GOOG, Nasdaq. Run 8 September 2026. Built from ten years of Alphabet 10-K filings, the Q2 2026 10-Q and the Q2 2026 earnings call, plus cited web market data for price. Base: ~$336 per share, ~34x normalised after-tax operating earnings.
Alphabet is one very large, very old profit engine bolted to one very large, very new capital project. The old engine is Google Services: Search, YouTube, Android, Play, devices and subscriptions, $342.7 billion of FY2025 revenue at a 41% operating margin (FY2025 10-K). It sells advertiser access to user intent through an auction, at close to zero incremental cost, with traffic acquisition cost of 20.3% of relevant revenue as the main variable cost of distribution (FY2025 10-K). The new project is Google Cloud plus the frontier-model stack behind it: $58.7 billion of FY2025 revenue, $24.8 billion in the second quarter of 2026 alone, growing 82% year over year at a 35.6% operating margin, up from 20.7% a year earlier (Q2 2026 call). A fourth line, Alphabet-level activities, absorbed $16.8 billion of shared AI research in FY2025 and $11.2 billion in the first half of 2026 alone, and is charged to no segment at all (FY2025 10-K; Q2 2026 10-Q).
85% of FY2025 revenue, 41% operating margin. Search and other alone was $224.5 billion, growing 17% in Q2 2026. YouTube ads $40.4 billion. Network $29.8 billion and shrinking. Subscriptions, platforms and devices $48.0 billion, +15%. This is what pays for everything else (FY2025 10-K; Q2 2026 call).
15% of FY2025 revenue, ~21% by Q2 2026. Backlog $514 billion, up more than $50 billion sequentially, with just over 50% expected to be recognised within 24 months. It also now sells TPU systems into customers' own data centres, with the bulk of that revenue expected in 2027 (Q2 2026 call).
Hold that picture: a mature, high-margin advertising machine that grows in the low teens, and a capital-hungry compute business growing at triple that rate whose costs arrive before its revenue does. Every lever below is really a question about which of those two is setting the pace.
Alphabet has two completely different volume mechanisms. In Search the unit is the paid click, and the filings show a long, orderly decay: clicks grew 40%, 54% and 62% in 2016 to 2018, then 23%, 19%, 23%, 10%, 7%, 5% and 6% in the years to 2025 (FY2016 to FY2025 10-Ks). Mid-single digits is now the run rate, and it is a genuine ceiling: there are only so many people, and Alphabet already has thirteen products with over a billion users. Management's own evidence that AI is adding rather than subtracting queries is real but qualitative, with AI Mode past one billion monthly active users and Search usage at an all-time high during the World Cup (Q2 2026 call). In Cloud the unit is a chip-hour, and the ceiling is physical: Alphabet says it is operating in a "supply constraint environment" and is renting third-party capacity in Q3 as a bridge while it builds its own. Capex of $195 to $205 billion in 2026, raised mid-year from $180 to $190 billion and set to "increase significantly in 2027", is not a financing detail here. It is the volume lever (Q2 2026 call). Against that sits Google Network, where impressions fell 7% in 2025 and revenue is in structural decline (FY2025 10-K).
This is where the case is most fragile, and the filings say so plainly. Cost-per-click on Google Search and other properties was negative every single year from 2016 through 2020: minus 13%, minus 21%, minus 25%, minus 7%, minus 10% (FY2016 to FY2020 10-Ks). It turned up in 2021 at plus 15%, went flat in 2022 and 2023 at minus 1% and plus 1%, and has now printed plus 7% in both 2024 and 2025 (FY2021 to FY2025 10-Ks). So Alphabet's demonstrated long-run price contribution in its largest business is roughly zero to negative, and the last two years are the exception, not the rule. Any scenario that carries plus 7% forward for five years is assuming that a two-year run is the new normal. There is real evidence behind the recent pace: management reports "encouraging" monetisation on queries showing AI Overviews even as coverage widened to more commercial queries, with retail and finance driving the largest contributions to a 17% Search revenue gain (Q2 2026 call). But Alphabet publishes no monetisation-per-query figure, so the claim cannot be checked from outside. On the Cloud side price is not Alphabet's to set: it competes with two well-capitalised rivals, and its own advantage is on cost, not price, as shown by cutting the cost of an AI Mode response to its lowest level since launch (Q2 2026 call).
Two mix shifts are running at once and they point in opposite directions. The helpful one: Cloud has gone from 7% of revenue in FY2020 to 14.6% in FY2025 to about 21% in Q2 2026, and its operating margin has gone from 20.7% to 35.6% in a year, which is now above Alphabet's consolidated 34% (FY2020 and FY2025 10-Ks; Q2 2026 call). Meanwhile Google Network, the lowest-quality revenue in the house because most of it is paid straight back out as traffic acquisition cost, is shrinking at 1 to 2% a year. Both of those quietly raise the average. The unhelpful one: the incoming mix is capital-heavy. TPU system sales are hardware, and hardware carries a different cost structure to selling chip-hours; inventory jumped from $2.4 billion to $10.0 billion in six months as those systems were built, and management expects the vast majority of that revenue in 2027 (Q2 2026 10-Q; Q2 2026 call). Alphabet has not disclosed the margin on it. Add the third-party bridging capacity that management says "will create modest margin pressure in the near-term", and the shared AI research bucket that grew to $11.2 billion in half a year and belongs to no segment, and the mix runway is genuinely ambiguous rather than simply favourable. Mix also interacts with the other levers: the Cloud shift is the volume lever, it dilutes the price lever because Cloud has less pricing power than Search, and it is the direct cause of the margin problem below.
Margin was Alphabet's single best lever of the last decade and is now running in reverse. Consolidated operating margin went from 20.1% in 2018 to 32.0% in 2025 and 34% in Q2 2026, roughly fourteen points of expansion, most of it operating leverage on headcount and the 2023 cost programme (FY2018 to FY2025 10-Ks; Q2 2026 call). That is close to the ceiling. Google Services is already at 41.8%, off a peak near 45%, and consolidated 33 to 34% is the highest Alphabet has ever printed. There is a named frontier above it (Microsoft has sustained mid-forties operating margins) but Alphabet is not structurally that business. Now the drag, which is arithmetic rather than opinion. Depreciation went from $15.3 billion in FY2024 to $21.1 billion in FY2025 to $13.6 billion in the first half of 2026 alone, up 43% year on year (FY2024 and FY2025 10-Ks; Q2 2026 10-Q). Management says roughly 60% of technical-infrastructure spend is servers and 40% data centres and networking (Q2 2026 call). On six-year server lives, every $100 billion of annual capex adds roughly $12 billion a year of run-rate depreciation once in service. At the guided ~$200 billion for 2026 that is ~$24 billion a year added, each year, and management has said 2027 goes higher (capex beyond 2026 is tagged approx., unverified). Set that against revenue growth: on a plausible build Alphabet adds roughly $70 billion of revenue a year, so incremental depreciation runs at roughly a third of incremental revenue, against 5.2% of revenue in FY2025. Operating expenses cannot leverage fast enough to offset that; they were 27.6% of revenue in FY2025, so even growing three points a year slower than revenue buys back only about two points of margin over five years. Management itself guides to the drag: higher depreciation and data-centre operating costs "will continue to put pressure on the P&L" and free cash flow "will remain under pressure" (Q2 2026 call).
Add the levers honestly. Volume ~8 to 9 points, price ~1.5 to 2.5 points, mix ~0 to 1 point gives plausible revenue growth of roughly 10 to 13% a year, which is consistent with a bottom-up build of the segments: Search at ~10%, YouTube at ~12%, subscriptions at ~14%, Network declining, and Cloud compounding at ~28% off its FY2025 base takes total revenue to roughly $750 billion by 2030, about 13.5% a year (author's build on FY2025 10-K segment revenue). Apply the margin lever at minus 1 to minus 3 points a year and plausible earnings growth is roughly 7 to 11% a year. The dividend adds ~0.26% ($0.22 per quarter declared, Q2 2026 call). So the business, on its own, supplies roughly 7 to 11% a year before the market does anything at all.
The target needs ~14.5% a year in earnings if the multiple simply holds at today's ~34x, and ~18.5% a year if it drifts back to Alphabet's own ten-year median of ~29x. That is a gap of roughly 4 to 7 points a year at the current multiple and 8 to 11 points at the median, and those points have no identified source in the levers. Worse, the funding branch makes it harder rather than easier. Alphabet bought back $45.4 billion of stock in FY2025 and zero in the first half of 2026; it issued $49.6 billion of equity including a $19 billion mandatory convertible preferred, opened a $40 billion at-the-market programme on 1 June 2026, and took long-term debt from $46.5 billion to $98.2 billion; shares outstanding went from 12,088 million to 12,230 million in six months, an annualised pace of about 2.4% (FY2025 10-K; Q2 2026 10-Q). Trailing free cash flow is $53.3 billion against trailing capex of $132.4 billion, and Q2 2026 free cash flow was negative $5.9 billion (Q2 2026 call). Growth here is consuming the cash flow, so the dividend and buyback that quietly sit inside a 15% target have to be set to zero and dilution added back: at the current multiple the requirement becomes roughly 16.5% a year of net income growth at a modest 1.5% dilution rate, and roughly 20% a year at the median multiple. This is a re-rating bet. Not because the business is weak, but because the arithmetic only reaches 15% if Alphabet is still valued at or above ~34x in 2031, which is the very top of the range it has traded in for ten years (~19x in 2022, ~29x median, ~35x peak in 2020).
On precedent, the honest answer is that there is almost none to consult. The requirement is roughly 15% a year of after-tax earnings growth for five years from a ~$122 billion base. Only three or four companies in history have had a base that size at all. Microsoft compounded net income at roughly 18% a year from FY2020 to FY2025, but from a base of about $44 billion, roughly a third of Alphabet's (macrotrends.net, approx., unverified). Alphabet itself grew operating income at roughly 26% a year from FY2020 to FY2025, also from about a third of today's base (FY2020 and FY2025 10-Ks). Apple's earnings have been roughly flat in the high nineties of billions for several years, which is the closest analogue for a company already at this base size (approx., unverified). Fewer than three genuinely comparable peers exist, so precedent analysis is thin here, and that thinness is itself the finding: no company has been observed compounding a $120 billion-plus earnings base at 15% a year for five years, because almost none has had one.
Move the four operating levers and the exit multiple. The page computes earnings growth, folds in the dividend, and shows the annualised return over 3 and 5 years, then splits it into its three sources. Read the case first; this is for pressure-testing it. Note that the multiple here is on normalised after-tax operating earnings (~$9.90 per share), not on Alphabet's reported EPS.
| Scenario | Vol | Price | Mix | Margin | Earnings CAGR | Exit mult. | 5-yr return |
|---|
Bear holds Search volume at its current mid-single-digit pace, lets CPC fade back toward its long-run norm, and carries the full depreciation drag, with the multiple reverting toward the 2022 trough. Base is the memo's central case: strong volume, price fading only halfway, and roughly two points a year of margin compression, exiting at the ten-year median. Bull is the steelman: Cloud compounding faster than the backlog schedule implies, price holding near its recent pace, and the capex arriving alongside its revenue so margin turns positive again, with the multiple above today's. Only Bull clears 15%, and it needs both the business and the multiple to cooperate.
No DCF, no discount rate, no terminal value. Total return decomposes into earnings growth, multiple change, share count change, and dividends. Earnings are normalised to after-tax operating income throughout, because reported net income currently contains very large unrealised equity gains. Every figure is rounded on purpose and carries a source or is tagged approx., unverified. This page states the bet the price represents; whether to take it is your decision.