI-literate services while models commoditize

today's evidence: strongly supports pick: $ACN
PLTR +11.8% · ACN +17.0% · IBM -15.6% 30d vs QQQ
→ watch: Watch whether hyperscaler/lab commentary quantifies thousands of new deployment-engineer hires to trigger the double-down criteria.
T4 // connectivity
Connectivity Is the New Bottleneck

AI bottleneck shifted from compute to data movement; CPO/photonics wins whether the buildout continues or rationalizes

today's evidence: strongly supports pick: $MRVL
MRVL -20.4% · ALAB -17.0% · COHR -18.3% 30d vs SMH
→ watch: 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.
T5 // debasement
Debasement

Deficits get inflated away; scarce assets reprice. Position built during the stress test, not the party

today's evidence: mixed pick: $BTC-USD
BTC-USD +5.1% · MSTR +2.1% · STRC +11.4% 30d vs QQQ
→ watch: Whether spot BTC ETF flows flip back positive tomorrow and if STRC can hold above the mid-80s.
T6 // musk_premium
The Musk Premium

The Musk empire trades on narrative, not cash. Tesla's auto cash cow is cracking — Q2'26 operating profit fell 57%, free cash flow went negative, and regulatory credits (once ~2pts of margin) collapsed as the EV credit expired — yet it's priced at ~145x earnings; the private crown jewels are only reachable at steep premiums to NAV. The bet: the premium compresses toward the cash the cows actually produce.

today's evidence: strongly supports short: $TSLA
TSLA -14.7% · DXYZ -7.3% · RKLB -21.2% 30d vs QQQ
→ watch: 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.
T7 // profit_for_size
Profit for Size

Companies generating outsized revenue and profit relative to their market cap — high margins, real free cash flow, low multiples — are mispriced. The market treats durable profit machines like they're stagnant; as the cash keeps compounding, they re-rate upward.

today's evidence: strongly supports pick: $MLI
SYF -5.9% · CROX +10.4% · MLI -5.3% 30d vs SPY
→ watch: 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.
T8 // power
Power Is the Next Bottleneck

The AI buildout's binding constraint is shifting from chips to electricity. Datacenters are signing decade-long power deals faster than anyone can build generation; firms that own or equip gigawatts capture pricing power for years

today's evidence: supports pick: $CEG
VST +1.0% · CEG +1.8% · GEV -5.0% 30d vs XLU
→ watch: Any concrete multi-GW hyperscaler PPA at premium pricing or capacity-auction clearing prices that would confirm escalating scarcity.
T9 // memory
Memory Is the Real Shortage

AI turned memory from a cyclical commodity into the binding constraint — HBM is sold out years ahead, conventional DRAM and NAND supply is being cannibalized to feed it, and pricing power has shifted to a three-player oligopoly for the first time in decades

today's evidence: strongly supports pick: $MU
MU -2.8% · SNDK -13.9% · LRCX -8.8% 30d vs SMH
→ watch: SK Hynix Q2 earnings on July 29 for confirmation of HBM sold-out status, contract pricing, and margin trajectory.
T10 // seat_software
AI Eats Seat-Based Software

Per-seat software loses its unit of account when agents do the work. Mid-market SaaS with seat pricing and thin data moats gets repriced first: seat counts stall, net revenue retention slips, and the multiple compresses. The bet: these names underperform the software index

today's evidence: supports short: $MNDY
HUBS +8.4% · MNDY +5.2% · DOCU +10.1% 30d vs IGV
→ watch: 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.

// earnings on deck — the theses get tested

// the tape — the AI's daily read, unedited

2026-07-24
T1 Chinese/open models hit record ~58% OpenRouter token share and Kimi K3/Qwen 3.8 landed as frontier-competitive open weights, while ServiceNow printed $1B AI ACV and raised sub revenue guidance.
T4 Eoptolink 1.6T ramp plus STM AI-optical read-through and copper-to-photonics bottleneck coverage strengthened CPO and interconnect legs, while Alphabet’s 2026 capex hike kept spend intact despite FCF-pressure headlines; 30d relative weakness vs SMH still capped the score.
T5 BTC ETF 7-day inflow streak snapped with ~$225M outflows while debasement/gold headlines stayed constructive and mNAV discount widened to 0.82x with STRC still at 86.75.
T3 a16z/Deloitte FDE expansions and unchanged 57% AI-ROI-failure studies plus ACN's record $21.1B bookings continue to confirm hard deployment and integrator capture.
T9 DRAM spot prices at record highs with Q3 contracts set to surge 25%+ and multi-year shortage commentary strongly affirm pricing power and supply tightness, partially offset by the basket's -8.5% 30d underperformance vs SMH.
T6 Q2 aftermath keeps confirming the cash-cow crack (op profit -57%, FCF negative, credits collapsed, op margin 1.4%) while SpaceX trades below IPO price and space names keep giving back mania gains.
T2 Brooklyn 3PL Ultra Robotics OP1 confirmed filling 30% of e-commerce orders as non-pilot plus Humanoid 34k pre-orders/$2.4B and nine Fortune 500 deployments lifted commercial signals, while Optimus stayed at zero units with Musk stressing scale difficulty and mass-production language removed.
T8 Upward BNEF revisions to 194 GW / 20% of US power by 2035 plus FERC data-center rules and DOE nuclear funding reinforce scarcity, while basket 30d avg still trails XLU by 0.7%.
T7 SYF's Q2 beat and raised 2026 EPS outlook plus MLI's 15.7% adjusted operating-income growth kept margin_durability at +2, while 30d basket vs SPY flattened to -0.2% lifting price_momentum from -1 to 0.
T10 Deloitte and Monday.com AI-agent restructuring headlines strongly back the replacement mechanism, but no NRR/seat-print cracks and the basket's +7.9% 30d outperformance vs IGV work against the short.
2026-07-23
T1 Open-weight token share hit majority/~58% Chinese/open lead on OpenRouter and Kimi K3 landed as Opus-class, while ServiceNow printed $1B AI ACV and raised guidance, reinforcing app-layer monetization.
T4 Alphabet's 2026 capex hike plus Goldman/Eoptolink 1.6T–3.2T optical demand pull-forward reinforced both the spend and CPO legs, while 30d relative price weakness vs SMH still capped conviction.
T5 BTC spot ETF inflows extended to a 6-7 day streak totaling ~$900M+, reinforcing scarce-asset demand, while mNAV held a 0.86x discount and STRC stayed depressed at 87.45 with mixed gold/debasement headlines.
T3 Fresh a16z/Deloitte FDE expansion plus unchanged 57% AI-ROI-failure studies and ACN's record $21.1B AI bookings keep reinforcing that deployment stays hard and integrators capture value.
T6 Q2 print confirmed the cash-cow crack (op profit -57%, FCF negative, credits collapsed) while SpaceX lost its IPO premium and traded below issue price, directly advancing the compression thesis.
T2 Humanoid's 34k pre-orders worth $2.4B and nine Fortune 500 industrial deployments reinforce backlog/commercial signals while Optimus unit count stays zero after earnings.
T7 SYF and MLI delivered clear Q2 beats with SYF raising 2026 EPS outlook and MLI posting 15.7% adjusted operating-income growth, lifting margin_durability while value-rotation headlines turned mildly constructive.
2026-07-22
T1 Open-weight models hit record ~58% OpenRouter token share and Kimi K3 landed as Opus-class open weights, strongly validating intelligence commoditization toward the app layer.
T4 Strong CPO/3.2T demand pull-forward and silicon-photonics mass-production news lifted the mechanism score while IDC raised 2026 AI infra spend to $497B; 30d relative price weakness vs SMH capped the move.
T5 BTC ETF inflows turned positive for a second week after prolonged outflows, mildly supporting scarce-asset demand, while MSTR sits at 0.85x mNAV discount and STRC at 87.45 signals ongoing stress.
T3 Day-one evidence strongly backs the thesis via multiple pieces on enterprise AI stalling at execution/ROI and Accenture posting record $21.1B bookings with sharp gen-AI jump.
T2 Humanoid's $152M raise with claimed 9 industrial deployments and 34k pre-orders mildly supports commercial traction while Optimus unit count remains zero ahead of earnings.

// how this works

the data

Prices and relative performance daily (vs each thesis's benchmark — QQQ, SMH or SPY), company fundamentals (growth, margins, valuation, cash flow), and fresh headlines collected per mechanism signal — the things each thesis actually depends on.

the ai

Every weekday Grok reads the day's evidence and scores each signal −2…+2 against the thesis mechanics, with a one-line explanation. I publish both unedited — even when it disagrees with me. That's the tape.

the deal

I wrote the kill and double-down criteria before taking the positions — they're shown verbatim on every page. Sustained negative evidence puts a thesis into kill review. No quiet goalpost-moving.

disclaimer

A one-person experiment in accountable investing. Nothing here is investment advice or a recommendation to buy or sell anything. Do your own work — that's the point of showing the data.