GOOGLE - Complete Company Analysis After Earnings
This is going to be long, so grab a coffee or use AI to summarize it, but I’ll explain my view on Google after its Q3 2026 results (28. 10.) and today’s stock drop.
For those who don’t want to read the whole post but want to understand the background to my investment thesis, here’s the conclusion:
My investment thesis is based on the company’s ability to adapt and, right now, on the Cloud business surpassing Search by 2030—and I’m willing to pay for that. The rest are “options” (bonuses) for me, generating profits and financing growth. In my calculations, this will happen around 2028–2030, and the fact that the market looks 12 months ahead doesn’t bother me.
In my view, Google is the winner in this Cloud battle, which is also why my allocation to this business is so high and I’m willing to take the risk.
The company adapts over time, and the market punishes it for that, because “the guy at the bank who values the company and calculates returns using free cash flow in a DCF analysis” obviously understands the business better than the company itself 😄 Ironically...
So, let’s get into it :)
What the hell is going on with Google...
----------------------------------------------------------------
Google is changing from a search company into an AI infrastructure company. Is that a bad thing? Or are markets simply discounting this business right now because of rising debt?..
Look: Cloud is growing 103%, has a $561 bln backlog, and according to my model, based on guidance + market expectations, its revenue will surpass Search in 2029 (in the BASE scenario), or as early as 2028 (in the BULL scenario).
The market isn’t overlooking the transformation. It’s discounting it because it’s being paid for with negative free cash flow, debt, and new shares, and the return on investment hasn’t been proven yet—it’s being punished just like other cloud businesses.
MY EXPERIENCE FROM BANKING: What hasn’t been proven is treated as if it doesn’t exist... institutions pay a lower multiple for it.
1. Q3 2026 RESULTS AND MARKET REACTION
=============================================================
[Revenue]
Result: $127,6 bln (+25 %)
Comparison: estimate ~$127,4–127,9 bln
Assessment: in line
[Search & other]
Result: $63,35 bln (+12 %)
Comparison: estimate ~$65,0 bln · Q2 +17 %
Assessment: MISS, slowing
[YouTube advertising]
Result: $11,2 bln (+9 %)
Comparison: Q2 +13 %
Assessment: slowing
[Google Cloud]
Result: $30,7 bln (+103 %), margin 36,2 %
Comparison: estimate ~$28,7 bln
Assessment: BEAT
[Operating profit]
Result: $40,6 bln (+30 %), margin 31,8 %
Comparison: estimate $43,2 bln (FactSet)
Assessment: MISS ~6 %
[EPS / excluding investment losses]
Result: $2,12 / ~$2,67
Comparison: estimate ~$2,97
Assessment: MISS
[Capex / free cash flow]
Result: $51,4 bln / −$9,1 bln
Comparison: Q2 FCF −$5,9 bln
Assessment: burning cash
[Cloud backlog]
Result: $561 bln (+256 % y/y)
Comparison: 55 % is expected within 24 months
Assessment: Strong
• Reaction on 29. 10.: shares −12 to −13 %, market value −$500 bln.
Worst earnings reaction since the 2004 IPO.
• Capex: 2026 remains at $195–205 bln; 2027 is expected to “increase significantly” (Goldman estimates ~$350 bln). Inventory related to TPU will weigh on cash flow through 2027.
• Cloud margins: third-party capacity will temporarily lower margins in Q4.
• TPU: nearly 100 customers, including Anthropic, Apple, Oracle and Crux AI.
• Why Search slowed: according to management, finance, retail and media, and lower monetization in AI Mode. AI is cannibalizing Google Search.
But the number of queries is growing at its fastest pace in 5 years.
• Financing: no buybacks. 36 mln new shares, $28,7 bln in bonds, $19 bln in preferred shares. Long-term debt of $125 bln (end of 2025: $46,5 bln). WARREN BUFFETT ALSO FINANCED THIS AT $350 PER SHARE.
(So he bought at a higher price 😄 )
• Analysts: Goldman Buy, target $435 → $385 · BofA Buy, $380 · UBS Neutral, $332 ·
=============================================================
2. BUSINESS SEGMENTS
=============================================================
[Search & other]
2026E revenue: ~$256 bln
Growth: 12–19 %, slowing
Description: Ads in Search, Maps and Gmail. A cash cow with the highest margins. AI is increasing the number of queries, but they’re still being monetized less effectively.
[YouTube advertising]
2026E revenue: ~$44 bln
Growth: 9–13 %
Description: Ads alongside videos, Shorts and on TV. Competition from TikTok and streaming.
[Google Network]
2026E revenue: ~$29 bln
Growth: ~0 %
Description: Ads on third-party websites. Low margins, in decline.
[Subscriptions, platforms, devices]
2026E revenue: ~$56 bln
Growth: 11–19 %
Description: YouTube Premium and Music (150 mln.+), Google One, Gemini, Play, Pixel.
[Google Cloud (incl. TPU)]
2026E revenue: ~$110 bln
Growth: 63 → 82 → 103 % GROWTH IS ACCELERATING.
Description: Computing capacity (GCP), Gemini API, Workspace and sales of TPU systems.
[Other Bets, Waymo, robotics, etc.]
2026E revenue: ~$1,5 bln
Operating loss: ~$7,5 bln
Description: Mainly Waymo (robotaxis)
[Holdings]
Anthropic (~14 %), SpaceX (~6 %). They affect earnings through revaluations (Q3 −$8,7 bln). WHEN you buy Google, you also buy Anthropic, SpaceX and other companies Google owns.
3. BASE PROJECTIONS (2026–2030)
Realistic assumptions:
Search grows by 9 → 7 % (AI partially cannibalizes monetization).
Cloud +72 % in 2027 (backlog and first major TPU revenue), then +38 %, +25 %, +18 %.
Services margin ~41 %, Cloud 33–34 % (depreciation, low-margin TPU hardware).
AI research $30–40 bln per year. Dilution ~1 % per year.
=============================================================
4. BULL PROJECTIONS (2026–2030)
Assumptions: Search stabilizes at 9–12 % (AI Mode learns to monetize). Cloud +113 % in 2027 (Morgan Stanley: TPU revenue of $84 bln in 2027, $108 bln in 2028), then +50 %, +32 %, +22 %. Cloud margins 35–38 %, Services 42–43 %.
Waymo scales up, the company resumes buybacks.
=============================================================
5. WHEN CLOUD WILL SURPASS SEARCH
=============================================================
Search vs. Cloud revenue (bln USD)
[BASE]
2026: Search 256 vs. Cloud 110
2027: Search 279 vs. Cloud 190
2028: Search 302 vs. Cloud 262
2029: Search 323 vs. Cloud 328
2030: Search 345 vs. Cloud 387
[BULL]
2026: Search 256 vs. Cloud 110
2027: Search 287 vs. Cloud 235
2028: Search 318 vs. Cloud 352
2029: Search 350 vs. Cloud 465
2030: Search 382 vs. Cloud 568
By revenue, Cloud will catch up with Search within 2–3 years. It will take longer in terms of profit, because Search margins are well above 40 % and Cloud will carry huge depreciation expenses and low-margin TPU sales (gross margin ~30 % according to Morgan Stanley).
Morgan Stanley expects Cloud to account for as much as 48 % of Alphabet’s operating profit by 2028, close to my BULL scenario.
=============================================================
6. ARE INVESTORS UNDERPRICING THE TRANSFORMATION BECAUSE OF DEBT?
=============================================================
• Goldman Sachs (Buy, $385): the market underestimates AI distribution through products with billions of users, computing scale and proprietary chips. According to Goldman, the EPS miss is mainly due to a non-cash write-down of investments.
• The stock trades at ~18× 2027 GAAP earnings and ~15× 2028 earnings (excluding net cash, Goldman). That’s the valuation of a slow-growth advertising company, not AI infrastructure.
• Cloud +103 % versus Azure +48 % and AWS +37 %, backlog $561 bln.. CRUSHING THE COMPETITION
• In the BULL scenario, Cloud alone, discounted to today, is roughly as large as all of today’s Alphabet (~$3,4 trln).
THE BEAR CASE: the market sees cash burning
• Cash is actually burning: FCF −$9,1 bln, and 2027 will be worse. Real money, not an accounting loss.
• Dilution: new shares through ATM and preferred shares, no buybacks.
• Debt of $125 bln. The market isn’t questioning solvency, but the pace of further borrowing.
• Pivot: “the first quarter in which AI starts seriously hurting Search.”
• Return on capital hasn’t been proven yet
VERDICT
The market sees Cloud, but applies a big “show-me” discount. It will narrow once FCF turns positive, the company stops issuing shares, and TPU delivers a real return on investment.
That’s where the BULL scenario’s potential lies.
=============================================================
7. SUM-OF-THE-PARTS VALUE IN 2030 AND DISCOUNT TO TODAY
[Google Services]
Method: 17× / 19× net income
BASE: 2 980 bln USD | BULL: 3 868 bln USD
[Google Cloud incl. TPU]
Method: 25× / 30× net income
BASE: 2 729 bln USD | BULL: 5 371 bln USD
[Alphabet AI research and costs]
Method: −15× net cost
BASE: -498 bln USD | BULL: -498 bln USD
[Waymo]
Method: estimated value
BASE: 150 bln USD | BULL: 300 bln USD
[Holdings (Anthropic, SpaceX), after tax]
Method: market value × 0,8 discount
BASE: 248 bln USD | BULL: 424 bln USD
[Net cash / debt]
Method: after years of negative FCF
BASE: -50 bln USD | BULL: 100 bln USD
[ALPHABET VALUE IN 2030]
Method: sum
BASE: 5 559 bln USD | BULL: 9 565 bln USD
Per-share value and discount
----------------------------------------------------------------
• Price target per share (end of 2030): BASE $434 | BULL $765
• Implied 2030 P/E: BASE 22,6× | BULL 27,5×
• Present value (discount rate 10 % / 9 %, 4,2 years): BASE $303 | BULL $514
• Expected annual return from $280: BASE ~11,3 % | BULL ~27,4 %
Rate cuts, an end to the war and cheaper oil: impact on the scenarios
This would increase the value by roughly 10–15 %. Mainly by lowering the discount rate and increasing valuation multiples.
=============================================================
8. RISKS THAT COULD MAKE CLOUD UNPROFITABLE
=============================================================
1. Customer concentration: a large part of the backlog comes from a few AI labs (Anthropic 5 GW, Apple, OpenAI, Oracle). If one slows down, switches to its own chips or can’t secure financing, the backlog will shrink.
2. Circular financing: Google invests in Anthropic, and Anthropic spends money on Google Cloud. Some of the revenue is indirectly paid for with Google’s money.
3. Price war and excess capacity: AWS, Azure, Oracle, CoreWeave and neoclouds are building out capacity at the same time. If supply gets ahead of demand, prices and margins will fall.
4. Low TPU hardware margins: they increase revenue, but with a gross margin of ~30 %.
5. Third-party capacity: renting other companies’ data centers lowers margins (the company warned about this for Q4).
6. Chip obsolescence and depreciation: see section 9.
7. Suppliers and energy: HBM sold out (SK hynix through 2029), limited capacity at TSMC and limited electricity.
8. Chip competition: Nvidia, Amazon’s Trainium, Microsoft’s Maia.
9. Regulation: antitrust scrutiny of ties with Anthropic and bundling of services.
=============================================================
9. DEPRECIATION, SIMPLY EXPLAINED (important)
=============================================================
Example: You buy a taxi for 600 000 Kč that will last 6 years. You pay the money up front (capex, which is why FCF is negative). But in the accounts, you spread the cost over 6 years at 100 000 Kč per year—that’s depreciation. The cash goes out today, but profits are reduced for another 6 years.
Google buys servers, TPUs and data centers for ~$200 bln per year, possibly $300–350 bln in 2027. Another layer of depreciation is added every year.
Depreciation risks:
• If depreciation grows faster than revenue: it could increase ~6× by 2030.
If Cloud slows down sooner, margins will collapse.
• Shortening useful life: if the company has to switch from 6 years to 4, 2030 depreciation will rise by $65 bln and EPS will fall by $4.
• Why it matters: today’s P/E looks low because earnings don’t yet include the full wave of depreciation from 2026–2027. That will come in 2027–2030.
=============================================================
10. HISTORY: WHAT FOLLOWED GOOGL DROPS OF 8 % OR MORE
=============================================================
GOOGL statistics since its 2004 IPO: 13 drops of 8 % or more in a day
[After earnings, 7 cases]
Next day: down in 7 out of 7 cases, average −2,1 %
+20 days: up in 4 out of 7 cases, average ~+0,8 %
+60 days: up in 6 out of 7 cases, median +2,9 % (average ~+2,2 %)
[Other cases, 2008 and 2020 panics, 6 cases]
Next day: up in 6 out of 6 cases
+60 days: mostly significantly higher (+21 to +37 %, except 2008)
What this tells us
----------------------------------------------------------------
• Today’s drop (~−12 to −13 %) is the biggest earnings reaction since the IPO, larger than all 7 previous ones (the worst was −9,8 % in 2008).
• After earnings drops, the stock fell again the following day every time, so there’s no need to rush into buying.
• After 60 days, it was almost always higher. Big rebounds followed market-wide panics, not disappointing results.
=============================================================
MY INVESTMENT THESIS
=============================================================
At ~$280, I bought the BASE scenario below fair value and the BULL scenario as an option with potential returns of ~27 % a year. The risk/reward ratio is VERY attractive IN MY VIEW, while keeping in mind all the risks that may emerge along the way. It’s quite likely that if market sentiment holds and investors don’t want to hear anything about AI, negative FCF and debt, there could be several more swings downward.
What could hurt Google:
• A significant further decline in Search.
• If Cloud growth slows below 40 % in 2027, or margins fall below 25 %.
• If 2027 capex is significantly above $400 bln without a corresponding increase in backlog... we’ll find out next quarter, and it could hurt a lot if we hear something crazy 😄
• Dilution above ~2 % per year.
• Losing a key TPU or Cloud customer, or Anthropic running into financing problems.
=============================================================
DISCLAIMER
Projections, multiples and scenarios are MY own model for my analysis, and I’m sharing my thoughts here about my own investment decision. I currently hold 46 shares of Google. Everyone has a different risk tolerance; I could be wrong, and I don’t recommend buying Google or any other stock based solely on someone else’s analysis. Do your own research.
This is not investment advice.
