How To Read This
A one-pager, not a repeat of the position page
The Holdings table shows what’s owned and its return. Each letter and news writeup covers what happened. This page is neither — it’s the underlying coverage case for every position, condensed to the fields a real initiating-coverage note would carry, so the full thesis for any name is readable in under two minutes.
Full Coverage List
32 positions, filterable by bucket
SNDK
Sandisk Corp
↓
Business Description
Spun off from Western Digital in Feb 2025, Sandisk is a pure-play NAND flash memory and storage products maker (SSDs, memory cards, retail flash). It runs a joint-venture fab arrangement with Kioxia in Japan rather than owning all its own capacity.
Investment Thesis
Largest single position in the book. Owns direct exposure to the NAND supply/demand imbalance created by AI-driven data-center storage demand colliding with underinvested industry capacity after years of memory-sector losses.
Bull Case
NAND pricing power extends through 2026-27 as hyperscaler SSD demand for AI training/inference data outstrips supply; Kioxia JV lets Sandisk expand output without full capex burden; enterprise SSD mix shift lifts margins structurally.
Base Case
Pricing normalizes off cycle peak but stays well above trough levels; Sandisk captures a fair share of AI-driven enterprise SSD growth while consumer/retail flash stays soft; earnings remain highly cyclical and volatile.
Bear Case
Memory is the most brutally cyclical sub-sector in semis; a supply response (Samsung, SK Hynix, Kioxia all expanding) or a slowdown in AI capex could crash pricing exactly as it did in 2022-23; Sandisk carries higher balance-sheet leverage as a recent spinoff.
Key Risks
Extreme cyclicality; capacity-expansion decisions by 3 competitors that don’t answer to this thesis; customer concentration in a handful of hyperscalers; JV structure with Kioxia limits full operating control.
Catalysts to Watch
Quarterly ASP (average selling price) commentary; hyperscaler capex guides; Kioxia capacity announcements; any TrendForce demand-normalization data points (already flagged as a near-term risk in recent research).
Valuation Snapshot
Trades at a low-teens-to-20s forward P/E on cycle-peak estimates, which the market historically discounts because NAND earnings are not run-rate — the multiple looks cheap only if the current pricing holds, which is the entire debate.
Why This Conviction Rating
Medium conviction despite being the largest position — sizing reflects a high-conviction cyclical call, but the 'Under Review' status flag signals the position is at or near its IPS cap and being actively reassessed after the recent TrendForce-driven pullback.
META
Meta Platforms
↓
Business Description
Meta operates Facebook, Instagram, WhatsApp, and Messenger, monetized almost entirely through advertising, plus Reality Labs (AR/VR/AI wearables), which remains loss-making. Now one of the largest AI capex spenders globally, building compute for both ad-ranking models and frontier AI (Llama family, Meta Superintelligence Labs).
Investment Thesis
Best-in-class digital advertising business with genuine AI-driven engagement and monetization gains (Advantage+, Reels ranking), funding an aggressive AI infrastructure buildout from enormous free cash flow rather than external capital.
Bull Case
AI-driven ad-load and ranking improvements keep lifting revenue per user well above peers; Meta’s own model stack (Llama, superintelligence effort) reduces dependence on external AI vendors; capex is self-funded, so the AI buildout doesn’t dilute or leverage the balance sheet.
Base Case
Core advertising keeps compounding at a mid-teens clip; Reality Labs losses persist but stay a manageable drag; AI capex stays elevated, compressing free-cash-flow conversion versus history without threatening the core franchise.
Bear Case
Capex intensity (reportedly entering the cloud-infrastructure/neocloud market per recent reports) pressures margins and FCF longer than the market expects; Reality Labs losses never inflect; regulatory pressure (antitrust, EU digital rules) caps monetization flexibility.
Key Risks
Advertising-cycle sensitivity to macro slowdowns; regulatory/antitrust overhang in US and EU; capex ROI uncertainty if AI infrastructure spend doesn’t translate to proportional revenue; competitive pressure from TikTok and AI-native content discovery.
Catalysts to Watch
Quarterly ad pricing/impression growth, capex guidance revisions, any concrete news on Meta’s cloud-infrastructure ambitions (the same story driving NBIS volatility), Reality Labs loss trajectory.
Valuation Snapshot
Mid-to-high-20s forward P/E for a business still compounding revenue in the mid-teens with expanding AI-driven margins — a premium multiple that assumes the ad-monetization gains are durable, not one-time.
Why This Conviction Rating
High conviction: an established, monetizing-today AI beneficiary rather than a pre-revenue infrastructure bet, which is why it sits alongside the index funds as a 'always-held' type position in the book’s own conviction framework.
AMD
Advanced Micro Devices
↓
Business Description
Designs x86 CPUs (server EPYC, client Ryzen) and GPUs/AI accelerators (Instinct MI300/MI350 series), competing directly with Nvidia in AI training/inference silicon and with Intel in CPUs. Fabless — manufactures via TSMC.
Investment Thesis
Second-source AI accelerator supplier as hyperscalers diversify away from single-vendor (Nvidia) dependence; a recent large cloud win (reported Microsoft Azure deal) is cited as a signal AMD is closing the competitive gap.
Bull Case
MI400-generation accelerators close the software/performance gap with Nvidia’s roadmap meaningfully enough that AMD captures durable double-digit share of the AI accelerator market; server CPU share gains against Intel continue; client PC recovery adds a second leg.
Base Case
AMD remains the credible #2 in AI accelerators with mid-single-digit-to-low-teens share, growing off a small base; server CPU share gains continue steadily; stock re-rates only modestly from here given a legitimate but not dominant position.
Bear Case
Nvidia’s CUDA software moat and next-gen Rubin platform keep AMD structurally behind on a per-GPU-economics basis; any AI capex digestion pause hits AMD’s smaller, less-diversified franchise harder than Nvidia’s.
Key Risks
Total dependence on TSMC for leading-edge nodes (same Taiwan concentration risk flagged for the whole semis sleeve); Nvidia’s dominant CUDA ecosystem lock-in; execution risk on MI-series ramp; cyclicality in PC/gaming segments.
Catalysts to Watch
MI400/MI450 ramp updates, hyperscaler capacity commitments (Microsoft, Oracle, Meta), server CPU share data versus Intel, any second major cloud AI-accelerator design win.
Valuation Snapshot
Forward P/E in the 30s-40s prices in continued AI-accelerator share gains; a re-rating catalyst (a second Microsoft-scale win) could justify further multiple expansion, but the stock is not statistically cheap on current numbers.
Why This Conviction Rating
Medium conviction: real, identifiable thesis (second-source AI silicon) but genuine execution/competitive uncertainty versus Nvidia keeps this a step below the book’s High-conviction names.
SPY
SPDR S&P 500 ETF
↓
Business Description
SPDR S&P 500 ETF Trust — passively tracks the S&P 500 index, the broadest large-cap U.S. equity benchmark.
Investment Thesis
Structural index-hedge ballast. Not a stock-picking position — it exists to dilute single-name and thematic concentration risk in a book that is otherwise heavily tilted toward semis and AI infrastructure.
Bull Case
Broad U.S. equity market continues compounding; SPY captures the upside of mega-cap tech/AI leadership without single-name blow-up risk.
Base Case
Market performs in line with long-run historical averages; SPY does its job as ballast, dampening portfolio-level volatility versus the concentrated sleeve.
Bear Case
A broad market drawdown still hurts — SPY offers no protection against a systemic correction, only against single-name idiosyncratic risk.
Key Risks
Full market beta; concentration within the index itself in mega-cap tech names that overlap thematically with the book’s other holdings (Meta, Broadcom-adjacent names all sit in the S&P 500 too).
Catalysts to Watch
Fed policy path, S&P 500 earnings season in aggregate, mega-cap tech weighting shifts.
Valuation Snapshot
Trades at the index’s blended forward P/E; not a name-specific valuation call.
Why This Conviction Rating
High conviction as a structural always-held hedge, not a return-driving bet — its role in the book is defined by the Investment Policy Statement’s index-hedge bucket, not by a stock thesis.
VOO
Vanguard S&P 500 ETF
↓
Business Description
Vanguard S&P 500 ETF — economically near-identical to SPY, tracking the same index at a lower expense ratio.
Investment Thesis
Same structural index-hedge role as SPY; holding both is largely a function of account/brokerage history rather than a differentiated thesis.
Bull Case
Same as SPY.
Base Case
Same as SPY.
Bear Case
Same as SPY.
Key Risks
Same market-beta and mega-cap concentration risk as SPY; redundant exposure with SPY (near-perfect 1.00 correlation per the book’s own matrix) is a portfolio-construction inefficiency worth noting.
Catalysts to Watch
Same as SPY.
Valuation Snapshot
Same as SPY, marginally cheaper expense ratio.
Why This Conviction Rating
High conviction, structural hedge — worth flagging that running both SPY and VOO at meaningful size adds sizing complexity without adding diversification, since the two are functionally the same exposure.
MU
Micron Technology
↓
Business Description
Micron is a major DRAM and NAND memory manufacturer, with growing exposure to High Bandwidth Memory (HBM) used in AI accelerators alongside Nvidia and AMD GPUs.
Investment Thesis
Direct HBM/DRAM beneficiary of the AI buildout, with better balance-sheet quality and manufacturing scale than smaller memory peers; the second leg of the book’s memory-cycle bet alongside SNDK.
Bull Case
HBM demand from AI accelerators keeps outstripping supply, driving structurally higher blended memory pricing and margins well beyond the historical cycle; Micron’s HBM3E/HBM4 roadmap keeps pace with Nvidia/AMD accelerator generations.
Base Case
Memory pricing stays elevated versus trough but moderates from cycle peak; Micron captures a fair share of HBM growth while conventional DRAM/NAND pricing normalizes.
Bear Case
Memory is memory — a capacity response from Samsung/SK Hynix or an AI-capex digestion pause collapses pricing the way it always eventually does in this sector’s history.
Key Risks
Extreme historical cyclicality; capital-intensive capacity expansion; customer concentration in a few AI-accelerator makers; Taiwan/geopolitical exposure through the broader supply chain.
Catalysts to Watch
Quarterly DRAM/NAND ASP trends, HBM revenue mix disclosure, hyperscaler AI capex guides, any TrendForce or industry-analyst demand-normalization calls (the same signal already weighing on SNDK).
Valuation Snapshot
Similar cycle-peak-earnings dynamic to SNDK — looks cheap on trailing numbers precisely because the market doubts current pricing is sustainable.
Why This Conviction Rating
Medium conviction — same cyclical-memory logic as SNDK, sized smaller, reflecting the same 'real thesis, real cyclicality risk' judgment.
PANW
Palo Alto Networks
↓
Business Description
Palo Alto Networks sells network security (firewalls), cloud security (Prisma), and security operations (Cortex/XSIAM) software, pursuing a 'platformization' strategy to consolidate customer security spend onto one vendor.
Investment Thesis
Best-of-breed cybersecurity platform consolidator — as enterprises rationalize point-tool sprawl, Palo Alto is positioned to be a primary beneficiary of platform bundling deals (reflected in its 0.88 and 0.72 correlation with CRWD and FTNT — the whole cyber block moves together).
Bull Case
Platformization deals (bundling firewall + cloud + SOC into one contract) accelerate, driving durable double-digit ARR growth and margin expansion as customers consolidate vendors.
Base Case
Steady mid-teens ARR growth continues; platformization is real but slower than bulls hope, competing against best-of-breed point solutions in each category.
Bear Case
AI-native security startups and hyperscaler-bundled security tools erode Palo Alto’s pricing power; platformization deals prove harder to close at scale than management guides.
Key Risks
Intensifying competition from CrowdStrike, Fortinet, and AI-native entrants; large-deal lumpiness; high multiple leaves little room for a growth deceleration.
Catalysts to Watch
Quarterly platformization deal count and ARR mix, next-gen security ARR growth, any competitive win/loss commentary versus CRWD.
Valuation Snapshot
Premium software multiple (high-30s/40s forward P/E) reflecting durable double-digit growth expectations — priced for continued execution, not for a slowdown.
Why This Conviction Rating
High conviction: an established competitive position in a structurally growing category (cybersecurity spend is one of the least discretionary IT budget lines).
INTC
Intel Corporation
↓
Business Description
Intel designs and manufactures (IDM model) x86 CPUs and is executing a foundry turnaround (Intel Foundry, 18A process node) intended to make it a leading-edge contract manufacturer alongside TSMC, backed by a U.S. government equity stake and CHIPS Act support.
Investment Thesis
A reshoring/foundry-turnaround bet: if 18A succeeds, Intel becomes the only U.S.-based leading-edge fab, a structurally different (and strategically valuable) risk profile than fabless peers entirely dependent on TSMC.
Bull Case
18A ramps successfully with real external foundry customers signed, U.S. government backing de-risks the balance sheet during the capex-heavy turnaround, and Intel re-establishes process leadership lost over the last decade.
Base Case
18A ships on a delayed but workable timeline with modest external customer wins; Intel stabilizes as a smaller, more focused company without fully reclaiming leadership, but the reshoring narrative supports a floor under the stock.
Bear Case
18A slips again or yields disappoint, external foundry customers don’t materialize at scale, and Intel remains structurally behind TSMC while burning cash on fab capex — the turnaround simply doesn’t turn.
Key Risks
Execution risk on the most complex process node transition in company history; heavy capex burden; foundry customer-acquisition risk (few companies want to bet critical silicon on an unproven node); geopolitical sensitivity given the explicit reshoring thesis.
Catalysts to Watch
18A yield and ramp updates, any named external foundry customer wins, government funding/equity developments, quarterly foundry segment losses narrowing (or not).
Valuation Snapshot
Hard to value on traditional earnings multiples given ongoing foundry losses; the market is effectively pricing a call option on turnaround success plus a strategic/reshoring premium.
Why This Conviction Rating
Medium conviction — explicitly framed in the book’s own stress-test as 'a bet on reshoring US-based leading-edge manufacturing,' a real but still unproven differentiator versus fully offshore-dependent peers.
EL
Estee Lauder Companies
↓
Business Description
Estee Lauder owns a portfolio of prestige beauty brands (Estee Lauder, MAC, Clinique, La Mer, Tom Ford Beauty) sold through department stores, specialty retail, travel retail, and e-commerce globally.
Investment Thesis
A turnaround/value name in the book’s Consumer sleeve — prestige beauty pricing power is real, but the company has been working through a multi-year China and travel-retail demand reset.
Bull Case
China prestige beauty demand and travel retail (duty-free) recover meaningfully, the cost-restructuring program (announced over the past two years) drives margin expansion, and brand reinvestment (Tom Ford, La Mer) reaccelerates growth.
Base Case
Gradual, uneven recovery — North America and EMEA stabilize while China/travel retail stays choppy; cost cuts support margins even as top-line growth stays modest.
Bear Case
Chinese consumer demand for prestige beauty structurally resets lower (shifting to domestic/mass brands), travel retail oversupply persists, and restructuring savings get reinvested rather than dropping to the bottom line.
Key Risks
Heavy China/travel-retail revenue concentration; brand-relevance risk versus newer indie beauty brands; multi-year restructuring execution risk; FX exposure given global footprint.
Catalysts to Watch
China same-store/prestige-beauty-category trends, travel-retail (duty-free) volume data, quarterly gross-margin progress on the restructuring plan, brand-level (La Mer, Tom Ford) growth.
Valuation Snapshot
Trades at a depressed multiple versus its own history, reflecting the market’s skepticism on the China/travel-retail recovery timeline — the modest 19.55% return reflects a name still mid-turnaround rather than re-rated.
Why This Conviction Rating
Medium conviction — a real, reasoned thesis (prestige brand pricing power, cost restructuring) with a specific, identified risk (China/travel-retail demand) that could keep compressing the multiple.
QQQ
Invesco QQQ Trust
↓
Business Description
Invesco QQQ Trust tracks the Nasdaq-100 — the 100 largest non-financial companies on Nasdaq, heavily weighted toward mega-cap tech and semiconductors.
Investment Thesis
A more concentrated, tech/semis-tilted index hedge than SPY/VOO — still a structural ballast position, but one that moves more in sympathy with the book’s own semiconductor and AI-infrastructure themes (0.68 correlation with AVGO, 0.60 with VRT).
Bull Case
Mega-cap tech and semis leadership continues, and QQQ captures broad AI-driven earnings growth across dozens of names at once, reducing single-stock risk within the theme.
Base Case
Tech/semis keep outperforming the broader market moderately, consistent with recent history.
Bear Case
A tech/semis-led drawdown hits QQQ harder than SPY precisely because of the same concentration that makes it attractive on the way up — it is a less diversifying hedge than SPY given its overlap with the book’s own thematic tilt.
Key Risks
Sector concentration risk (tech/semis dominate the index); less true diversification benefit for this specific book than a broader index fund would provide.
Catalysts to Watch
Mega-cap tech earnings season in aggregate, semiconductor sector sentiment.
Valuation Snapshot
Trades at the Nasdaq-100's blended forward multiple, elevated versus SPY given its tech tilt.
Why This Conviction Rating
High conviction as a structural hedge, though worth flagging it provides less true diversification for this particular book than SPY/VOO/SCHD do, given the overlap with the semiconductor and AI-infrastructure theme running through the rest of the portfolio.
SPCX
SpaceX
↓
Business Description
SpaceX (privately held) — Falcon 9/Falcon Heavy launch services, Starship development, and the Starlink satellite broadband constellation, now reportedly the dominant global commercial launch provider.
Investment Thesis
A speculative-sleeve bet on the dominant private space company, held through secondary/pre-IPO markets rather than public exchange — access itself is part of the thesis given SpaceX’s continued private status.
Bull Case
Starship reaches full operational reusability, Starlink scales into a major standalone profit center, and any eventual IPO or tender offer re-rates the implied valuation materially higher.
Base Case
Falcon 9 launch cadence and Starlink subscriber growth continue steadily; Starship development proceeds on a bumpy but forward-moving timeline; no near-term liquidity event.
Bear Case
Starship development setbacks persist, Starlink competition (Amazon Kuiper, OneWeb) pressures the broadband economics, and illiquidity means the position can’t be exited even if the thesis deteriorates.
Key Risks
Complete illiquidity — this is the one position in the book without same-day exit capability, a genuine outlier versus every other holding; no public financial disclosure; valuation is whatever the last private transaction says it is, not a market-clearing price; key-person risk.
Catalysts to Watch
Starship test flight outcomes, Starlink subscriber/revenue disclosures (rare), any secondary-market valuation marks, IPO speculation.
Valuation Snapshot
Priced off private secondary-market transactions rather than public multiples — the current unrealized loss reflects a markdown from the entry-round valuation.
Why This Conviction Rating
Low conviction/sizing by design: the book’s own compliance notes flag this as the only illiquid, non-modelable holding, appropriately capped as a small speculative-sleeve position.
TLN
Talen Energy Corporation
↓
Business Description
Talen Energy owns and operates power generation assets, most notably the Susquehanna nuclear plant, and has signed data-center power-supply agreements (including with Amazon) to feed electricity directly to AI infrastructure buildouts.
Investment Thesis
A direct power-for-AI-data-centers thesis — nuclear baseload capacity is scarce and increasingly valuable as hyperscalers scramble to secure firm power for new AI data centers.
Bull Case
More hyperscaler co-location/power-purchase deals get signed at attractive long-term rates, nuclear capacity value re-rates further as data-center power demand keeps outstripping grid buildout, and Talen becomes a template for the sector.
Base Case
Existing power agreements perform as contracted; incremental deals get signed at a steady pace; the stock tracks the broader power-for-AI theme without a major re-rating catalyst.
Bear Case
Regulatory or interconnection-queue delays slow new data-center power deals, nuclear relicensing/operational risk resurfaces, or a slowdown in AI capex growth reduces the urgency (and pricing power) behind data-center power deals.
Key Risks
Regulatory approval risk on power agreements; nuclear plant operational/safety risk; customer concentration in a small number of hyperscaler counterparties; interest-rate sensitivity as a capital-intensive utility-like business.
Catalysts to Watch
New power-purchase-agreement announcements, nuclear plant uprate/relicensing news, PJM capacity market pricing, hyperscaler data-center capex guidance.
Valuation Snapshot
Re-rated meaningfully off the AI-power thesis over the past two years; current multiple reflects real, contracted cash flows rather than pure speculation, though further upside depends on additional deal announcements.
Why This Conviction Rating
Medium conviction — a real, contracted thesis (data-center power agreements are signed, not hypothetical) but with identified counterparty-concentration and regulatory risk.
GOOGL
Alphabet Inc
↓
Business Description
Alphabet’s core businesses are Google Search, YouTube, Google Cloud, and Android/Play, monetized primarily through advertising and increasingly through cloud infrastructure and Gemini-branded AI products.
Investment Thesis
A high-conviction core holding: Search remains the most profitable advertising business ever built, Google Cloud has reached durable profitability, and Gemini gives Alphabet a credible frontier-AI position rather than a defensive one.
Bull Case
Gemini-powered AI Overviews/Search products defend and extend Search’s monetization rather than cannibalizing it, Google Cloud keeps taking AI-infrastructure share from AWS/Azure, and antitrust remedies land lighter than feared.
Base Case
Search grows steadily despite AI-search disruption fears, Cloud continues its profitable growth trajectory, and the business compounds at a solid low-double-digit rate overall.
Bear Case
AI chatbots (ChatGPT and peers) structurally erode Search query volume and ad monetization faster than Alphabet can offset with its own AI products, and antitrust remedies (search default-placement rulings) meaningfully impair the core business model.
Key Risks
Antitrust/regulatory overhang (ongoing search and ad-tech litigation); AI-driven disruption risk to the core Search query/ad model; heavy capex ramp to compete in AI infrastructure and models.
Catalysts to Watch
Search revenue growth and AI Overviews monetization data, Google Cloud growth/margin trajectory, Gemini model releases and enterprise adoption, antitrust remedy rulings.
Valuation Snapshot
Trades at a below-market-average multiple for its growth rate specifically because of the AI-disruption and antitrust overhangs — a valuation gap the bull case argues is overstated.
Why This Conviction Rating
High conviction: an established competitive position (still the dominant search engine and a top-3 cloud provider) with limited near-term uncertainty relative to the book’s more speculative AI-infrastructure names.
NBIL
GraniteShares 2x Long NBIS Daily ETF
↓
Business Description
GraniteShares 2x Long NBIS Daily ETF — a leveraged single-stock ETF designed to deliver approximately 2x the daily return of Nebius Group (NBIS), reset daily.
Investment Thesis
A leveraged, capped-size way to amplify conviction in the Nebius/AI-cloud-infrastructure theme without allocating additional capital to the underlying — explicitly the book’s leveraged-derivative sleeve, capped at 5% of the portfolio.
Bull Case
Nebius shares continue their strong run and the leveraged structure amplifies gains; used tactically and kept small, it adds convex upside without threatening portfolio survivability.
Base Case
Nebius stays volatile (as AI-infrastructure names have been, including the recent Meta-cloud-entry-driven derating), and daily-reset leverage decay causes the ETF’s longer-term return to lag a simple 2x of NBIS’s cumulative move — an inherent structural feature of daily-reset leveraged products, not a flaw specific to this fund.
Bear Case
A sustained NBIS drawdown is amplified 2x, and daily volatility decay compounds losses faster than the underlying stock alone — leveraged single-stock ETFs are specifically not designed to be held through extended choppy or declining periods.
Key Risks
Daily-reset compounding/volatility-decay risk (the single most important thing to understand about this instrument); complete dependence on Nebius-specific news flow and sentiment; not suitable as a long-term buy-and-hold vehicle by design.
Catalysts to Watch
Anything that moves NBIS itself — Nebius earnings (next report July 29, 2026), Meta cloud-infrastructure-entry headlines, broader neocloud sector sentiment.
Valuation Snapshot
Not a traditional valuation case — it’s a derivatives-based amplifier of the NBIS thesis; sizing discipline (capped, currently under-limit at 2.9%) matters more than any multiple.
Why This Conviction Rating
Low conviction/sizing by design — explicitly the book’s most levered, tactical position, appropriately capped well under its 5% IPS limit.
VRT
Vertiv Holdings Co
↓
Business Description
Vertiv makes critical digital infrastructure for data centers — power management, thermal/liquid cooling, and IT management systems — a direct picks-and-shovels supplier to the AI data-center buildout.
Investment Thesis
A core AI-infrastructure beneficiary one layer removed from the chips themselves: every AI data center, regardless of which GPU/accelerator wins, needs power and cooling equipment, giving Vertiv broader exposure across the whole AI capex cycle rather than a single-vendor bet.
Bull Case
Liquid-cooling adoption for next-gen high-density AI racks accelerates faster than expected, Vertiv’s backlog keeps growing as hyperscalers commit to multi-year data-center buildouts, and margins expand with the higher-value liquid-cooling mix.
Base Case
Steady backlog growth and margin expansion continue in line with the broader data-center capex cycle, without a step-change catalyst.
Bear Case
AI data-center capex growth decelerates or gets pushed out, competitive pressure in cooling/power equipment compresses margins, and the extremely tight correlation with power/utilities names (0.55 with AVGO, 0.60 with QQQ) means it falls hard in any AI-infrastructure-wide derating.
Key Risks
High sensitivity to hyperscaler capex cycles (a second-derivative bet on AI spending continuing); customer concentration in a handful of large data-center operators; competitive entry into liquid cooling from established thermal-management players.
Catalysts to Watch
Quarterly backlog/orders growth, liquid-cooling revenue mix disclosure, hyperscaler capex guidance (the same catalysts driving NBIS/CRWV sentiment), margin trajectory.
Valuation Snapshot
Premium industrial multiple reflecting AI-infrastructure growth rates rather than typical industrial-sector multiples — priced for the buildout to continue, with limited room for a capex pause.
Why This Conviction Rating
Medium conviction — a real, diversified AI-infrastructure thesis (exposure across all GPU vendors, not one) with the specific identified risk of hyperscaler capex-cycle sensitivity.
DDOG
Datadog Inc
↓
Business Description
Datadog provides cloud-native observability and monitoring software (infrastructure, application performance, log management, security) that helps engineering teams run and debug modern cloud applications.
Investment Thesis
An emerging-growth software compounder benefiting from both cloud migration (still not fully penetrated) and rising AI-workload observability needs, as companies need to monitor increasingly complex AI/agentic systems.
Bull Case
AI-workload monitoring (LLM observability, agent tracing) becomes a major new product category that Datadog leads, land-and-expand motion accelerates net-revenue-retention back toward historical highs, and the platform consolidates more point-tool spend.
Base Case
Steady 20%+ revenue growth continues as cloud migration and multi-product attach persist, without a major new AI-driven growth leg materializing as quickly as bulls hope.
Bear Case
Net-revenue-retention keeps decelerating as cloud-cost-optimization headwinds persist, competition from hyperscaler-native monitoring tools (AWS, Azure, GCP) and newer AI-observability startups intensifies, compressing growth and multiple simultaneously.
Key Risks
Customer cloud-cost-optimization cycles directly pressure usage-based revenue; intensifying competition in the observability category; high multiple leaves little room for a growth deceleration surprise.
Catalysts to Watch
Net-revenue-retention trend, new product (AI/LLM observability) attach rates, usage-based revenue reacceleration signals, competitive win/loss commentary.
Valuation Snapshot
Premium growth-software multiple (high revenue multiple) reflecting durable 20%+ growth expectations — a name priced for continued execution.
Why This Conviction Rating
High conviction: an established, profitable, category-leading software franchise with a genuine new growth vector (AI-workload observability) layered on top of the existing cloud-migration tailwind.
NOW
ServiceNow Inc
↓
Business Description
ServiceNow provides enterprise workflow automation software (IT service management, HR, customer service) and is pushing hard into agentic AI (AI agents that autonomously execute workflow tasks) as its next growth platform.
Investment Thesis
A durable enterprise-software platform with genuine, monetizing-today AI upside — ServiceNow’s AI agent products (Now Assist, AI Agent orchestration) are already being sold as incremental ACV, not just a future promise.
Bull Case
AI Agent/Now Assist products drive a meaningful incremental ACV growth leg on top of the core ITSM/HR/CSM platform, expanding both seat count and per-seat pricing as agentic workflows take over routine enterprise tasks.
Base Case
Core platform continues its consistent 20%+ subscription growth; AI products add a modest but real incremental contribution without yet being the dominant growth driver.
Bear Case
Enterprise AI-agent adoption proves slower and more fragmented than ServiceNow’s roadmap assumes, competitors (Microsoft, Salesforce, Workday) bundle competing agentic capabilities into their own suites, and the premium multiple compresses on any growth deceleration.
Key Risks
Intensifying competition from platform peers building their own AI-agent capabilities; premium valuation leaves limited room for disappointment; large-enterprise sales-cycle lumpiness.
Catalysts to Watch
Quarterly subscription revenue and cRPO growth, AI Agent/Now Assist ACV disclosure, large-deal (8-figure+) contract counts, competitive commentary versus Salesforce/Microsoft.
Valuation Snapshot
Premium enterprise-software multiple reflecting best-in-class execution and the AI-agent growth narrative — priced for continued category leadership.
Why This Conviction Rating
High conviction: an established competitive position in enterprise workflow software with a credible, already-monetizing AI growth vector, consistent with the book’s preference for AI beneficiaries with limited near-term uncertainty.
CAKE
The Cheesecake Factory
↓
Business Description
The Cheesecake Factory operates full-service casual dining restaurants under its namesake brand plus North Italia, Flower Child, and other emerging concepts.
Investment Thesis
A speculative-sleeve consumer/restaurant turnaround and multi-concept growth story, smaller and lower-conviction than the book’s core holdings.
Bull Case
Same-store sales momentum continues, the emerging concepts (North Italia, Flower Child) scale into meaningful earnings contributors, and margin expansion from cost discipline drives multiple expansion off a historically modest restaurant-sector valuation.
Base Case
Steady low-single-digit same-store sales growth and gradual emerging-concept unit growth continue without a major re-rating catalyst.
Bear Case
Consumer discretionary spending softens in a slowdown, labor/food-cost inflation pressures restaurant margins broadly, and emerging concepts fail to scale profitably fast enough to move the overall earnings needle.
Key Risks
Consumer discretionary spending sensitivity; labor and commodity (food) cost inflation; execution risk scaling newer restaurant concepts profitably; intensely competitive casual-dining category.
Catalysts to Watch
Quarterly same-store sales, emerging-concept unit growth and unit economics, labor/food cost commentary, consumer spending data broadly.
Valuation Snapshot
Modest restaurant-sector multiple that has already re-rated meaningfully off the position’s cost basis; further upside depends on emerging-concept execution rather than the core brand alone.
Why This Conviction Rating
Low conviction — a smaller, speculative-sleeve position where the thesis (multi-concept growth) is real but genuinely less certain than the book’s core holdings.
ARM
Arm Holdings plc
↓
Business Description
Arm Holdings licenses CPU architecture and chip designs used in the vast majority of the world’s mobile devices, and is expanding aggressively into data-center and AI-accelerator-adjacent silicon (compute subsystems, custom chip designs for hyperscalers).
Investment Thesis
A royalty-model, high-margin way to participate in AI-driven silicon growth beyond mobile — as hyperscalers design more custom AI chips, Arm’s architecture increasingly underpins those designs too, without Arm bearing the manufacturing or full R&D cost.
Bull Case
Data-center and AI-accelerator design wins (custom silicon for hyperscalers building their own chips) scale meaningfully, driving royalty-rate increases as Arm’s v9 architecture and compute-subsystem licensing gain share beyond its mobile base.
Base Case
Mobile royalty base continues growing steadily with v9 mix shift, while data-center/AI licensing adds a smaller but growing incremental contribution.
Bear Case
Data-center design-win momentum disappoints, RISC-V open-architecture competition erodes Arm’s long-term licensing position at the margin, and the stock’s premium multiple compresses on any growth deceleration in the core mobile royalty business.
Key Risks
Emerging RISC-V competition as an open-source architecture alternative; customer concentration risk (a small number of hyperscalers and chip designers drive incremental growth); premium valuation leaves little room for disappointment.
Catalysts to Watch
v9 architecture royalty-rate mix shift, data-center/compute-subsystem design win announcements, quarterly royalty revenue growth, any RISC-V competitive-share commentary.
Valuation Snapshot
Very high multiple reflecting royalty-model operating leverage and the AI-silicon growth narrative — one of the more richly valued names in the book’s semiconductor sleeve.
Why This Conviction Rating
Medium conviction: a real structural thesis (royalty exposure across the entire chip ecosystem) but genuine uncertainty around how fast data-center licensing actually scales relative to the rich multiple already priced in.
AVGO
Broadcom Inc
↓
Business Description
Broadcom designs custom AI ASICs (application-specific chips) for hyperscalers including Google (TPUs) and reportedly Meta, alongside networking silicon and the VMware enterprise software business acquired in 2023.
Investment Thesis
The premier custom-silicon partner for hyperscalers building their own AI chips as an alternative to Nvidia GPUs — a structurally different, higher-visibility (multi-year contracted) business model than merchant-silicon competitors.
Bull Case
Custom AI ASIC demand from Google, Meta, and additional hyperscaler customers keeps expanding the multi-year backlog, VMware cross-sell into the existing enterprise networking customer base continues, and Broadcom cements itself as the default custom-silicon partner beyond Nvidia.
Base Case
Existing hyperscaler ASIC programs (Google TPU, Meta) continue scaling as contracted, VMware integration proceeds steadily, and growth remains strong but somewhat more visible/contracted than the more speculative AI names in the book.
Bear Case
AI capex growth decelerates industry-wide, a disappointing outlook (as seen in a recent guidance reiteration that triggered a broader semis sell-off) signals the custom-ASIC growth story is maturing faster than expected, and customer concentration (a small number of hyperscaler ASIC programs) creates lumpiness.
Key Risks
Customer concentration in a handful of hyperscaler ASIC programs; any AI capex digestion pause hits the backlog-dependent growth story; VMware integration/customer-retention risk; extremely high market cap already prices in continued execution.
Catalysts to Watch
Quarterly AI/custom-silicon revenue growth and forward guidance, any new hyperscaler ASIC customer announcements, VMware ARR growth, hyperscaler capex commentary broadly.
Valuation Snapshot
Premium multiple reflecting the custom-AI-silicon growth narrative — recent guidance reiteration (rather than a raise) already triggered one sector-wide derating, underscoring how much growth is priced in.
Why This Conviction Rating
High conviction: an established, multi-year-contracted competitive position (Google TPU relationship alone is a differentiated moat) even after a recent guidance-driven pullback.
WYFI
WhiteFiber Inc
↓
Business Description
WhiteFiber provides AI cloud/GPU infrastructure and data-center colocation services, having pivoted from a crypto-mining-adjacent heritage into AI compute infrastructure.
Investment Thesis
A small, early-stage, high-risk/high-reward bet on continued AI compute infrastructure demand — smaller and more speculative than CRWV/NBIS, reflecting genuinely earlier-stage execution.
Bull Case
AI GPU cloud demand keeps outstripping hyperscaler supply, forcing more workloads to smaller/independent neocloud providers, and WhiteFiber scales its data-center footprint and customer base faster than the market expects from a name this size.
Base Case
Steady but modest scaling of AI infrastructure capacity and customer contracts, tracking the broader neocloud sector’s growth without a standout catalyst.
Bear Case
Larger, better-capitalized neocloud competitors (CoreWeave, Nebius, and hyperscalers themselves) out-compete for GPU allocation and customer contracts, and a smaller/earlier-stage balance sheet leaves less room for error through any AI-capex digestion period.
Key Risks
Smallest-cap, least-established name in the AI-infrastructure sleeve — genuinely higher execution and funding risk than CRWV or NBIS; heavy sensitivity to GPU allocation/availability from Nvidia; customer concentration risk.
Catalysts to Watch
New data-center capacity announcements, customer contract wins, any funding/balance-sheet updates, broader neocloud sector sentiment (the same Meta-cloud-entry story moving NBIS).
Valuation Snapshot
Small-cap, early-stage valuation with limited historical earnings to anchor a multiple — priced substantially on growth-narrative and sector-sentiment rather than current fundamentals.
Why This Conviction Rating
Low conviction — explicitly a newer, smaller, speculative-sleeve position where the AI-infrastructure thesis is directionally real but execution is genuinely less certain than the book’s larger neocloud names.
FTNT
Fortinet Inc
↓
Business Description
Fortinet sells network security appliances (firewalls built on its own custom ASICs), SD-WAN, and a broadening security operations/SASE platform.
Investment Thesis
Part of the book’s cybersecurity block (0.72 correlation with PANW, 0.71 with CRWD) — a well-capitalized, profitable network-security incumbent benefiting from the same platform-consolidation dynamics as its peers.
Bull Case
SASE/SD-WAN and security-operations attach accelerates on top of the core firewall franchise, Fortinet’s custom-ASIC cost advantage supports superior margins versus peers, and the firewall refresh cycle (customers upgrading aging hardware) provides a multi-year tailwind.
Base Case
Steady firewall-refresh-driven growth continues alongside modest SASE/SD-WAN attach, in line with recent historical growth rates.
Bear Case
Cloud-native, software-only security competitors erode the hardware-firewall business model’s relevance over time, and a slower enterprise refresh cycle (customers extending hardware life) compresses growth below current expectations.
Key Risks
Hardware-refresh-cycle dependency creates lumpiness; intensifying competition from cloud-native security vendors; customer concentration in enterprise/mid-market firewall replacement cycles.
Catalysts to Watch
Firewall/hardware refresh cycle data, SASE/SD-WAN attach and billings growth, competitive commentary versus Palo Alto and Zscaler.
Valuation Snapshot
Reasonable growth-software multiple relative to PANW/CRWD, reflecting its more hardware-centric (versus pure-software) revenue mix.
Why This Conviction Rating
High conviction: an established, profitable competitive position in a structurally growing, non-discretionary IT spending category.
CRWV
CoreWeave Inc
↓
Business Description
CoreWeave is an AI-focused cloud infrastructure provider (GPU cloud), heavily backed by and dependent on Nvidia GPU allocation, with major customer relationships including Microsoft and OpenAI.
Investment Thesis
A pure-play bet on continued GPU-cloud/neocloud demand outstripping hyperscaler capacity — genuinely early-stage and high-risk given customer concentration, but positioned at the center of the AI-compute buildout.
Bull Case
Microsoft/OpenAI and other hyperscaler-adjacent customer contracts keep expanding, CoreWeave secures durable multi-year GPU capacity commitments from Nvidia, and the business scales into a profitable, diversified neocloud leader.
Base Case
Existing large customer contracts (Microsoft, OpenAI) continue as contracted while new customer diversification proceeds slowly, keeping growth strong but concentration risk elevated.
Bear Case
Customer concentration proves the central vulnerability — any pullback or renegotiation from Microsoft or OpenAI hits results disproportionately, and broader AI-infrastructure sector sentiment (the same Meta-cloud-entry-driven derating hitting NBIS) compresses the multiple sector-wide.
Key Risks
Extreme customer concentration in a handful of large contracts; heavy capital intensity and debt used to fund GPU purchases; total dependence on Nvidia GPU allocation and pricing; highly correlated with NBIS (0.79) and the broader neocloud derating theme.
Catalysts to Watch
Customer contract renewals/expansions (especially Microsoft/OpenAI), new customer diversification announcements, GPU supply/allocation commentary from Nvidia, quarterly capacity utilization and backlog data.
Valuation Snapshot
Priced on forward growth and contracted backlog rather than current earnings — highly sensitive to sector-wide neocloud sentiment swings, as seen in the recent Meta-cloud-entry-driven sector derating.
Why This Conviction Rating
Low conviction — a newer, smaller-in-book-context, speculative-sleeve position where the AI-infrastructure thesis is real but customer-concentration risk is genuinely elevated.
NBIS
Nebius Group N.V.
↓
Business Description
Nebius Group is an AI cloud infrastructure company built from former Yandex assets, providing GPU compute and AI infrastructure services, listed on Nasdaq.
Investment Thesis
A direct neocloud/AI-infrastructure bet — recently pressured by Bloomberg reports that Meta may enter the cloud-infrastructure market, triggering a sector-wide derating of neoclouds rather than a Nebius-specific issue; July 29, 2026 earnings is the near-term catalyst to watch.
Bull Case
The Meta-cloud-entry threat proves overstated or years away from materially competing for Nebius’s customer base, Nebius keeps signing new GPU-cloud contracts, and the July 29 earnings report reassures the market on demand trends, driving a relief rally off the recent pullback.
Base Case
Nebius continues growing its GPU-cloud business steadily, the Meta-cloud-entry overhang persists as a sector-wide sentiment drag without a specific near-term impact on Nebius’s own contracts, and the stock stays volatile around AI-infrastructure sector news flow.
Bear Case
Meta (or another hyperscaler) does enter cloud infrastructure competitively, pressuring pricing and customer-acquisition economics for independent neoclouds broadly, and Nebius’s July 29 earnings disappoint on growth or margin trends, extending the recent derating.
Key Risks
Highest-profile name-specific risk in the book right now: the Meta-cloud-entry overhang; customer concentration; heavy capex intensity funding GPU purchases; extremely high correlation (0.79-1.00) with CRWV and NBIL, meaning the whole neocloud sleeve moves together on this one risk factor.
Catalysts to Watch
July 29, 2026 earnings (explicitly flagged as the key near-term catalyst), any further Meta cloud-infrastructure-entry news, GPU supply/Nvidia allocation commentary, new customer contract announcements.
Valuation Snapshot
Has derated meaningfully off the Meta-cloud-entry headlines; the July 29 print will be the first real test of whether the pullback was sector-sentiment noise or a legitimate repricing of the competitive threat.
Why This Conviction Rating
Medium conviction — a real, identified near-term catalyst (earnings) and a real, identified risk (Meta competitive entry) make this the most actively monitored name in the book heading into month-end.
SCHD
Schwab U.S. Dividend Equity ETF
↓
Business Description
Schwab U.S. Dividend Equity ETF tracks an index of U.S. companies with a history of consistent dividend payments and quality fundamentals screens (profitability, leverage).
Investment Thesis
The book’s dedicated income/quality-value ballast — does almost all the work behind the portfolio’s modest 0.24% weighted dividend yield, diversifying away from the growth/AI-infrastructure tilt running through the rest of the book.
Bull Case
Dividend-paying value/quality stocks outperform in a rotation away from high-multiple growth names, providing a genuine offset if the book’s AI/semis thesis cools.
Base Case
SCHD performs in line with its historical quality-dividend factor tilt, lagging in strong growth-led markets but providing ballast in choppier periods.
Bear Case
A prolonged growth-led market (exactly the environment this book is otherwise positioned for) leaves SCHD structurally lagging, though that’s the acknowledged tradeoff of holding it as a diversifier rather than a return driver.
Key Risks
Factor-timing risk — dividend/quality strategies structurally underperform in strong growth markets, which is somewhat in tension with the rest of the book’s growth tilt; limited by design.
Catalysts to Watch
Value-versus-growth factor rotation broadly; dividend-payer earnings season.
Valuation Snapshot
Trades at a below-market multiple by construction (the index screens for value/quality factors), consistent with its role as a diversifier.
Why This Conviction Rating
High conviction as a structural diversifier — its purpose is providing genuine ballast against the book’s dominant growth/AI tilt, not generating standalone alpha.
CRWD
CrowdStrike Holdings
↓
Business Description
CrowdStrike provides cloud-native endpoint security and threat detection (the Falcon platform), expanding into identity security, cloud security, and AI-driven security operations (Charlotte AI).
Investment Thesis
A best-in-class cybersecurity platform that recovered from a major 2024 global IT outage caused by a faulty software update — the thesis is that the incident, while reputationally costly, did not structurally impair the franchise’s competitive position.
Bull Case
Customer retention and new-logo growth fully recover from the 2024 outage’s reputational impact, Falcon platform module attach (identity, cloud security) accelerates, and Charlotte AI drives a genuine incremental AI-security growth vector.
Base Case
Steady module attach and retention continue at a healthy but not accelerating pace, with the outage impact fully in the rearview mirror but not forgotten by the largest enterprise customers.
Bear Case
Enterprise customers who experienced direct outage-related business disruption continue slow-walking renewals or diversifying vendors as a risk-management practice, and intensifying platform competition from Palo Alto/Microsoft compresses growth.
Key Risks
Residual reputational/trust risk from the 2024 outage among the largest enterprise accounts; intensifying platformization competition from Palo Alto and Microsoft; premium multiple leaves limited room for a growth disappointment.
Catalysts to Watch
Net-new ARR and module attach rates, large-customer retention commentary, Charlotte AI adoption data, any residual outage-related customer commentary.
Valuation Snapshot
Premium cybersecurity-software multiple, reflecting a full recovery narrative that the market has largely already priced in.
Why This Conviction Rating
High conviction: an established, category-leading competitive position where the 2024 outage is judged to be a reputational rather than structural/competitive impairment.
AAL
American Airlines Group
↓
Business Description
American Airlines is one of the largest U.S. legacy carriers, operating a global route network with significant debt taken on through the pandemic era.
Investment Thesis
A speculative-sleeve, deep-value airline turnaround bet — high operating leverage to any improvement in demand, pricing, or debt paydown, appropriately small given the sector’s structurally thin margins and cyclicality.
Bull Case
Premium/loyalty revenue mix keeps improving, fuel costs stay benign, debt paydown continues, and industry capacity discipline supports better pricing power than the airline sector has historically managed to sustain.
Base Case
Steady operational execution and gradual balance-sheet improvement continue without a major re-rating catalyst, consistent with the airline sector’s typical thin-margin, highly cyclical profile.
Bear Case
A demand slowdown (business or leisure travel pullback), fuel-cost spike, or renewed industry capacity war compresses already-thin margins, and the high debt load leaves limited room to absorb a downturn.
Key Risks
Extremely high operating leverage to fuel costs and travel demand; large debt load from the pandemic era; intensely competitive, historically value-destructive industry structure; macro/recession sensitivity.
Catalysts to Watch
Fuel cost trends, premium/loyalty revenue mix data, industry capacity (ASM) growth discipline, debt-paydown progress.
Valuation Snapshot
Trades at a low absolute multiple typical of the airline sector, reflecting its structurally thin margins and cyclicality rather than a specific mispricing.
Why This Conviction Rating
Low conviction — a smaller, speculative-sleeve position where the turnaround thesis is real (debt paydown, premium mix) but genuinely less certain given the sector’s historically poor capital-allocation track record.
VST
Vistra Corp
↓
Business Description
Vistra Corp is an integrated power generation and retail electricity company with a significant nuclear fleet (including Comanche Peak), benefiting from the same data-center power-demand thesis as Talen.
Investment Thesis
A second, more diversified power-for-AI-data-centers position alongside Talen — broader generation mix (nuclear, natural gas, some renewables) and a retail electricity business provide more diversification than a pure nuclear-only play.
Bull Case
More data-center power-purchase agreements get signed at attractive rates across Vistra’s broader generation fleet, nuclear capacity value keeps re-rating on scarcity, and the retail electricity business adds a stable earnings floor beneath the generation upside.
Base Case
Existing generation and retail businesses perform steadily, with incremental data-center power deals adding to but not transforming the earnings base.
Bear Case
Regulatory/interconnection delays slow new data-center power deals, and a broader AI-capex slowdown reduces the urgency behind the power-for-data-centers theme that has driven the stock’s re-rating.
Key Risks
Regulatory approval risk on power agreements; commodity (natural gas) price exposure in the non-nuclear generation fleet; interest-rate sensitivity as a capital-intensive utility-like business; correlation with TLN (0.81) means the whole power sleeve is exposed to the same thematic risk.
Catalysts to Watch
New data-center power agreement announcements, nuclear fleet operational updates, PJM/ERCOT capacity market pricing, natural gas price trends.
Valuation Snapshot
Re-rated meaningfully off the AI-power-demand thesis, similar to Talen — current multiple reflects real contracted cash flows with additional upside tied to further deal announcements.
Why This Conviction Rating
Medium conviction — the same real, contracted power-for-AI thesis as Talen, with somewhat more generation-mix diversification but the same regulatory and counterparty-concentration risks.
SYM
Symbotic Inc
↓
Business Description
Symbotic designs and deploys AI-powered warehouse automation robotics systems, with Walmart as its largest and most prominent customer and equity partner.
Investment Thesis
An emerging-growth automation/robotics bet on warehouse and supply-chain labor-cost pressure driving continued automation capex, anchored by a marquee Walmart relationship.
Bull Case
Walmart deployment continues scaling as contracted, additional large retail/logistics customers sign on, and Symbotic’s backlog converts to revenue faster than the market expects as automation ROI cases prove out.
Base Case
Walmart deployment proceeds on its existing contracted timeline while new customer diversification happens slowly, keeping growth steady but concentrated in one large relationship.
Bear Case
Walmart deployment pace slows or gets renegotiated, new customer wins remain scarce, and the business remains too concentrated in a single customer relationship to be considered a diversified automation platform.
Key Risks
Extreme customer concentration in Walmart; long, lumpy sales cycles for large warehouse automation deployments; execution risk on complex robotics system installations; unproven ability to diversify beyond the anchor customer.
Catalysts to Watch
Walmart deployment/backlog progress updates, any new large customer announcements, quarterly revenue recognition on system deployments.
Valuation Snapshot
Growth-stage valuation dependent on backlog conversion and customer diversification rather than current run-rate earnings.
Why This Conviction Rating
Low conviction — a newer, smaller, emerging-growth position where the automation thesis is real but customer concentration makes it genuinely less certain than the book’s core holdings.
RKLB
Rocket Lab USA
↓
Business Description
Rocket Lab operates the Electron small-satellite launch vehicle, is developing the larger Neutron rocket to compete for medium-lift launch contracts, and has expanded into satellite/space-systems manufacturing.
Investment Thesis
A speculative-sleeve bet on the broader commercialization of space launch and satellite infrastructure, positioned as a smaller, more diversified (multiple business lines) alternative to a pure-play launch bet.
Bull Case
Neutron reaches successful commercial launch and starts winning medium-lift contracts (including potential government/defense business), the space-systems/satellite-components business scales profitably, and Rocket Lab becomes a credible multi-line space company beyond Electron.
Base Case
Electron launch cadence continues steadily while Neutron development proceeds on a bumpy but forward-progressing timeline, without a major near-term re-rating catalyst.
Bear Case
Neutron development suffers meaningful delays or a launch failure, competitive pressure (SpaceX’s dominance, other new entrants) limits Rocket Lab’s addressable market, and the space-systems segment fails to reach meaningful profitability.
Key Risks
High execution risk on Neutron’s development and first launches; intense competition from SpaceX’s dominant, lower-cost position; cash-burn/funding risk during the development-heavy phase; small-cap volatility.
Catalysts to Watch
Neutron development milestones and first-launch timing, Electron launch cadence and contract wins, space-systems segment revenue/margin progress, any government/defense contract announcements.
Valuation Snapshot
Priced substantially on the Neutron growth narrative rather than current Electron-driven earnings — high sensitivity to any development-timeline news.
Why This Conviction Rating
Low conviction — a smaller, speculative-sleeve position where the multi-line space thesis is real but execution risk on Neutron specifically is genuinely high.
TEM
Tempus AI Inc
↓
Business Description
Tempus AI combines genomic sequencing, clinical data, and AI/machine-learning models to support precision medicine — primarily oncology diagnostics and data licensing to pharmaceutical companies and health systems.
Investment Thesis
A healthcare-AI emerging-growth bet on precision medicine and clinical-data monetization, the book’s smallest sector exposure and one of its more speculative positions given the company’s still-developing path to profitability.
Bull Case
Data-licensing deals with pharmaceutical companies scale meaningfully, diagnostic test volume growth accelerates, and the AI-model-driven precision-medicine platform reaches profitability faster than the market expects.
Base Case
Steady diagnostic test volume and data-licensing revenue growth continue, with the path to profitability remaining a multi-year story rather than a near-term inflection.
Bear Case
Reimbursement pressure on diagnostic testing, slower-than-expected pharma data-licensing deal flow, and continued cash burn without a clear near-term profitability inflection weigh on the stock.
Key Risks
Path to profitability remains unproven; healthcare reimbursement and regulatory risk; competitive pressure from other precision-medicine/genomics companies; execution risk scaling data-licensing partnerships.
Catalysts to Watch
Data-licensing deal announcements, diagnostic test volume growth, quarterly progress toward profitability, any regulatory/reimbursement developments.
Valuation Snapshot
Growth-stage valuation dependent on future data-licensing and diagnostic-volume growth rather than current earnings, consistent with its small, speculative-sleeve sizing.
Why This Conviction Rating
Low conviction — the book’s smallest position and one of its most speculative, reflecting genuine uncertainty around the path to profitability despite a real underlying healthcare-AI thesis.
NFLX
Netflix Inc
↓
Business Description
Netflix is the largest global subscription streaming service, monetizing through subscription tiers (including an ad-supported tier) and licensed/original content across film, TV, and increasingly live programming and gaming.
Investment Thesis
A core, established streaming leader — the password-sharing crackdown and ad-tier rollout successfully reaccelerated growth in recent years, and the thesis is whether that growth reacceleration is durable or was a one-time catch-up.
Bull Case
The ad-supported tier scales into a meaningful, high-margin revenue stream as ad-tech infrastructure matures, live programming (sports, events) further expands the addressable audience, and content spending efficiency keeps improving margins.
Base Case
Subscriber growth normalizes to a steadier, more moderate pace post-crackdown, with the ad tier and price increases providing continued but less dramatic revenue growth.
Bear Case
Subscriber growth decelerates meaningfully now that the password-sharing crackdown’s one-time benefit has been fully realized, content-cost inflation (competing for premium IP, live sports rights) pressures margins, and competitive intensity from Disney+, Amazon, and others caps pricing power.
Key Risks
Post-crackdown growth-normalization risk (the recent negative return reflects the market questioning growth durability); content-cost inflation, especially for live sports/events; intensifying streaming competition; consumer discretionary spending sensitivity.
Catalysts to Watch
Subscriber growth trends (though the company has reduced subscriber-count disclosure), ad-tier revenue and ARPU data, live-programming/sports-rights announcements, content spending guidance.
Valuation Snapshot
Premium media/streaming multiple that has come under pressure as the market questions whether recent growth reacceleration was durable or a one-time catch-up effect, consistent with the position’s negative return to date.
Why This Conviction Rating
Medium conviction: an established, category-defining competitive position, but with a specific identified risk (post-crackdown growth durability) that could compress the multiple further.
Not Owned
Watchlist coverage — 15 names, same depth, honestly labeled
These aren’t positions, so they don’t get a bull/base/bear breakdown or a conviction rating tied to real money — that would imply a level of commitment that isn’t there. What they get instead is the same real fundamentals and the same three honest questions every Watchlist deep dive already answers: why the name is on the list, why it isn’t in the book yet, and what would actually move me to buy. Content here is pulled directly from each name’s own watchlist page, not written separately for this view.
NVDA
NVIDIA Corp
↓
Fundamentals
P/E 29.95 · EV/EBITDA 29.43x · PEG 0.28. ~$4.8T (implied, 24.22B shares outstanding) market cap.
Why It's On the List
NVIDIA is the supplier every other AI-infrastructure name in this book ultimately depends on — GPUs from NVIDIA are the input CoreWeave, Nebius, and the hyperscalers are all racing to buy and deploy. I have heavy indirect AI-compute exposure through AMD, INTC, NBIS, NBIL, and CRWV, but no direct exposure to the company most people would name first when asked about this theme.
Why It's Not in the Book Yet
It's the consensus trade — the name every AI-infrastructure article already leads with — and I've been more interested in owning the less-obvious parts of the buildout (memory, foundry turnaround, power) where I think the market's read is less settled. Sizing is also a real question: adding NVDA on top of AMD and INTC risks doubling up on the same "AI chip demand" bet from a different angle rather than actually diversifying it.
What Would Move Me to Buy
A pullback that makes the entry point less consensus-crowded, or a clearer view on how much of my existing semiconductor sleeve is already effectively a NVDA bet through AMD/INTC's own GPU and foundry ambitions.
TSM
Taiwan Semiconductor
↓
Fundamentals
P/E 39.3 (trailing 33.33, forward 22.77) · EV/EBITDA 18.02x · PEG 1.55. $1.99T market cap.
Why It's On the List
Nearly every advanced chip in this book — NVIDIA's GPUs, AMD's data-center silicon, Apple's chips that Intel is now also chasing — is manufactured at TSMC or depends on TSMC-adjacent capacity. It's a way to own the AI-infrastructure buildout without betting on any single chip designer winning the architecture race.
Why It's Not in the Book Yet
Geopolitical risk around Taiwan is real and not something I can size away — it's a structural risk sitting underneath the whole position, not a normal business risk. I've been sitting with that rather than dismissing it.
What Would Move Me to Buy
A sizing framework I'm comfortable with for the geopolitical tail risk specifically, separate from the normal position-cap math in the IPS.
ASML
ASML Holding
↓
Fundamentals
P/E 58.18 (trailing 54.53) · PEG 2.69. €580.98B market cap.
Why It's On the List
ASML is the sole global supplier of extreme ultraviolet (EUV) lithography machines — the equipment TSMC and every other leading-edge foundry needs to make the chips this entire book is exposed to. It's arguably a more defensible monopoly than any single chip designer, since there's no second source.
Why It's Not in the Book Yet
European listing/ADR structure and a very high absolute share price make position sizing less intuitive for an account this size, and I haven't yet done the work to compare its valuation properly against the U.S.-listed names I already understand better.
What Would Move Me to Buy
Doing the valuation homework properly rather than avoiding it because the ticker is unfamiliar — that's a process gap on my end, not a real reason to stay away.
ORCL
Oracle Corp
↓
Fundamentals
P/E 24.08 · PEG 1.23. $407.87B market cap.
Why It's On the List
Oracle's cloud infrastructure business (OCI) has become a real AI-capacity story, landing large compute contracts that read a lot like the deals NBIS and CRWV sign — except backed by a company with decades of enterprise database cash flow underneath it, not a pure-play neocloud balance sheet.
Why It's Not in the Book Yet
I don't yet have a clean read on how much of Oracle's re-rating is durable OCI growth versus enthusiasm carrying over from the same AI-capex narrative lifting everything else. That's exactly the kind of "is the thesis real or is it the theme" question the correlation-matrix piece was about.
What Would Move Me to Buy
A quarter where OCI revenue growth and backlog disclosures give me enough to underwrite the capacity story independent of the broader AI-infrastructure sentiment.
APLD
Applied Digital Corp
↓
Fundamentals
P/E n/m (unprofitable, -57.77) · PEG N/A (not consistently profitable on EBITDA basis). ~$10.66B market cap.
Why It's On the List
Applied Digital is building out GPU-cloud and data-center capacity for AI workloads — the same basic business as NBIS and CRWV, at a much smaller market cap. If I believe in the AI-infrastructure buildout thesis broadly, a smaller, earlier-stage name is a legitimate way to express more of that conviction, not just a repeat of positions I already hold.
Why It's Not in the Book Yet
Smaller-cap infrastructure builders carry more execution and financing risk than NBIS or CRWV, and I'm already at a meaningful weight in this exact theme once NBIS, NBIL, and CRWV are added together. Adding a fourth name in the same cluster would concentrate risk further, not diversify it.
What Would Move Me to Buy
Trimming or a thesis break in one of my existing AI-infrastructure names that frees up room in that bucket without breaching the clustering risk already flagged in the July letter.
PLTR
Palantir Technologies
↓
Fundamentals
P/E 149.32 (forward 73.85) · EV/EBITDA 147.10x · PEG 1.58. $305.15B market cap.
Why It's On the List
Palantir has built genuinely sticky government and enterprise software around AI-driven data analysis, with renewal rates and contract expansion that suggest real product lock-in rather than just a good AI story.
Why It's Not in the Book Yet
The valuation is the whole conversation with this stock — it trades at a multiple that prices in years of continued hypergrowth, and I don't yet have a version of the bull case where I'm comfortable being wrong about the multiple, only the growth rate.
What Would Move Me to Buy
A meaningful multiple compression that isn't accompanied by a deceleration in the underlying government/commercial contract growth — i.e., the business staying intact while the price gets more reasonable.
CRM
Salesforce Inc
↓
Fundamentals
P/E 17.6 · EV/EBITDA 12.93x · PEG 0.32. $145.7B market cap.
Why It's On the List
Salesforce is betting its next growth leg on Agentforce, its AI-agent product layered onto the existing CRM install base. If it works, it's a way to own enterprise AI adoption through a company with an already-massive, sticky customer base rather than a newer, unproven platform.
Why It's Not in the Book Yet
"If it works" is doing a lot of the work in that sentence — AI-agent products across the software industry are still early, and I haven't seen enough usage/retention data yet to distinguish real adoption from a good demo.
What Would Move Me to Buy
Management disclosing real Agentforce usage and revenue-attach metrics, not just deal-count announcements.
UBER
Uber Technologies
↓
Fundamentals
P/E 17.68 (forward 22.22) · PEG 6.03. $151.51B market cap.
Why It's On the List
Uber is a genuine network-effect business that's turned the corner into real free cash flow generation, with autonomous-vehicle partnerships giving it optionality on a future where its own driver network becomes less central to the model — upside without needing to bet on any single AV company winning.
Why It's Not in the Book Yet
It's a different kind of business than almost everything else in this book — not an AI-infrastructure or semiconductor story — and I haven't done enough work on ride-share/delivery unit economics specifically to have real conviction versus just liking the story.
What Would Move Me to Buy
Actually doing that unit-economics homework, and a valuation that reflects the current cash-flow profile rather than the growth-stock multiple it traded at years ago.
COIN
Coinbase Global
↓
Fundamentals
P/E 54.25–57.60 · PEG 3.26. $43.45B market cap.
Why It's On the List
Coinbase is the primary regulated on-ramp for crypto in the U.S., and I already carry real BTC and ETH exposure through the market-snapshot ticker's proxy instruments — Coinbase would be a way to own the infrastructure layer around that exposure rather than just the assets themselves.
Why It's Not in the Book Yet
Coinbase's revenue is still heavily tied to trading volume and crypto-price volatility, which makes it a leveraged bet on crypto sentiment more than a diversifying position — closer to adding risk than spreading it.
What Would Move Me to Buy
Continued growth in Coinbase's non-trading revenue (subscriptions, custody, stablecoin infrastructure) reducing that correlation to raw trading volume.
IONQ
IonQ Inc
↓
Fundamentals
P/E 69.5x (recently turned positive) · PEG 0.55. $16.83B market cap.
Why It's On the List
Quantum computing is a genuine long-horizon technology bet, and IonQ is one of the more credible pure-play public companies in the space — the kind of small, speculative, real-optionality position the IPS's speculative sleeve is built for.
Why It's Not in the Book Yet
This is pre-revenue-at-scale, story-driven investing almost by definition — the speculative sleeve is already at 6.6% of the book (RKLB, WYFI, SPCX, TEM), and I don't think a fifth deeply speculative name is the right next move before that bucket's existing positions prove out.
What Would Move Me to Buy
Room opening up in the speculative sleeve, or a clear commercial milestone (a real paying enterprise customer, not just a research partnership) that changes this from a pure science bet to something closer to an early-revenue company.
MSFT
Microsoft Corp
↓
Fundamentals
P/E 23.16 (forward 21.00) · PEG 1.33. $2.77T market cap.
Why It's On the List
Azure is one of the three hyperscaler clouds and a direct beneficiary of enterprise AI adoption through Copilot and the OpenAI partnership, yet I have zero direct hyperscaler exposure despite AI infrastructure being my largest thematic bet in the book.
Why It's Not in the Book Yet
It's the single most "obvious" AI mega-cap trade, priced accordingly, and it would partially duplicate cloud-capex exposure I already carry indirectly through NBIS and CRWV, which sell capacity to hyperscalers like Microsoft rather than competing with them directly.
What Would Move Me to Buy
A valuation reset, or a specific catalyst that separates Azure's AI-driven growth from the broad mega-cap multiple the whole stock trades on.
AMZN
Amazon.com Inc
↓
Fundamentals
P/E 29.00 (trailing 31.21) · EV/EBITDA 18.72x · PEG 1.47. $2.60T market cap.
Why It's On the List
AWS is the largest cloud infrastructure provider and directly comparable to the power-purchase-agreement theme already in the book — AWS itself signed the 1.9 GW nuclear PPA with Talen Energy that's central to the TLN thesis, a real, direct link between this watchlist name and a position already held.
Why It's Not in the Book Yet
AWS is one segment of a much larger, slower-growing retail and logistics conglomerate, which dilutes the "clean" AI-infrastructure thesis relative to a pure-play name.
What Would Move Me to Buy
More visibility into AWS's standalone growth and margin trajectory, separated cleanly from the retail business's much lower-margin economics.
AAPL
Apple Inc
↓
Fundamentals
P/E 37.23 · EV/EBITDA 28.39x · PEG 2.68. ~$4.6T (implied, ~14.69B shares outstanding) market cap.
Why It's On the List
Apple is now directly tied to the Intel foundry turnaround thesis — the June 2026 Apple-Intel manufacturing deal — that's already the single best-performing position in this book. Owning Apple would be a way to play the other side of that same real relationship.
Why It's Not in the Book Yet
Apple's own AI story (Apple Intelligence) has been the market's biggest source of skepticism about the company's execution, and the stock trades on much more than the foundry relationship alone.
What Would Move Me to Buy
Clearer evidence Apple's own AI strategy is working, not just its manufacturing diversification into domestic foundry capacity.
SMCI
Super Micro Computer
↓
Fundamentals
P/E 16.10–20.13 (forward 10.38) · PEG 0.27. $19.86B market cap.
Why It's On the List
Super Micro builds the physical servers that house the GPUs everyone else in this book is chasing exposure to — a picks-and-shovels-of-picks-and-shovels business one level removed from NVIDIA itself.
Why It's Not in the Book Yet
The company has a real history of accounting and governance controversy, and margins have been volatile and thin (3.2% in a recent quarter) even as revenue grows at a triple-digit annual pace. This is a name where "will the business model hold together" is as important a question as the growth story.
What Would Move Me to Buy
A longer track record of clean, timely financial reporting alongside the growth — governance risk resolving, not just the numbers looking good.
MRVL
Marvell Technology
↓
Fundamentals
P/E 83.15 (trailing 84.73, forward 54.07) · PEG 2.63. $214.58B market cap.
Why It's On the List
Marvell designs custom AI silicon (ASICs) for hyperscalers as an alternative to buying off-the-shelf NVIDIA GPUs — a different, complementary bet on the same AI-buildout theme than anything else currently in the book.
Why It's Not in the Book Yet
The stock already trades at a very rich multiple (80x+ trailing earnings) pricing in a lot of custom-silicon success, and I don't yet have high conviction on which hyperscaler AI-chip programs will actually scale versus stay pilot-sized.
What Would Move Me to Buy
More disclosed detail on customer concentration and program-specific revenue, since custom-silicon deals are lumpy and customer-concentrated by nature — I want to see the backlog, not just the narrative.