NOTE · VII / COMPANY RESEARCH

Alphabet (Google): Business,
Investment Case and Outcomes

Alphabet is Google's parent company. Search advertising still supplies most of its economic power, and that cash is now funding AI, Cloud and other long-term projects. The investment is not simply a bet that Google can build good AI. It is a bet that Google can protect the commercial value of search while earning an adequate return on much heavier investment.

OUR VIEW

Google's core business is not collapsing: Search & other advertising revenue is still growing quickly, and Cloud is profitable under the company's segment reporting. What remains unproven is whether AI search can preserve profit per commercial query and whether the roughly $195–205 billion 2026 capital-spending plan can become free cash flow per share. Google can remain an important AI company without its shareholders automatically winning. At the August 2026 historical price, the base path implies about 5% a year, below our roughly 9% minimum return requirement.

01 · The thirty-second answer

If you read only one section, read this one. The rest of the note explains these four answers.

WHAT THE COMPANY DOES

It turns attention and intent into revenue

Search, YouTube, Android and Chrome attract users, then Google charges advertisers. Google Cloud sells computing, data and AI services to companies. Subscriptions, devices and projects such as Waymo make up the rest.

WHERE THE MONEY COMES FROM

Search is still the main cash engine

Google Search & other—mainly search advertising—produced about $63.3 billion of Q2 2026 revenue. Google Services earns most of the company's profit. Cloud is growing rapidly, but segment profit is not the same as a full return on capital.

WHAT WE ARE BETTING ON

Old profit holds while new investment pays back

AI can make search better, but it can also reduce ad clicks and raise the cost of each answer. New-business profit after depreciation and shared research must offset lost search profit, while free cash flow per share keeps rising.

RISK AND RETURN

The purchase price still matters

Using $345.90 on 14 August 2026 as the starting point, the simplified three-year annualised outcomes are roughly −16% bear, +5% base and +16% bull. These are stress tests, not targets. The base path falls below our return requirement.

02 · What kind of company is Google?

Alphabet is the listed holding company; Google is its main subsidiary. The simplest picture is an advertising cash engine, a Cloud and AI growth engine, and a set of long-term experiments sharing the same pool of capital.

Google Services operating profit, AI and Cloud investment, and the per-share value test
Search and other mature services provide cash. Management reinvests it in AI search, Google Cloud and long-term projects. Success ultimately has to show up in free cash flow per share, not only in impressive products.
Search and advertising

People search for products, services and answers. Advertisers pay to appear at the right moment. Google also sells ads on YouTube and partner sites. This remains the main source of revenue and profit.

YouTube, subscriptions and platforms

YouTube earns advertising and membership revenue. Google Play, Google One and devices also charge customers. They matter, but no single one can yet replace search profit.

Google Cloud

Companies rent computing, storage, databases, analytics and AI services. It is the fastest-growing engine and requires large spending on data centres, chips and power.

Long-term projects (Other Bets)

Projects such as Waymo are long-duration experiments. Success could be valuable; failure keeps consuming cash. They should not be treated as free upside.

The listed shares: Alphabet is the issuer. GOOGL is Class A with one vote per share; GOOG is non-voting Class C. Founders retain control mainly through Class B shares with ten votes each. The economic rights of the common classes are broadly the same. Every price and scenario on this page refers to GOOGL.

03 · Where does the profit really come from?

Google does not disclose search profit on its own. We should not call all Services profit “search profit,” even though search remains the strongest economic foundation.

Q2 2026RevenueOperating resultHow to read it
Google ServicesAbout $94.5bnAbout $39.5bn profitIncludes search, YouTube, partner-site advertising, subscriptions, platforms and devices. It is the main profit source, not a pure search segment.
Search & otherAbout $63.3bn, up 17%Not disclosed separatelyThe core business is still growing. There is no financial evidence that AI has already destroyed search.
Google CloudAbout $24.8bn, up 82%About $8.8bn segment profitGrowth and profit are strong, but acquisitions and proprietary AI-chip sales affected growth. Shared AI research is not allocated here, and capital employed is not disclosed, so the 35.6% segment margin cannot establish the investment return.
Long-term projects (Other Bets)About $0.4bnAbout $1.8bn lossThese remain highly uncertain projects rather than a dependable source of current value.

Why search earns so much

The most valuable searches reveal commercial intent: what to buy, where to go or which service to use. Advertisers bid for limited positions, and a click sends the user to a merchant. Google does not merely sell answers; it charges for access to customers at the moment they are ready to act.

Why AI is both an opportunity and a threat to the old business

A more complete AI answer can keep users loyal, but it can also remove the need to click through many links. Move too slowly and users may switch to another assistant; move too quickly and Google may remove some of its own ad opportunities. Generative answers also use more chips and electricity than traditional retrieval. More usage therefore does not guarantee more profit per use.

04 · What is our investment case?

Our case is not “Google has AI, therefore it wins.” The question is whether it can reinvest cash generated by Google Services while increasing free cash flow per share.

Management's 2026 capital-spending guide is about $195–205 billion, mainly for servers, data centres and networking; research expense is separate. Q2 capital spending was about $44.9 billion and free cash flow was about negative $5.9 billion. Free cash flow is operating cash less this kind of long-term construction spending.

One negative quarter does not mean Alphabet is short of money. Trailing-twelve-month free cash flow was still about $53.3 billion, and cash, cash equivalents and marketable securities were about $242.5 billion. New shares and debt are better understood as a choice about how to fund expansion than as a liquidity emergency. Depreciation spreads the cost of servers and data centres bought today across future income statements, so cash leaves first and profit pressure can arrive later. The question is whether that choice improves future value per share.

FIVE THINGS HAVE TO HAPPEN
  1. Commercial intent stays inside Google's products
  2. AI answers preserve ad, transaction or subscription revenue
  3. Cloud growth comes from broad, durable enterprise demand
  4. After depreciation and shared research, new-business profit exceeds lost search profit
  5. After capital spending and the higher share count, free cash flow per share keeps rising
TWO TESTS THE INVESTMENT MUST PASS

Operations: new-business profit after depreciation and shared research must exceed lost search profit. Shareholders: operating cash less capital spending, divided by the higher share count, must keep rising. Debt principal is not subtracted directly in free cash flow, but more debt increases risk.

How the company can win while the stock loses

Traditional search resembles a high-margin toll road. AI Cloud and computing resemble a power plant that needs continual expansion. Both can grow while the latter consumes more capital, carries more depreciation and earns a lower margin. If the market stops valuing Google as a light-capital advertising business and starts valuing it as a heavier infrastructure platform, earnings can grow while the valuation multiple falls.

Why it still deserves attention

REASON ONE

The core has not failed first

Search revenue, paid clicks and cost per click are still rising. Google has time to move users gradually from traditional search into AI search from a position of strength.

REASON TWO

Several paths can add profit

Cloud, YouTube, subscriptions, AI ads, transaction fees and Google's proprietary TPU AI chips do not each need to replace search. Their combined incremental profit matters.

REASON THREE

It owns entry points and infrastructure

Google has consumer distribution, advertising, models, chips and Cloud. That creates options, but it does not guarantee a good return on every dollar invested.

How much trust does management deserve with this capital?

The answer is neither complete trust nor complete distrust. YouTube and proprietary AI chips show that long investment can create important assets. Other Bets still loses money, while the present Cloud acquisitions and the roughly $195–205 billion 2026 construction plan are much larger than earlier decisions. Public evidence is not strong enough for a high-confidence verdict on capital allocation, so we do not count the return before it appears.

05 · What is proven, and what is not?

Current results show that revenue and demand exist today. They do not prove broad customers, durable demand or a decade of capital returns.

QuestionCurrent viewEvidence and limit
Has AI already made search decline?No financial evidence of thatSearch revenue rose 17%, paid clicks 13% and cost per click 3%. That says current monetisation is strong, not that future unit profit is safe.
Is Cloud demand real?Revenue and demand exist todayRevenue rose 82% and segment profit increased sharply. The rate includes the acquired cloud-security company Wiz and proprietary AI-chip sales; customer concentration, low-value raw-compute mix and full-cost profit are not disclosed.
Can AI search match old search margins?Not provenGoogle does not disclose AI-mode ad revenue, query cost, commercial clicks or cash contribution per query. It also does not disclose traditional search profit separately, so the old and new forms cannot be compared directly.
Will the build-out turn into shareholder cash?Not provenCapital spending is rising quickly and quarterly free cash flow turned negative, but trailing-twelve-month free cash flow remained positive and liquidity was ample. The risk is weaker return per share, not an immediate cash shortage.
Is capital discipline strong enough?It needs continued testingThe company has issued shares and added debt while founders retain control through multiple share classes. Outside shareholders have little power to slow investment; the record includes both successful assets and continuing losses.

Why reported net income is a poor shortcut

Alphabet's equity holdings rose sharply in value during Q2, producing unusually large “other income” and reported net income. This is legitimate investment revaluation, not cash earned by Search or Cloud; the 10-Q does not identify every major private holding.

The estimated “clean EPS” is calculated as follows: the research record captured a 2026 consensus EPS of $20.58, then removes about $8.63 per share of after-tax equity-revaluation gains already recognised in the first half of 2026, leaving $11.95. The consensus was captured from a live page on 11 August 2026 and cannot be reproduced exactly after that page updates. The result is our estimate, not an Alphabet figure.

Two important limits: Removing revaluation gains from recurring earnings does not mean the underlying investments have no value. Cloud's reported margin was also about 35.6%, while roughly $5.8 billion of company-level net costs mainly reflected shared AI research; capital employed by Cloud is not disclosed. Segment margin therefore cannot establish the full investment return.

06 · Valuation and return

This is not a target-price exercise. It shows how much success the starting price already required.

The dated starting point

The test starts at the $345.90 GOOGL closing price on 14 August 2026 and uses estimated 2026 clean EPS of $11.95, or about 28.9 times earnings. It assumes that in August 2029 the market applies a multiple to expected full-year 2029 clean EPS. Returns cover exactly three years and exclude dividends. Prices and estimates have since changed.

CaseWhat happens to the business and cash2029 clean EPS × multipleAssumed price after three yearsAnnualised from start
BearSearch remains relevant but profit per commercial query weakens. Cloud does not offset new costs, free cash flow per share stays under pressure and clean EPS grows about 5% a year.$13.83 × 15About $208About −16%
BaseSearch broadly holds and Cloud keeps growing. AI earns an ordinary return, free cash flow per share gradually recovers and clean EPS grows about 15% a year.$18.17 × 22About $400About +5%
BullGoogle protects commercial intent and scales Cloud and AI monetisation. Free cash flow per share rises clearly as capital spending slows, and clean EPS grows about 20% a year.$20.65 × 26About $537About +16%

How to use the table: A 15-times multiple represents a mature, capital-heavy business with slower growth; 22 times represents a good but heavier business; 26 times requires sustained growth and strong capital returns. These are our stress-test assumptions, not company guidance. The table does not value cash, equity investments and debt separately; their net value is simplified into the multiple and may therefore be over- or understated.

OUR RISK-RETURN VIEW

At this historical starting point, the bear path is about −16% a year, the base path +5% and the bull path +16%. The three paths have no reliable probabilities and cannot be combined into an expected return. Even so, the base path falls below our roughly 9% annual hurdle. We would need a lower price or clearer evidence on revenue and cost per use, together with free cash flow per share.

07 · How does the investment win or lose?

Product success, company profit and stock returns are three different tests. They need not happen together.

WIN

Search holds and new businesses earn good returns

AI search preserves commercial intent and advertiser bidding; Cloud demand broadens; profit still grows after shared research and depreciation; free cash flow per share rises when construction slows.

LIMITED WIN

The business performs, but the purchase price was high

Search, Cloud and YouTube all grow, but the new mix is heavier and more expensive. Results merely meet what the starting price already assumed, leaving ordinary stock returns.

LOSE

Search and new investment disappoint together

Commercial intent moves to other assistants; Google adds free AI to defend access; ad opportunities and profit per use fall; Cloud behaves like price-competitive compute while depreciation, financing risk and valuation all come under pressure.

A result that is easy to misread: Cloud keeps growing quickly

Same headlineWhat happens alongside itWhat it means
Cloud grows quicklyCompany margin is stable, construction slows and free cash flow per share risesPositive: revenue growth is becoming free cash flow per share.
Cloud grows quicklyShared research, depreciation, debt and the share count rise fasterNegative: revenue is growing, but enough profit and cash have not yet followed.

08 · How the risks hurt shareholders

The risks below are ranked by their importance to the case. A bad headline matters only through the path by which it reduces value per share.

PriorityRiskHow the stock losesWhat to watch
OneProfit per commercial search weakensQueries still rise, but ad opportunities, bidding or profit per use fall. Earnings and the valuation multiple are cut.Search & other revenue, paid clicks, cost per click, AI ad disclosure and Google Services operating margin.
TwoAI investment earns too littleUtilisation or pricing disappoints; depreciation and shared research rise; free cash flow fails to recover.Capital spending, depreciation, operating cash, free cash flow per share and growth after spending peaks.
ThreeCloud revenue is lower qualityGrowth depends on concentrated AI customers or low-margin compute. Segment profit looks strong while company returns weaken.Customer mix, signed contracts not yet recognised as revenue, company margin, shared costs and external financing. Disclosure is limited.
FourDistribution and antitrust casesDefault-search advantages weaken, distribution gets more expensive, or ad-tech rulings restrict operations or asset structure.Implementation and appeal of the first-instance US search judgment, later compliance and the separate ad-tech case.
FiveControl and capital disciplineFounders continue expanding investment, debt or issuance while outside shareholders cannot apply a brake.Net issuance, buybacks, debt, acquisitions, pay and cash growth per share rather than for the whole company.
WHAT WOULD CHANGE OUR VIEW

We use four consecutive quarters rather than one quarter. We would become more constructive if Search & other kept growing, Google Services operating margin stopped falling, Cloud remained strong after acquisitions and hardware were separated as far as disclosure allows, and trailing free cash flow per share rose as capital-spending growth slowed. We would treat the case as damaged if Services margin kept falling despite healthy search revenue, or if free cash flow per share had not recovered four quarters after spending growth slowed.

09 · Sources, method and disclaimer

Operating data primarily use Q2 2026. Valuation uses a dated August 2026 snapshot solely to show the risk-return relationship at that time. Company facts, reported accounting and research estimates are separated wherever practical.

Clean EPS, scenarios, full-cost segment economics and risk ranking are research estimates, not Alphabet guidance. This note is general research and information only. It is not investment advice, financial product advice, a securities recommendation, an offer or a solicitation.