Alphabet Q3 2026 Earnings: Capex and Cloud Results

Alphabet held 2026 capex at $195–205 billion after a $51.4 billion third quarter, told investors spending will rise “significantly” in 2027 without a number, and used Cloud’s doubling and a $561 billion backlog to argue the AI build is already earning its keep—even as free cash flow went negative and third-party compute starts to press Cloud margins.

Results were released October 28, 2026, for the quarter ended September 30. Alphabet does not publish a separate quarterly shareholder letter. The primary documents are the earnings release (8-K Exhibit 99.1), the CEO’s edited prepared remarks on the Google blog, the earnings-call transcript, and the Form 10-Q filed October 29, 2026.

Capex, the mix, and what 2026–27 actually says

The constraint is no longer “will they spend,” it is the mix and the cash timing. Purchases of property and equipment were $51.355 billion in Q3 (management rounded this to $51.4 billion), more than double the year-ago quarter, and $131.953 billion for the first nine months. Operating cash flow was $42.290 billion, so free cash flow—operating cash flow less capex—was negative $9.065 billion. Management also flagged a TPU inventory build and a $5.2 billion cash payment of a finalized European Commission fine as additional drains on cash.

Anat Ashkenazi’s split of the quarter’s technical-infrastructure spend, which she said was the vast majority of capex:

• About 65% servers and networking equipment, generally depreciated over six years.
• The remaining 35% data centers, with useful lives of seven to 40 years.

That is a flow mix, not the stock. The 10-Q says that as of September 30, 2026, about 60% of technical-infrastructure assets on the balance sheet were servers and network equipment, with the rest data-center land, buildings, and related assets.

Full-year 2026 capex guidance was left unchanged at $195 billion to $205 billion. With $132.0 billion already spent through September, landing inside that range implies roughly $63–73 billion of Q4 capex (arithmetic from the filing and the guidance range, not a company forecast).

On 2027, the company gave direction, not a number. Ashkenazi: “We continue to expect our CapEx to increase significantly in 2027, and we’ll provide further details on our Q4 call.” Figures circulating in the press that 2027 could exceed $300 billion are investor speculation, not guidance. Depreciation and data-center operating costs, including energy, are expected to keep weighing on the P&L as the new capacity comes online.

Longer-dated supply is already locked in. The 10-Q reports $802.1 billion of contractual obligations with remaining terms greater than one year as of September 30, 2026—mostly long-term supply agreements for technical infrastructure and inventory components, with the significant majority of those payments expected through 2031, plus energy agreements that can run into the 2050s.

Google Cloud: growth, margin, backlog

Cloud is the line where AI shows up as both revenue and profit at the same time. Google Cloud revenue was $30.709 billion, up 103% from $15.157 billion. Operating income more than tripled to $11.121 billion, and operating margin expanded to 36.2% from 23.7% a year earlier. Management said growth was led by Google Cloud Platform—enterprise AI infrastructure, enterprise AI solutions, and core GCP—and that TPU system sales were a meaningful contributor. On the call, Thomas Kurian said GCP was the vast majority of Cloud revenue, and that Cloud growth excluding TPU system sales was still very strong.

Backlog, from the 10-Q (Note 2), not the rounded call figure:

• Total remaining performance obligations: $567.7 billion.
• Of which Google Cloud: $560.8 billion (Ashkenazi rounded this to $561 billion).
• Just over 55% expected to be recognized as revenue over the next 24 months, up from “just over 50%” at June 30, when Cloud backlog was $513.9 billion of a $519.5 billion total.
• Includes $16.8 billion tied to a multi-year commercial agreement with Crux AI, the Blackstone joint venture and a related party.

Recognition timing depends on contract duration, Alphabet’s ability to deliver, and when customers actually use the capacity—so the 55% is an estimate, not a schedule.

The margin expansion is not yet fully stress-tested against the bridging strategy. Third-party compute capacity only came online at the very end of Q3, so it had limited negative impact on the quarter’s 36.2% margin. Ashkenazi said usage ramps in Q4 and will create near-term Cloud margin headwinds.

What Pichai said on demand, constraints, TPUs, and returns

Pichai’s framing is that demand is not the scarce input—capacity and allocation discipline are. From the earnings release and the blog/call remarks:

• Alphabet is at a half-trillion-dollar annual revenue run rate, with revenue up 25% to $127.577 billion, the 13th straight quarter of double-digit growth.
• “Our AI investments are driving overall Search query growth, including commercial queries. In fact, our year-over-year query growth rate is the highest it’s been in the last five years.” Search and other revenue was $63.350 billion, up 12%.
• On Cloud: customers see Google as the leading provider of AI infrastructure and solutions, “so there’s huge demand. We are working hard to bring more capacity online to meet this long-term demand.”
• TPU systems: “strong demand… across a wide range of customers, and we expect that to continue as we scale.”
• Capacity is already binding on the model side. Argon, launched about two weeks before the call, ramped on developer platforms three-and-a-half times faster than Gemini 3 in its first week, “even with capacity constraints.” First-party model APIs were processing nearly 32 billion tokens per minute, up 40% quarter over quarter and more than four times a year ago. The Gemini app reached one billion users.

On allocation, in Q&A, he said compute allocation is something they spend a lot of time on because of the forward planning required. He rejected the idea of a short-term rationing exercise: each business runs long-range plans under a “disciplined ROIC framework.” Models are the foundation across Search, YouTube, Cloud, and Waymo, and the full stack is meant to create both efficiency and optionality. Demand signals are “very, very robust,” and they are “investing for that moment.”

On economics of selling systems outside Google’s own data centers, he said the ability to deploy TPUs both in Google data centers and as systems in customers’ data centers scales the product and “directly translates into better economics.”

What Ashkenazi said on constraints, TPU cash timing, and returns

Ashkenazi separated three clocks that the market was collapsing into one number: capex, Cloud margin, and cash from TPU system sales.

Supply and margins. The environment is still supply-constrained. Alphabet started using third-party compute at the end of Q3 and will ramp it in Q4 as a bridge while internal capacity is built. That capacity is expected to pressure Cloud margins in the near term. Higher depreciation and data-center costs such as energy will also hit the P&L.

2026 and 2027 spend. The $195–205 billion 2026 range is unchanged. 2027 capex is still expected to increase significantly; the number comes on the Q4 call.

TPU systems are not traditional capex. A system sale is hardware, compute, storage, networking, and software for training and inference—not a chip. When a contract is signed it hits backlog first; Alphabet then builds inventory before it bills. Cash goes out before revenue comes in. Some system sales have already been recognized; sales are expected to keep scaling in Q4, with a meaningful revenue ramp in 2027. Kurian added that inventory is procured against signed contracts, and that placing systems in customer or partner data centers (Oracle; Crux AI with Blackstone) both widens the market—including customers who cannot move data to the cloud—and reduces the need for Google to pre-build hosting for those systems.

Returns. Before Q&A she addressed ROIC and free cash flow directly. The pitch is that the same infrastructure is monetized across products that already reach billions of users and enterprises: Search at its highest query growth in five years; Google Cloud with “tens of billions of dollars in AI revenues year-to-date”; Google One as the fastest-growing subscription, driven by AI plans. She said they expect an attractive ROIC that stays meaningfully above the cost of capital, and strong operating cash flow, even while technical-infrastructure investment and the TPU inventory build pressure free cash flow.

What the filings add that the call compressed

The 10-Q (filed October 29, 2026) is the cleaner source on backlog, cash, and commitments than the rounded call language.

• Revenue backlog $567.7 billion, Cloud $560.8 billion, just over 55% inside 24 months, including $16.8 billion with Crux AI.
• Q3 purchases of property and equipment $51.355 billion; nine-month total $131.953 billion; Q3 operating cash flow $42.290 billion.
• Inventory on the balance sheet rose sharply (the cash-flow statement shows a $10.959 billion inventory use of cash over nine months), consistent with the TPU build-ahead-of-bill story.
• Accrued purchases of property and equipment were $21.398 billion at quarter-end, versus $8.877 billion at year-end 2025—capex cash and capex accruals are not the same thing.
• Long-term contractual obligations with terms over one year: $802.1 billion.
• In Q3 the company issued senior notes for net proceeds of $28.7 billion. Long-term debt was $125.1 billion at quarter-end. The earnings release also notes ATM equity issuance in the quarter ($4.3 billion of Class A and $8.1 billion of Class C) and $27.6 billion still available under the program. There were no share repurchases in the first nine months of 2026, versus $39.810 billion in the year-ago nine months.
• The 10-Q states that Alphabet has signed a limited number of agreements to supply TPU systems to customers who need or provide on-premises infrastructure, began recognizing revenue from those agreements in Q2 2026, and expects the significant majority of that revenue in 2027—consistent with, and slightly more precise than, the call’s “meaningful revenue ramp in 2027.”

Consolidated operating income was $40.576 billion, up 30%, at a 31.8% margin (the release table rounds this to 32%). Net income available to common stockholders was $26.170 billion, or $2.12 diluted EPS, pulled down mainly by an $8.2 billion other-income loss, largely an unrealized loss on the SpaceX stake. Google Services operating margin was 40.7% on $96.302 billion of revenue.

How to read this if you are underwriting the spend

The company’s own case is that returns are already visible in query growth, Cloud AI revenue, and subscription mix, and that ROIC is being managed on a multi-year basis rather than against one quarter of negative free cash flow. The open items are the ones they deferred: the 2027 capex number, how large the Q4 third-party compute drag on Cloud margin is, and how fast TPU system inventory converts to cash in 2027. A competitor or investor should treat the $195–205 billion 2026 range as firm, treat “significantly higher in 2027” as directional only until the Q4 call, and not confuse the 65/35 quarterly technical-infrastructure spend split with the 60/40 stock of assets already on the books.

Sources:

1. Alphabet Q3 2026 earnings release (October 28, 2026) — https://abc.xyz/files/doc_financials/2026/q3/2026q3-alphabet-earnings-release.pdf
2. Alphabet Q3 2026 earnings call transcript (October 28, 2026) — https://s206.q4cdn.com/479360582/files/doc_events/2026/Oct/28/2026_Q3_Earnings_Transcript.pdf
3. Alphabet investor relations, Q3 2026 earnings call page — https://abc.xyz/investor/events/event-details/
4. Sundar Pichai, “Q3 2026 earnings call: Remarks from our CEO” (October 28, 2026) — https://blog.google/company-news/inside-google/message-ceo/alphabet-earnings-q3-2026/
5. Alphabet Form 10-Q for the quarter ended September 30, 2026 — https://www.sec.gov/Archives/edgar/data/1652044/000165204426000084/goog-20260930.htm
6. CNBC, Alphabet Q3 2026 earnings coverage (October 28, 2026) — https://www.cnbc.com/2026/10/28/alphabet-googl-earnings-q3-2026.html
7. Reuters, Alphabet Q3 2026 (October 28, 2026) — https://www.reuters.com/business/alphabets-google-cloud-unit-beats-quarterly-revenue-estimates-steady-ai-demand-2026-10-28/

---

Recent Findings Supplement (October 2026)

Alphabet reported Q3 2026 results on October 28, 2026 (quarter ended September 30), with total revenue of $127.58 billion (+25% YoY, reported and constant currency), marking its 13th consecutive quarter of double-digit growth and reaching a ~$500 billion annual run rate.[1][2][3]

Google Cloud drove outsized growth amid heavy AI infrastructure investment, while core Search showed moderating but still solid expansion. The results featured a significant acceleration in capex and negative free cash flow, with management maintaining 2026 capex guidance while signaling a sharp step-up in 2027 alongside TPU system sales ramping meaningfully.[4][5]

Capex reached $51.4 billion in Q3 (more than double YoY), primarily for technical infrastructure supporting AI; FY 2026 guidance held at $195–205 billion, with expectations of a significant increase in 2027 (one reference to exceeding $300 billion).[4][6][7]

Approximately 65% of technical infrastructure spend went to servers and networking equipment (useful lives ~6 years), with ~35% to data centers. Free cash flow turned more negative at –$9.07 billion (vs. expectations near breakeven), driven by capex, TPU inventory build, and other factors like a European fine payment; TTM FCF fell 73% to $19.7 billion.[5][8]

This reflects Alphabet’s aggressive buildout to address sustained AI demand, shifting from prior years’ lower spending levels. Competitors or entrants face pressure to match scale or find differentiated efficiency, as the capex intensity highlights the capital barriers in AI infrastructure.

Google Cloud revenue surged 103% to $30.71 billion (beating estimates), with operating income more than tripling to $11.1 billion and margin expanding to 36.2% (from 23.7%).[9][8][10]

Growth was led by Google Cloud Platform (GCP) strength in enterprise AI infrastructure/solutions and core services; TPU system sales were a “meaningful contributor.” Backlog reached $561 billion (Alphabet total ~$568 billion), with just over 55% expected to be recognized over the next 24 months.[6][9]

TPU inventory rose sharply (to $13.3 billion in one analysis) as the company builds ahead of deliveries, with ~100 customers including AI labs like Anthropic. This positions Cloud as the primary growth engine, with TPU systems extending the moat beyond internal use into external sales. New entrants or smaller players must navigate supply constraints and compete on full-stack AI offerings (silicon to software) to capture similar enterprise commitments.

No detailed updated full-year 2026 revenue guidance was provided beyond capex; Q4 faces larger FX headwinds and lapping of prior Search acceleration.[11][12]

Initial 2027 commentary emphasized capex rising significantly, TPU system sales continuing to scale in Q4 with a “meaningful revenue ramp” in 2027, and AI-driven productivity gains (Pichai described 2027 as a potential “golden year of productivity” from models like Argon and Carbon, citing internal gains such as 35% higher code deliveries).[5][13]

Near-term Cloud margin pressure is expected from third-party compute capacity (ramping in Q4 after starting end of Q3) as a bridge amid supply constraints.[9][12]

Sundar Pichai highlighted robust demand signals and AI momentum: AI investments drove the highest Search query growth rate in five years (including commercial queries); Cloud saw “extraordinary demand” for AI infrastructure/solutions; TPU systems showed strong customer demand across AI labs, enterprises, and others, with scaling expected to continue.[14][15]

He noted ongoing work to bring more capacity online and careful compute allocation between internal frontier needs, consumer/enterprise serving, and external opportunities (e.g., partnerships like Oracle and Crux AI JV).[8]

Anat Ashkenazi emphasized capex discipline within the unchanged 2026 range while flagging 2027 increases, TPU inventory build pressuring FCF, and near-term margin headwinds from third-party capacity; she underscored the $561 billion Cloud backlog and >55% recognition timeline.[16][12]

Thomas Kurian (Google Cloud CEO) reinforced that GCP drove the majority of Cloud growth (excluding TPU systems impact), with broad demand from digital natives, AI labs, enterprises, and financial services; TPU systems sales are scaling as a distinct offering.[9][17]

Overall, the earnings underscore accelerating AI-driven demand (especially in Cloud and via TPUs) offset by elevated capex intensity and near-term cash flow/margin pressures from supply bridging and inventory. This creates a high-bar environment where scale, custom silicon (TPUs), and full-stack capabilities differentiate leaders, while pressuring returns timelines into 2027+.