task-gdpval-2

PHASE 1: Interpret thesis and ingest inputs Purpose: Read the rules that govern every downstream calculation and survey all files before modelling. Rationale: We read and pin the thesis rules first because the market screen, the Mega-Investment Exclusion, the forecast assumptions, and the Tax Uniformity rule drive every later number; if we modelled before fixing them, each workbook could apply a different convention. Steps: 1.1 We read Fund_investment_thesis.docx and record the two market-screen thresholds (>$1T addressable within five years AND CAGR >=10%), the Mega-Investment Exclusion rule (strip the after-tax P&L of any single external investment over $1B), the forecast assumptions (5% QoQ off latest actual exclusion EPS, private stakes valued around Jun 1, public prices +2.5%/quarter, annual EPS = 4x quarterly), and the Tax Uniformity rule (apply each firm's last-quarter ETR to all streams). [rubric 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7] 1.2 We read the prompt and enumerate the artifacts we owe: the investment selection, a 13-F report, a gain model, revenue waterfalls, a private-investment EPS model, and a summary deck, so we scope all six before building any one. 1.3 We load the current-holdings workbook, the May-2026 price deck, the market and earnings PDFs, the company reports (10Qs), and the estimates image into a single source register, so every input has a known provenance. 1.4 We create one assumptions register and have all models draw thesis rules, tax rates, valuation dates, and price inputs from it, so the same conventions flow consistently into every workbook. [output: assumption tabs in the Excel models] PHASE 2: Screen markets and select investments Purpose: Convert the thesis screen into a defensible recommendation of two Magnificent Seven companies. Rationale: We let the thesis thresholds, not intuition, drive selection, and we document the market-to-segment mapping so each pick is justified by the screen rather than asserted. Steps: 2.1 We apply both thresholds (>$1T within five years and CAGR >=10%) to the market data and isolate exactly the qualifying markets: e-commerce and data centers. [rubric 2.1, 2.2, 2.3] 2.2 We map Amazon's Online stores to e-commerce and AWS to data centers, and recommend Amazon for the 4% stake. [rubric 2.4, 2.6] 2.3 We map Microsoft's Azure and server/cloud exposure to data centers, recommend Microsoft for the 2.5% stake, and explain choosing it over Alphabet on playing in a faster growing enterprise SaaS market (vs. Ads for Alphabet). [rubric 2.5, 2.7] PHASE 3: Construct post-investment 13-F Purpose: Translate the investment mandate into a regulator-style holdings report and reconcile the funding from cash. Rationale: We read the 4% and 2.5% mandates as percentages of each company's shares outstanding, not of fund NAV, because that is what determines share counts, position cost, and the cash drawdown that must reconcile. Steps: 3.1 We interpret the stakes as percentages of shares outstanding and compute share counts from reported shares (SEC filings): Amazon about 430.0 million shares and Microsoft about 185.7 million shares. [rubric 3.1, 3.2] 3.2 We value the new positions at the May-2026 average build prices of $268.74 for Amazon and $417.59 for Microsoft. [rubric 3.3] 3.3 We set each position value as shares times price (Amazon about $115.6B, Microsoft about $77.6B) and confirm internal consistency. [rubric 3.4, 3.5] 3.4 We add the two holdings to the legacy 13-F table, produce the information table with the required fields and a correct portfolio total, and reduce cash from $250.0B to about $56.9B so the post-investment balance sheet reconciles. [rubric 3.6, 3.7] [output: UniCorn_Fund_Form_13F.docx] PHASE 4: Gain model to Q2 2026 earnings Purpose: Forecast each new position's value using normalized EPS, then calculate percent and dollar gains against the May build price. Rationale: We forecast off Mega-Investment-Exclusion EPS rather than reported EPS so that one-off investment marks do not distort the operating earnings power we capitalize, and we apply each firm's own last-quarter ETR for tax uniformity. Steps: 4.1 We derive latest-actual exclusion EPS by removing qualifying marks over $1B: for Amazon we remove the Anthropic +$16,807mm gain and the Rivian -$889mm loss (base exclusion EPS ~ $1.67), and for Microsoft we treat OpenAI as equity-method already excluded from the non-GAAP base (base exclusion EPS ~ $4.27). [rubric 4.1, 4.2] 4.2 We apply each firm's uniform last-quarter effective tax rate to the excluded items (Amazon 24.0%, Microsoft 19.2%) rather than a generic or spot rate. [rubric 4.3, 4.4] 4.3 We grow exclusion EPS 5% QoQ to the Q2 2026 estimate (Amazon ~ $1.75/qtr, Microsoft ~ $4.48/qtr) and annualize as 4x the quarterly figure (Amazon ~ $7.01, Microsoft ~ $17.93). [rubric 4.5, 4.6, 4.7, 4.8] 4.4 We apply the median Magnificent Seven forward P/E of 25x from a tier-1 broker (Goldman Sachs report on AI), setting forecast price = forward P/E times annualized exclusion EPS (Amazon ~ $175.33/sh, Microsoft ~ $448.18/sh). [rubric 4.9, 4.10, 4.11] 4.5 We calculate gain per share, gain percentage, dollar gain, and forecast position value versus the May average build cost. [rubric 4.12, 4.13, 4.14] [output: UniCorn_Gain_Model.xlsx] 4.6 We confirm the workbook is fully formula-driven and recalculates without errors, tying to Amazon about -34.8% / -$40.2B, Microsoft about +7.3% / +$5.7B, and combined net about -$34.5B. [rubric 4.15] PHASE 5: Revenue waterfall Purpose: Explain each selected company's Q1 2026 versus Q1 2025 revenue growth by top business lines and a remainder bucket. Rationale: We disaggregate growth into the top three lines plus a Rest bucket and preserve the sign on declining lines, so the bridge reconciles start to end rather than masking offsetting moves. Steps: 5.1 We pull disaggregated revenue for the comparable periods (calendar Q1 2026 versus Q1 2025) from each company's report (SEC filings). [rubric 5.1] 5.2 For Amazon we identify Online stores, third-party seller services, and AWS as the top three, with everything else in Rest; for Microsoft we identify Server products and cloud services, Microsoft 365 Commercial, and Gaming, with everything else in Rest. [rubric 5.2, 5.3] 5.3 We compute the YoY dollar and percentage change for each line with correct signs, keeping Microsoft Gaming as a decline rather than forcing all lines positive. [rubric 5.4] 5.4 We tie Amazon from $155.7B to $181.5B and Microsoft from $70.1B to $82.9B, ensuring top-three plus Rest equals total growth. [rubric 5.5, 5.6] 5.5 We build a waterfall/bridge chart per company that distinguishes increases, decreases, and endpoint totals, and confirm the workbook is formula-driven and recalculates without errors. [rubric 5.7] [output: UniCorn_Summary_Deck.pdf] PHASE 6: Private-investment EPS impact Purpose: Estimate whether qualifying private investments over $1B distort Q2 2026 EPS for Amazon or Microsoft. Rationale: We test this as a wildcard because the same marks we excluded for forecasting could still move reported EPS, and the accounting treatment (net income vs OCI vs equity method) determines whether they do. Steps: 6.1 We identify the qualifying private stakes over $1B as of June 1, 2026: Anthropic and OpenAI for Amazon, OpenAI for Microsoft. [rubric 6.1] 6.2 We record pinned June 1, 2026 valuation assumptions from credible public sources with retrieval dates noted, including Anthropic around $965B post-money and OpenAI around $852B. [rubric 6.2] 6.3 We apply the correct accounting treatment: Amazon preferred-investment marks flow through net income, Amazon convertible notes flow through OCI until conversion, and OpenAI is treated under equity-method logic. [rubric 6.3] 6.4 We compute the Amazon/Anthropic net-income-recognized after-tax EPS impact at about +$3.06 per share, and flag the full-mark upper bound at about +$7.10 per share. [rubric 6.4, 6.5] [output: UniCorn_Private_Investment_EPS.xlsx] 6.5 We conclude approximately zero incremental Q2 impact for the flat-valuation stakes (Amazon/OpenAI and Microsoft/OpenAI), with the Microsoft equity-method impact near zero, and confirm the workbook is formula-driven with sourced, auditable input cells. [rubric 6.6, 6.7] PHASE 7: Summary deck Purpose: Synthesize the recommendation, position sizing, valuation, revenue bridges, and EPS wildcard into a decision-ready deck. Rationale: We write the deck as a set of conclusions supported by the workbooks, stating each conclusion before its support, so a decision-maker reads the recommendation rather than a spreadsheet walkthrough. Steps: 7.1 We build a cover, exactly four content slides, and an appendix: thesis and selection; positions and valuation; revenue engines; and the private-AI-stake wildcard. [rubric 7.1] 7.2 We make every displayed number reconcile to a workbook or source and state the conclusion before the support. [rubric 7.2] 7.3 We append methodology, key assumptions, and sources. [rubric 7.3] 7.4 We run layout QA for overflow, overlap, and chart legibility, showing revenue as a waterfall consistent with Phase 5. [rubric 7.4] [output: UniCorn_Summary_Deck.pdf]