Phase Space AI

Alphabet Valuation Analysis 2026-07-27

Alphabet [GOOGL]

Alphabet Inc. [GOOGL] — Valuation Analysis

Task 3 · investment-memo skill v1.4.0 · 2026-07-27 · spot $326.56

Model: Alphabet_Financial_Model_2026-07-27.xlsx. Every figure below was read back from the live workbook in Excel; formulas were not trusted because openpyxl writes them without evaluating them.

Read Section 0 first. Alphabet's GAAP earnings are contaminated by mark-to-market on non-marketable equity securities to a degree that breaks conventional valuation. This section is not a caveat; it is a prerequisite.

---

0. The GAAP contamination problem, and how this memo handles it

Q2-2026 other income was a net gain of $98.0bn, "primarily the result of net unrealized gains on our equity securities" (8-K Ex-99.1). GAAP EPS was $9.11, up 294%, on operating income up 30%.

Consequences, stated plainly: 1. Any headline P/E on GOOGL is meaningless. 2. Any earnings-surprise (SUE) or estimate-revision series — which the Factor Scorecard would normally use — inherits the distortion. This is disclosed in Section 4 rather than silently computed. 3. The published consensus of $14.22 (2026) / $14.68 (2027) is on a basis that includes some level of other income and is therefore not cleanly comparable to an operating EPS. The house model reports both, and the comparison in Section 3.2 is explicitly labelled as not like-for-like.

This memo therefore values Alphabet on operating income and unlevered cash flow. EV/EBIT is the primary multiple.

---

1. DCF — and why its point estimate should not be used

Sensitivity — and this is the point:

$144 to $312 — a 2.2x range across defensible parameters. The DCF point estimate is not usable on this name, and pretending otherwise would be the most dishonest thing in this memo. Three compounding reasons: 1. Unlevered FCF is negative in FY2026E (−$40.1bn) and FY2027E (−$21.0bn), so almost all value sits beyond the explicit period. 2. 82.2% terminal-value weight. 3. The regressed beta of 1.42 — genuinely computed, genuinely what the stock has done, and inflated by large idiosyncratic moves including −7.1% on 23 July. A $4tn company with a 1.42 beta is itself a finding, but using it mechanically produces an 11.6% discount rate that few would apply to Alphabet's cash flows.

Reported DCF range: $160–$310, with no point estimate offered. The scenario table and EV/EBIT comps carry the weight.

1.1 Free cash flow — the number that moved the stock

The FY2026E working-capital line is calibrated to the H1-2026 actual CFO of $84,859m and labelled as such in the workbook — without that calibration the model would have produced a CFO ~$40bn above what Alphabet actually reported in the first half.

Alphabet is modelled as free-cash-flow negative for full-year 2026 and barely positive in 2027. H1-2026 actual FCF was $4,261m and Q2-2026 alone was −$5,855m. This is not a forecast; it is an extrapolation of a disclosed trajectory against guided capex.

1.2 Depreciation — computed from the placed-in-service roll-forward

Seeded on the 10-Q R64 disclosure: in-service technical infrastructure $247,177m, office/other $57,133m, and assets not yet in service $122,814m (from $78,592m at Dec-2025). The not-in-service pool is modelled as transferring into service explicitly.

This is the mechanical reason the Street models 2027 EPS growth at +3.2%. The depreciation wave is in consensus, not missing from it.

---

2. Comps

Implied values: peer-median EV/EBIT × house FY2027E operating income, bridged to equity = $384.06. Peer-median P/E × house FY2027E operating EPS ($13.97) = $324.03.

Sourcing honesty: GOOGL and MSFT FY+1 P/E use sourced consensus. AMZN/META/ORCL/NVDA multiples are estimates, flagged as such. As with Microsoft, the comp set is the AI-capex cohort re-rating together — any relative conclusion is a sector-regime statement, not a stock-specific one.

---

3. Street & positioning

3.1 Consensus

Alpha Vantage was NOT called (quota exhausted); consensus sourced from public aggregators per references/consensus-bridge.md.

3.2 House vs Street — the Gate 2A test

Why this is an artifact and not a variant, spelled out because it would be very easy to sell as one. The house series excludes unrealised equity marks in both years. The Street's 2026 base of $14.22 sits on a year containing $127.6bn of such marks in the first half alone; its 2027 estimate necessarily contains none, because nobody forecasts marks. So consensus growth of +3.2% is measured from an inflated base, and house growth of +17.2% from a clean one. Two different bases, not two different views.

Adjusting crudely for that, the house operating trajectory and the Street's implied operating trajectory are close. There is no demonstrable earnings variant on either 2026 or 2027.

3.3 The consensus bridge — decomposing a 32% gap (required)

The Street target of ~$430 is 31.7% above spot.

The gap is 100% multiple and ~0% numbers. The Street is paying ~30x forward operating earnings for a business whose free cash flow is negative this year; this memo's base case pays 24x. Neither is provably right.

The real risk to the house view if the Street is right: Cloud holds above 60% growth with a 36%+ margin, 2027 capex comes in at the low end, free cash flow inflects positive in 2027 rather than 2028, and the market re-rates the whole capex programme from "cost" to "moat." That path is worth 30–50% from here. Given that Cloud has now accelerated for two consecutive quarters and expanded margin for six, the bull case is a live possibility, not a straw man — and this memo has no evidence against it, only an unwillingness to pay 30x for it before the cash flow turns. That is scepticism, not a variant, and it is labelled as such per references/trade-construction.md.

3.4 Short interest, ownership, options

• Short interest: negligible; days-to-cover ~1.8 on the most recent figure located (dated, and therefore stale). The current FINRA bi-monthly figure could not be sourced and is not fabricated. No crowding, no squeeze risk.
• Institutional ownership: ~80%, index-dominated. Passive capital. Dual-class structure means founder control is insulated from the June-2026 issuance.
• Options market — the mirror image of Microsoft. From Alpaca snapshots, 2026-07-27:

Realised: 31.4% (1-year), 38.8% (3-month). Alphabet's implied volatility sits BELOW its trailing realised volatility across the curve — because its catalyst passed on 22 July and the event premium has already collapsed. Skew is close to flat (Sep $325 put 31.2% vs $325 call 31.6%; Jan-27 $300 put 33.4% vs $350 call 33.4%). Options are, unusually, not expensive here. That materially changes the vehicle analysis relative to Microsoft, and it is the one genuinely actionable asymmetry between the two names.

---

4. Factor & Anomaly Scorecard (REQUIRED)

Direction assessed for a long. All inputs computed from Alpaca SIP bars and EDGAR XBRL.

Synthesis — the factors split, and two readings are actively unreliable. Profitability (0.404) and accruals (−0.055) are excellent and momentum (+102%) is positive: on those three, GOOGL looks like a quality-momentum long. Against that, +32.2% asset growth is one of the most negative single factor readings available, and the trend filter is balanced on a knife-edge at +0.8%.

Two readings must be discounted rather than counted: the revision signal is contaminated by equity marks, and PEAD is inverted (a 2.6% revenue beat produced a 7.1% decline). Honesty requires saying that the two expectations-based factors, which would normally be the most decision-relevant, cannot be read on this name right now. That is not a bullish or bearish conclusion; it is a statement that two of the eleven instruments are broken.

---

5. Regime-Change Test (references/regime-change-test.md)

Tiered evidence

Verdict: the regime-change claim PASSES, primarily on Tier 2 rather than Tier 1 — the opposite balance to Microsoft. The technical/physical position is stronger; the contractual evidence is weaker because it is undisclosed, not because it is absent.

Duration matching (Principle 3)

Because duration is undisclosed, the model assumes a 20% 12-month conversion rate — anchored on Microsoft's disclosed 25% and adjusted for Alphabet's larger cover — and this is labelled an estimate, not a source, in the workbook. The backlog is modelled at contracted terms through FY2029 and then allowed to decline ($660bn → $650bn → $630bn → $610bn), with no renewal at similar terms assumed.

The sensitivity of this assumption is the honest headline: at a 20% conversion, the backlog covers 98.7% of modelled 2026 Cloud revenue. That figure is almost tautological (5.2x cover × 20% ≈ 100%) and it means the answer to "how contracted is Alphabet's cloud growth?" is driven entirely by a parameter the company does not disclose. That is the finding.

Two-tranche output — contracted share of Cloud revenue

Declining, as Principle 4 requires.

Principle 5 shape check

The signature is WEAKER than Microsoft's, and the reason is instructive. The bear case is −44.6%, versus Microsoft's −35.2%, on a company with a larger nominal backlog. The backlog does truncate the left tail somewhat — but because its duration is undisclosed, it cannot be relied on to support a specific year, so the bear case is not floored the way Microsoft's is. A $519.5bn backlog with no disclosed duration provides less downside protection than a $627bn backlog with 25% disclosed as converting within twelve months. That is the practical value of disclosure, and it is scored against Alphabet here rather than glossed.

Base rate being overridden: essentially no company above $300bn of revenue has compounded >15% for five years. The model decays total growth from 24.3% (2026E) to 7.3% (2032E) rather than extrapolating. Two overrides are named: Cloud revenue through FY2029 (on the backlog) and the Cloud operating-margin path, which rests on six consecutive observed quarters of sequential expansion — strong evidence for the near years, assumption in the outer ones.

---

6. Valuation summary

The probability-weighted value is $323.91 against a spot of $326.56 — a 0.8% gap. Scenario-weighted expected return is −0.8% gross, −5.8% after the 4.7% cash hurdle and frictions.

The distribution is extraordinarily wide for a $4tn company — $181 to $494, a 2.7x spread — and that width is itself the most honest output of this analysis. It reflects a genuine, unresolvable-today disagreement about whether $457bn of two-year capital expenditure is a moat or a mistake.