--- title: "GOOGL: The 'Money Machine' myth is back. Can the 'sell shovels' play hold up?" type: "Topics" locale: "en" url: "https://longbridge.com/en/dolphin/post/44612483.md" description: "$Alphabet(GOOGL.US) released its Q3 2026 results after-hours on Oct 29 (Beijing time). Core metrics in search, cloud, and profit missed estimates, further souring an already unfavorable narrative for Google.Specifically: 1) search decelerated as the AI cannibalization story began to play out. Q3 search ad growth was only 12%, with a notable QoQ slowdown.In a period of heavy AI investment, with profit and cash flow expected to be under pressure, a slowdown in the core franchise is the most damaging to the thesis. This validates the AI cannibalization narrative..." datetime: "2026-10-29T02:11:10.000Z" locales: - [en](https://longbridge.com/en/dolphin/post/44612483.md) - [zh-CN](https://longbridge.com/zh-CN/dolphin/post/44612483.md) - [zh-HK](https://longbridge.com/zh-HK/dolphin/post/44612483.md) author: "[Dolphin Research](https://longbridge.com/en/dolphin.md)" generator: "portal-rs" --- # GOOGL: The 'Money Machine' myth is back. Can the 'sell shovels' play hold up? $Alphabet(GOOGL.US) released its Q3 2026 results after the U.S. market close on Oct 29 Beijing time, with core metrics — search, cloud and profit — all missing. For Alphabet, where the narrative was already on the back foot, this is salt in the wound. **Specifically:** **1) Search decelerates; the AI cannibalization narrative starts to play out** Q3 search ad revenue grew 12%, a clear QoQ slowdown. With heavy AI capex expected to weigh on margins and FCF, any stall in the core franchise does the most damage to the Alphabet thesis, reinforcing fears that AI is eroding search. That said, past quarters showed AI lifting search share, suggesting Q3’s miss may also reflect Gemini’s lag and a hit to product experience leading to user churn. Gemini’s lag, in turn, ties to internal vs. external compute allocation. In 1H, GCP tilted compute toward external clients, and we often heard DeepMind staff flag internal shortages for frontier model training. At the same time, as World Cup tailwinds faded, YouTube ads slowed meaningfully under a high base. Ahead of earnings, third-party data already showed slower query growth and share loss, with the EU’s DMA driving weaker travel and alcohol searches in the region. The stock’s pullback captured some of this concern, but the actual print was worse. Versus Meta Platforms, which keeps growing ad revenue via higher pricing, Google’s search pricing power looks thinner. Since 2026 began, CPC growth has cooled sharply, hinting AI may be accelerating budget shifts from search to social. **Notably, as Muse penetrates key search-heavy verticals like travel and finance,** Meta has a shot at offline service categories where it historically lacked an edge. This broadens the competitive threat to search. **2) Cloud and TPU still surge, but lofty expectations were embedded** Q3 cloud revenue (incl. TPU sales, primarily v7 deliveries in Q3) rose 103%, accelerating QoQ. However, buy- and sell-side expectations were even higher; the BBG consensus (+91%) didn’t fully capture the step-up after first-time recognition of TPU hardware sales (sell-side +105%; buy-side +110%). Backlog was 561 bn in Q3. Adding recognized in-quarter amounts implies ~77.7 bn in new contract value, roughly flat vs. Q2 and up ~1.0 bn, indicating demand remains healthy. **Dolphin Research believes the cloud miss partly reflects two straight quarters of lag in frontier model iterations, weighing on API revenue.** Gemini 4 Argon, launched end-Sep and opened to the public mid-Oct, is a big step up vs. its prior version, but still trails SOTA peers on coding and remains suboptimal on price/performance — cf. our prior note <[link](https://longbridge.cn/en/dolphin/post/44162543?channel=SH000001&invite-code=355628&app_id=longbridge&utm_source=longbridge_app_share&locale=zh-CN&share_track_id=7d418dc4-b745-4402-8743-bdb9626425a5)\>. **Post-print, Gemini 4 Carbon dropped and, on overall capability, jumped to SOTA territory quickly.** Per-task cost is still not best-in-class on value, but somewhat lower than Opus 5.5; watch for real-world user tests. **Separately, based on channel checks, the expectation gap may also stem from TPU shipment timing (e.g., large-scale v8i deliveries clearly pushed by one quarter to YE).** The split within cloud was not disclosed; watch the call for how much TPUs contributed (management said ‘significant’) and the impact from v8 delays. **3) Profit weakened under heavy AI spend** Group OPM fell 200bps QoQ in Q3, driven by higher D&A and R&D, reflecting compute infra build-out and large-model training. Beyond lower margins on TPU hardware vs. the legacy group, management noted short-term reliance on third-party compute due to supply constraints. **For example, compute short-leased from SpaceX tends to be expensive given on-demand needs, further pressuring margins.** By segment, Google Services OPM fell 100bps QoQ, while Cloud OPM expanded QoQ. Note model training costs are partially booked in corporate R&D, not in Cloud — one of the few upside surprises. **4) Heavy capex widens the FCF gap** Q3 capex was 51.4 bn, up 2x YoY and still accelerating, reaching ~40% of revenue. FCF fell deeper into negative territory at -9.1 bn. **The market expects capex of ~220 bn in 2026 and ~350 bn in 2027** (company stance: 195–205 bn for 2026 unchanged, implying 63–73 bn in Q4 and continued QoQ acceleration; 2027 to grow meaningfully). **On this trajectory, full-year FCF is likely still negative next year.** Alphabet continued executing its 85 bn financing plan set at the start of the year, issuing $12.5 bn of equity in Q3, with 27.6 bn capacity remaining. It also issued 28.7 bn of unsecured notes in Q3; cash, ST investments and remaining capacity should largely cover funding needs for the next two years. **5) Key metrics vs. expectations** Alphabet runs multiple businesses and has reshaped its reporting several times; for readers less familiar, here’s the business structure for context. It helps frame the drivers discussed above. ![Figure auto-generated description](https://pub.pbkrs.com/cms/2026/0/PmzVzhDyToVXDyuowZ1EJHVTWxjVgX1M/editor.jpg?x-oss-process=style/lg) **In brief, the long-term fundamental logic:** a) Ads are the revenue anchor and main profit pool. Search faces medium- to long-term encroachment from feed-based ads, with high-growth YouTube filling the gap. b) Cloud, incl. TPU hardware, is the primary growth engine and the clearest AI beneficiary. It underpins the current topline acceleration narrative. **Dolphin Research view** Over the past two quarters, short interest in Alphabet rose rapidly among Big Tech. As AI’s erosion of search deepens, last year’s ‘AI full-stack’ bull case has been questioned, especially after talent departures and visible lag in frontier model iterations. Technically, the gap is catchable, especially on the application side focusing on agents and ecosystem. With rich user scenarios, distribution and multimodal strengths, Alphabet can still carve out advantages. But the core issue from this print is: **AI is turning Alphabet more into a picks-and-shovels supplier, while the cash-cow search engine is slowing.** With ads still ~60%+ of the base, AI’s cannibalization has been evidenced; turning a supertanker takes time, and earnings may deteriorate more than expected during the transition. Meanwhile, management frames next year’s capex as a significant step-up. Street models assume 6–7GW of new compute deployment at ~4.3 bn/GW unit cost, plus 50–60 bn for existing DC maintenance, totaling ~350 bn. On that math, next year’s FCF could be around -100 bn. Ignoring long-term debt, liquidity looks sufficient near term; but if spend ramps further, search weakens, and from a post-2028 lens, Alphabet still faces nearly 1 tn of contractual obligations, ~700+ bn due within five years on a quasi-hard basis (Q2-end data; Q3 to be updated in the 10-Q). Offsetting this are customer backlog and Anthropic’s ~114.5 bn compute commitment to Alphabet. **If TPUs ship and collect on schedule, FCF could inflect around 2028 (our model sees 2H27 turning positive).** The risk is a two-sided mismatch — purchase commitments are rigid, while a sizable chunk of backlog comes from cash-burning AI firms like Anthropic. If **Anthropic** struggles to fundraise, Alphabet’s collection risk rises. Alphabet is tightening its tie-up with Anthropic to counter ChatGPT, embedding it deeper into Google Workspace for seamless integration — e.g., Anthropic buttons directly in Google Docs. On consumer subs, they may bundle plans offering both models, forcing users to choose between OpenAI and Anthropic+Gemini. That said, the Anthropic partnership is non-exclusive. Anthropic also offers Office integrations and Microsoft 365 connectors for cross-app workflows within the Microsoft ecosystem. Absent ad support, cloud plus TPUs alone make it harder to cover obligations. So beyond keeping frontier models on pace, we also want to see more AI agent products that leverage Alphabet’s native channels to form a new traffic entry point, rather than relying solely on legacy search. **On valuation, our prior take was: ‘Aggressive spend will worsen near-term fundamentals (margins and FCF), and as valuation and lofty expectations reset... a window of dislocation could emerge.’ After-hours -10% now puts mkt cap at ~3.6 tn, which lands in our strike zone of sub-20x next-year P/E.** **But Q3 negatives came in worse than we expected; if ads keep deteriorating, further EPS cuts are possible.** Sentiment-wise, Gemini Carbon may offer some relief, but not a cure-all. **Therefore, while 3.6 tn may sit in a relatively low valuation band long-term, near-term rebound catalysts look thin.** With overshoot risk, shares may grind lower before a crowded short squeeze drives a technical bounce. On fundamentals, we look to a personal AI agent to cement a new entry point, and to more TPU wins plus large-scale deliveries to elongate Cloud’s ramp. \ Risk disclosure and statements: [Dolphin Research disclaimer and general disclosures](https://support.longbridge.global/topics/misc/dolphin-disclaimer) **Dolphin Research 'Google' archive:** **Earnings season** Jul 23, 2026 call notes '[Google (Trans): All-in on Gemini 4, capex to rise sharply next year](https://longbridge.cn/topics/42872083?channel=SH000001&invite-code=355628&app_id=longbridge&utm_source=longbridge_app_share&locale=zh-CN&share_track_id=e74a6e92-94bc-41e7-8406-55a4026f0b3f)' Jul 23, 2026 earnings take '[Google: Is AI cracking? A savior emerges](https://longbridge.cn/topics/42870745?channel=SH000001&invite-code=355628&app_id=longbridge&utm_source=longbridge_app_share&locale=zh-CN&share_track_id=cca98b91-de86-4a06-a655-c27aa5c393f8)' **Hot topics** Jul 27, 2026 '[‘Trial’ of Google: Flighty market or guilty for hyper-investing?](https://longbridge.cn/topics/42954949?channel=SH000001&invite-code=355628&app_id=longbridge&utm_source=longbridge_app_share&locale=zh-CN&share_track_id=cc63cb13-226f-4640-b89c-db6e970d5b8a)' Jun 2, 2026 '[Google adds another 8.5 bn of debt — extension rally or bubble litmus?](https://longbridge.cn/topics/41375509?channel=SH000001&invite-code=355628&app_id=longbridge&utm_source=longbridge_app_share&locale=zh-CN&share_track_id=48394c65-07b0-4bc5-88c3-f901306b4ac1)' Dec 19, 2025 '[Google vs. NVIDIA: Is the comeback story credible?](https://longbridge.cn/en/topics/37259603?invite-code=032064)' Deep dives Sep 21, 2026 '[Challenging NVIDIA’s dominance — what powers Google’s ‘Optical Net’](https://longbridge.cn/topics/43994983?channel=SH000001&invite-code=355628&app_id=longbridge&utm_source=longbridge_app_share&locale=zh-CN&share_track_id=008d69c5-fbb7-4eb7-b151-71a1f5bc5302)' ### Related Stocks - [META.US](https://longbridge.com/en/quote/META.US.md) - [GOOGN.US](https://longbridge.com/en/quote/GOOGN.US.md) - [GOOGM.US](https://longbridge.com/en/quote/GOOGM.US.md) - [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md) - [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md) - [GGLL.US](https://longbridge.com/en/quote/GGLL.US.md) - [GGLS.US](https://longbridge.com/en/quote/GGLS.US.md) - [GOOY.US](https://longbridge.com/en/quote/GOOY.US.md) - [GOOX.US](https://longbridge.com/en/quote/GOOX.US.md) - [GOOP.US](https://longbridge.com/en/quote/GOOP.US.md) ## Comments (1) - **TAISHO99 · 2026-10-30T14:34:39.000Z**: That's enough. --- > **Disclaimer: This article is for reference only and does not constitute any investment advice.**