Wait for proof. No position.

1. One causal thesis

Alphabet's AI capex converts into GAAP D&A faster than consensus EPS incorporates it; if EBITDA growth does not absorb the charge, quarterly prints force FY2027-28 EPS cuts and a multiple reset.

D&A is an earnings-revision catalyst, not the direct DCF downside. In FCFF it is added back; the cash leaves through capex. A rising D&A line is therefore not confirmation by itself. The signal is the portion that remains unoffset in EBIT, reaches EPS, changes the analyst distribution, and then changes the multiple.

2. Implementation gate

Robinhood read-only market data shows 30-day GOOGL ADV of 31.1m shares, approximately $11.0bn at spot. That supports ordinary underlying liquidity; it does not establish short locate, carry, squeeze capacity, option execution, hedge basis, or an exit plan for a specific book.

3. What is priced in—and what is not proven

At spot, the frozen FMP FY2027 EPS feed implies 23.6x. The base EPS thought experiment implies 24.3x. Those are observable arithmetic, not proof that the market explicitly believes any single WACC, terminal margin, or revenue CAGR.

4. Placed-in-service D&A

The rebuilt schedule starts from reported Q2 D&A of $7.5bn and adds depreciation only after explicit commissioning lags. Asset mix, lives and lags are blue analyst assumptions in the workbook. Historical/opening D&A is separated from forecast-vintage D&A; the memo no longer claims all modeled depreciation comes from forecast capex.

FY2026 capex is $200bn, the midpoint of the $195-205bn guide. FY2027 is an explicit $230bn assumption, +15%, consistent with management's direction that spending would increase significantly. There is no kill criterion that the base case trips on day one.

5. Illustrative scenario skew

Scenario values are EPS × multiple and reflect the actual causal chain. The DCF is a cross-check, not the target-setting mechanism. Expected value is shown because it is formula-linked, but it does not clear the trade gate and does not produce a position size.

6. Quarterly catalyst bridge

Before each print freeze quarterly EBITDA, EBIT, implied D&A, tax, below-line income, diluted shares and EPS; record actual D&A, EBITDA surprise/offset, FY2027 EPS before/after, stock reaction and post-print multiple.

7. Conditional action rules

8. Valuation and capital-structure controls

• Dated stub: the June 30 bridge is rolled to August 7 with an explicitly estimated interim FCFF; only August 8 onward is discounted. Reported H1 is not counted again.
• Terminal normalization: next-period NOPAT is reduced by g / terminal ROIC reinvestment before applying Gordon growth. Base WACC is 8.81%, built from a 4.65% Treasury rate, 0.95 beta, 4.50% ERP and market-value capital weights.
• Shares: 12.416bn point diluted shares = 12.230bn point-in-time common + ordinary dilution + minimum mandatory-convertible dilution. Preferred value is not deducted again.
• Financing language: the $40bn ATM is registered capacity available but unsold at June 30, not charter-authorized capital. Unsupported aggregate financing claims have been removed; completed common and preferred issuance is separated from preliminary or undrawn capacity.
• Multiple: enterprise earnings use EV/NOPAT; there is no price/NOPAT-per-share "economic P/E."

Appendix A. Source ledger

Appendix B. Known limitations

• No broker-level quarterly consensus ledger or revision history.
• No live borrow/carry, short-interest/crowding, option-chain execution or hedge analysis.
• Capex asset mix, useful lives, commissioning lags, beta, ERP, terminal ROIC, scenario probabilities and exit multiples are analyst assumptions.
• The opening-vintage D&A run-rate is held flat through the explicit horizon because remaining lives by historical vintage are not disclosed.