Revenue is compounding 24% and Cloud 82% against a $514B backlog — yet trailing free cash flow fell 20% to $53.3B and consensus has 2026 FCF near $8.6B as capex heads to roughly $200B. The market is pricing one question: does the watts-and-silicon wave convert into Cloud and TPU profit before depreciation arrives? Six analyst lenses, three scenarios, four time horizons — with the next test on Oct 28.
Gray line = Alphabet Class C actual closes into today ($236.69 52-week low on Oct 10 ’25 → $404.47 intraday high in mid-May ’26 → $347.86 on Oct 9 ’26); built only from dated month-end closes and the two intraday extremes, so intermediate wiggles are not drawn. Colored paths = synthesized scenario midpoints, each EPS × exit multiple, probability-weighted 30% bear · 45% base · 25% bull (bear-tilted: negative-FCF capex cycle, unrealized-gain-inflated GAAP EPS). Linear price axis from $100; history on a month scale, forecast on a year scale. Open circle = Street 12-month consensus $420.36 (+21%; 62 analysts, range $340–$485 on the Class C page).
Those probabilities are a judgment call — so make them yours. Drag to set how likely the bear and bull cases are (base takes the remainder); the blended target below, the dotted line on the chart, and the prob-weighted row of the scenario cards all update live.
The same fundamentals support wildly different conclusions depending on which framework you trust. Each lens is a synthesized perspective worked through independently, with its own 12-month target. Panel mean $352 (range $260–$425) — below the Street’s $420.
Revenue +24% on a ~$480B run-rate, Cloud +82% against a $514B backlog (just over half converting inside 24 months), and a 34% operating margin even as capex doubled. Gemini is at 950M monthly users and AI Mode past 1B — queries are at records, so AI is expanding Search rather than eating it. $15+ of 2027 earnings power earns a premium multiple.
Trailing FCF of $53.3B is down 20% and equals a 1.2% yield on $4.28T; consensus FY26 FCF is only ~$8.6B. The 17.5x headline P/E rests on ~$136B of unrealized gains — clean TTM EPS is ~$11.2, so ~31x. Debt rose to $120.8B (net borrowing ~$70B TTM). A great business priced as though capex were free.
$195–205B of 2026 capex turns into depreciation for years. A 35.6% Cloud margin may be flattered by early TPU system sales; if D&A outgrows revenue by 2028 the operating leverage reverses. Meanwhile agentic assistants can route around the Search click that funds the whole machine.
Own chips (TPU, now sold externally), own models (Gemini), own distribution (Android, Chrome, YouTube, Search) and own cash flows to fund the build. Few rivals run training, serving and monetization inside one P&L, and TPU system sales add a merchant-silicon option on top. Moat is cost per token.
$347.86 sits above the 50-day ($343.39) and 200-day ($338.38) averages with RSI ~56 and +42% over 52 weeks — but 14% below the May high that never held. Closes have ranged ~$326–$377 since late July while 2027E EPS revisions drift up ($15.21). Neutral trend, positive estimate momentum.
The Sept 2 ad-tech ruling spared an AdX breakup (behavioral decree; joint proposed judgment filed Oct 2), but the Search case is on cross-appeal and the EC’s €3B ad-tech fine is under appeal. Finland ordered work halted on two data centers. Power, permitting and policy now pace the capex curve as much as demand does.
What the sell-side expects over the next year. Bars are sorted low to high; the dashed line is today’s $347.86 — every target shown sits above it, and none of the 62 covering analysts rates the stock Sell.
Selected sell-side 12-month targets as last reported: Citizens $515 (Oct 9), TD Cowen $485 and Needham $450 (Oct 8), BMO $465 and RBC $425 (Oct 9), Wells Fargo $417 (Oct 6; 27.5x 2027E EPS), Morgan Stanley and Oppenheimer $400, and the two Hold-rated names — UBS $379 and DA Davidson $350 — as of their Jul 23 cuts. Street rating mix (S&P Global, Oct 2026): 44 Strong Buy, 13 Buy, 5 Hold, 0 Sell across 62 analysts. Sources conflict: the Class C page shows a $420.36 mean, $425 median and $485 high; the Class A page shows $429.47, $430 and $515. This report uses the Class C figures for the GOOG ticker. Firm targets are as reported by aggregators and not independently verified.
Synthesized scenario midpoints (mid-year), each computed as clean EPS × exit multiple. Returns shown vs. today’s $347.86. These are illustrative frameworks, not predictions — the outcome hinges on whether AI capex pays back in Cloud, TPU and Search profit.
Where the money actually goes. The bull and the bear theses both live in the gap between the capex bar and the free-cash-flow bar.
Revenue compounds ~15–24% a year while capex goes from $32B (2023) to a guided ~$200B (2026E, midpoint of the $195–205B range raised on Jul 22) — and free cash flow, steady near $70B for three years, collapses to a consensus ~$8.6B. That squeeze is the bear’s whole argument; the $514B backlog is the bull’s answer. Total debt (slate) has quadrupled from $30B to $120.8B as of Jun 30, 2026 (latest reported; plotted in the 2026E slot), but cash and investments of $242.5B leave a net-cash position of $121.7B. 2023–25 are reported (S&P Global via StockAnalysis); 2026E revenue and FCF are consensus; debt is gross; FCF = operating cash flow less capex.
The price targets aren’t pulled from the air — each is an EPS estimate times an exit multiple. Here is the earnings ladder the base case is built on.
Read this chart carefully: gray bars are reported GAAP EPS, and 2025’s $10.81 already includes some equity-investment gains, while 2026E–2027E ($12.11, $15.21) are Street consensus on a clean basis that excludes them — so the apparent 12% step into 2026 understates underlying growth. Reported 2026 GAAP EPS will look far higher: Q1’26 ($5.11) and Q2’26 ($9.11) carried $37.7B and $98.0B of other income, mostly unrealized gains on equity stakes. 2028E–2031E are this desk’s extension (growth easing from ~20% to ~12%). The base case’s ~$27.5 of 2031 EPS at ~21x ≈ the $578 base-case 5-year target.
Q2 FY26, year-over-year — read these against a stock sitting 14% below its high.
Every revenue and profit line is growing double digits, with Cloud (+82%, operating income more than tripled to $8.8B) compounding fastest off a $25B quarterly base. Capex (clay) is the cost side of the same story: it doubled to $44.9B in the quarter. The disconnect — accelerating operations, a stock below its May high, and shrinking free cash flow — is both the bull’s opening and the bear’s. Cloud growth includes the first TPU system sales (management says growth also accelerated excluding them).
The entire valuation argument compresses into one disagreement: is $200B a year of AI infrastructure a moat being poured, or a bill being run up?
Where each risk sits, not just how big it is. The hot upper-right corner — likely and high-impact — holds a single risk: the capex squeeze. Most of Alphabet’s other serious risks sit one row down, in “possible.” Horizon: 3–5 years.
Capex runs ~$200B against consensus FCF near $8.6B; if backlog converts slowly, depreciation and debt service outrun Cloud profit and the buyback and dividend capacity shrinks.
AI assistants and agents answer queries without the ad-bearing click; Search & other ($63.3B, +17%) stops compounding and funds less of the build.
The $514B backlog needs power, chips and sites on schedule; delivery slips or counterparties renegotiate and recognized revenue lags the capex already spent.
Search cross-appeal, the ad-tech decree and EU DMA cases keep adding cost, constraints and distraction even without a breakup.
Finland halted two data-center builds Oct 7; permitting and grid constraints delay capacity and push up unit costs.
Fleet scaling now needs outside debt ($5B term loan); a safety incident or slower unit economics drags Other Bets and sentiment.
A consumer or macro slowdown hits Search and YouTube ad budgets while fixed AI costs keep rising.
Roughly $136B of unrealized equity gains over two quarters can reverse; cash-neutral, but GAAP EPS and sentiment swing.
An appeals court orders structural remedies in Search or ad tech — the 10-Q flags the DOJ proposal could materially harm the business.
A foundry, memory or geopolitical disruption strands TPU and GPU supply while the company is already supply-constrained.
Hover the dotted terms in the metrics, or scan the desk’s working definitions here.