The Quarter the AI Buildout Outran the Cash Flow: Cloud +82% and a $514B Backlog, Paid For With Equity

Results vs. Consensus

Year-over-Year Comparison

Quarter-over-Quarter Comparison

Revenue Assessment

The composition of the quarter's growth shifted decisively toward Cloud. Of the $23.4B of YoY revenue growth, Cloud contributed $11.1B (48%), the first quarter in Alphabet's history where Cloud supplied nearly half of consolidated growth. The +81.8% Cloud print against a +63% Q1 marks a second consecutive quarter of sharp acceleration (+30% in Q4 2025, +63% in Q1, +81.8% now) on a base that is now approaching $100B annualized. Management attributed the acceleration to GCP growing faster than the segment overall, with core GCP, AI solutions, and AI infrastructure all contributing, plus the first revenue from TPU system sales delivered to customer data centers. Critically, the CFO stated Cloud growth "accelerated meaningfully even after excluding the impact of TPU system sales," pre-empting the obvious quality question.

Services grew 14.5%, with the internal mix telling two stories. YouTube ads (+12.9%) accelerated from Q1's +11% on World Cup strength: 1.7B unique viewers watched World Cup content, 550M of them on television screens, making it the most-viewed World Cup in YouTube's history. Subscriptions, platforms and devices (+15.2%) continues compounding on Google One AI plans and YouTube Premium. Search & other at +16.8% remains the franchise's core engine but broke its four-quarter acceleration streak, and the regional table shows where the strength concentrates: US revenue grew 32% while EMEA grew 12% in constant currency. The US skew is consistent with AI Mode monetization scaling domestically first; it also means the international monetization ramp is still ahead.

Margin Assessment

Consolidated operating margin of 34.0% expanded 160bps YoY but contracted 210bps sequentially, and the sequential direction is the forward signal. The YoY expansion is carried entirely by Cloud (margin +1,490bps YoY to 35.6%, operating income 3.1x); Services margin expanded just 180bps YoY to 41.8% and fell 350bps sequentially, with Services operating income declining QoQ in dollars. Three cost vectors are compounding: depreciation (+42% YoY, and the CFO guided it steeper as the capex wave lands), AI-talent compensation inside R&D (+32%), and content/World Cup costs plus legal charges. The other-cost-of-revenues line (+22%) now carries a new item: inventory costs from TPU system sales, a lower-margin revenue stream that did not exist a quarter ago.

Cloud margin at 35.6% is the structural surprise of the print, again. It has now expanded from 20.7% to 35.6% in four quarters while the segment nearly doubled. But management explicitly guided this line lower for Q3: to bridge the supply gap, Alphabet will rent third-party compute capacity (the arrangement one analyst identified on the call as the SpaceX deal) at costs that "will create modest margin pressure in the near term," stacked on the continuing Wiz integration headwind. We read the guide as Cloud margin returning to the low-30s for two to three quarters, an investment-phase dip rather than a structural break, but the days of 270bps sequential margin gains are paused.

EPS & Cash Flow Assessment

The $9.11 GAAP EPS is a bookkeeping artifact and should be discarded for modeling. The $99.0B gain on equity securities (against $36.9B in Q1) flows from the private-holdings portfolio: non-marketable securities on the balance sheet grew from $68.7B at year-end to $131.5B at June 30, and the cash-and-securities line now separately discloses $87.1B of marketable equity securities. Netting the after-tax gain of roughly $77B (~$6.27/share at a ~22% marginal rate) leaves operational EPS of approximately $2.84, against $2.89 consensus. Alphabet also disclosed $21.1B of new purchases of non-marketable securities in the quarter, a tripling of its investment pace with no attribution of the recipients. The investment book is now large enough to swing GAAP results by entire multiples of operating income in either direction; a market drawdown would produce headline GAAP losses of similar scale.

The cash flow statement is where the quarter's real news lives. Operating cash flow of $39.1B grew 41% YoY, but capex of $44.9B exceeded it, producing the first negative FCF quarter in Alphabet's modern history at $(5.9)B. Add the working-capital signature of the TPU hardware business ($6.7B sequential inventory build to $10.0B) and the funding gap is structural for as long as the guide holds: FY26 capex of $195-205B against perhaps $175-185B of operating cash flow implies a negative-FCF year. The June financing package covers it: $30.5B of common stock, $19.1B of 6.25% mandatory convertible preferred, $24.8B of debt in the quarter, and a $40B ATM authorized. Total cash and marketable securities stand at $242.5B against $98.2B of long-term debt. The balance sheet is fortress-grade; what changed is that the equity holder is now paying for the fortress through dilution (shares outstanding +142M in six months, with buybacks at zero) rather than being paid from it.

Segment Performance

Google Services: $94.5B, +14.5% YoY; margin 41.8%, down 350bps sequentially

Services delivered a solid quarter flattered by the sports calendar and pressured by its own cost base. Advertising grew 14.4% with, per the CFO, growth across all major verticals; retail and finance drove the largest contributions to Search. The World Cup was the quarter's engagement event: Search usage hit an all-time high during the tournament, and YouTube's 1.7B unique World Cup viewers converted into channel takeovers, Gemini-powered sponsorship formats, and YouTube ads growth of +12.9%, accelerating from Q1's +11%. Subscriptions continues to be the quiet compounder, with YouTube subscription revenue growing faster than YouTube ads.

On the advertiser-tooling side, the AI product suite crossed adoption thresholds worth recording: AI Max is out of beta with 500,000 advertisers adopted, campaigns using AI Max or Performance Max see roughly 15% more conversions at similar ROAS per management, over half of SMB customers globally now use AI to create or optimize creative, and shopping ads saw a 20% improvement in surfacing relevant matches from Gemini query understanding. The agentic-commerce layer advanced from protocol to product: UCP is live with Target and Steve Madden, and a universal cart plus Buy with Google Pay on connected TVs extend checkout into Google surfaces.

"People love our AI experiences like AI Mode and AI overviews and they're coming back to search more... since expanding AI mode globally last October, we have surpassed 1 billion monthly active users. And just like AI Overviews, AI mode is driving an incremental increase in search queries overall, and we are now sending billions of clicks to websites every week through AI features in Search." — Sundar Pichai, CEO

Assessment: The revenue engine is intact but the segment's margin trajectory has flipped from expansion to absorption: Services is now carrying AI serving costs, content costs, and marketing for the Gemini app against a growth rate half of Cloud's. We model Services FY26 at $383-390B (+13-14%) with margin in the 42-44% band rather than the 45%+ exit rate Q1 implied.

Google Cloud: $24.8B, +81.8% YoY; margin 35.6%; backlog $514B

Cloud is now the center of the Alphabet story by every measure: growth contribution (48% of consolidated YoY growth), margin trajectory (35.6%, +1,490bps YoY), and forward visibility ($514B backlog, just over 50% converting to revenue within 24 months per the CFO, unchanged from Q1's conversion framing). The demand indicators management stacked up: new-customer acquisition velocity more than doubled YoY, existing customers are exceeding their commitments by more than 50% (an acceleration versus Q1), marketplace transactions grew 7x YoY, nearly 500 customers each processed over 1T tokens in the past year (up from 330 disclosed at Q1), and nearly 90% of the Fortune 100 now use Gemini Enterprise.

"We also began to recognize revenues from TPU system sales, which we delivered to customer data centers for the first time in Q2. Cloud revenue growth accelerated meaningfully even after excluding the impact of TPU system sales." — Anat Ashkenazi, CFO

The backlog arithmetic deserves precision. The $52B sequential add is a striking deceleration from Q1's $222B add, but the Q1 figure included the initial TPU hardware agreements and, we believe, one or more anchor-scale infrastructure commitments; $52B of net new bookings on top of that base is still more than two quarters of current revenue booked in ninety days. At just over 50% converting within 24 months, roughly $260B of revenue is now contractually visible through mid-2028. The composition remains majority "typical GCP contracts for a broad mix of customers," per the CFO, which is the healthiest possible answer to the concentration question.

Assessment: Cloud is compounding at a rate and margin no hyperscaler peer is matching, with the demand book to sustain it. The Q3 margin will step down on the third-party capacity bridge and TPU hardware mix, and we would treat a print in the low 30s as on-plan rather than as deterioration. We raise FY26 Cloud to $102-107B (+63-70% YoY) and FY27 to $150-165B.

Other Bets: $382M revenue, $(1.8)B loss; Waymo's 6th-generation vehicle arrives

Other Bets revenue was roughly flat YoY at $382M with the operating loss widening to $1.8B as Alphabet, per the CFO, continues to scale Waymo and invest in key bets. The quarter's disclosures were operational: Waymo introduced its newest vehicle, the first powered by the sixth-generation Waymo Driver, to public riders; Wing crossed 1M cumulative home deliveries with Walmart, DoorDash and Papa Johns partnerships; and Isomorphic Labs raised over $2B of external capital for its AI drug-design engine, continuing the pattern (Verily, GFiber, now Isomorphic external raises) of moving Other Bets funding off Alphabet's P&L. Asked directly under what conditions Waymo would make sense outside Alphabet, management declined to engage with structure and re-anchored on execution.

Assessment: The segment is economically immaterial to the model but strategically consequential: Waymo remains the most valuable non-consolidated-revenue asset in the company, and the external-capital pattern across the portfolio (Isomorphic's $2B this quarter) suggests the monetization path for the bets is third-party marks rather than Alphabet funding. Those marks, of course, are exactly what is inflating the GAAP P&L through the investment book.

Key Topics & Management Commentary

Overall Management Tone: Management was confident and unhurried on demand, product and infrastructure, treating an 82% Cloud quarter and a $514B backlog as the expected output of a machine working to plan; the supply-constraint framing ("demand still outpaces that investment") was delivered as validation rather than apology. Where the call was less sure-footed was money: the equity raise, the buyback halt and the negative free cash flow were narrated matter-of-factly in prepared remarks but never framed for shareholders, and the capital-structure answer in Q&A leaned on "resilient balance sheet" language rather than a return-on-equity argument. The gap between operational conviction and financial framing is the quarter's tonal signature, and the after-hours tape suggests the market noticed it too.

1. FY26 Capex Raised a Fourth Time, to $195-205B; 2027 Still "Significantly" Higher

The FY26 capex guide has moved from $175-185B (set at the Q4 2025 print) to $180-190B at Q1 and now $195-205B, the fourth consecutive quarter the capex outlook has stepped higher, with the increase attributed to "an acceleration in the delivery of capacity to meet growing demand." Q2 capex of $44.9B doubled YoY, with roughly 60% in servers and 40% in data centers and networking. The 2027 commitment to a significant increase was reiterated without quantification. On the ROIC question, management's framing was unchanged from prior quarters: invest as long as returns are attractive, with the growth rates in Search and Cloud offered as the proof point.

"We are updating our full year 2026 CapEx guidance range to $195 billion to $205 billion, up from our previous estimate of $180 billion to $190 billion. The increase in the range is primarily due to an acceleration in the delivery of capacity to meet growing demand." — Anat Ashkenazi, CFO

Assessment: A $15B raise on a base this large is not itself the story; the story is cadence. Four raises in four quarters means the plan is being re-written upward faster than the company can articulate a ceiling, and the market's tolerance for open-ended spend guides visibly expired this quarter. The demand cover remains genuine ($514B backlog, supply constraints acknowledged industry-wide), but the burden of proof has shifted: from here, each incremental capex dollar needs to arrive with visible revenue conversion, not just backlog accumulation.

2. The Capital-Structure Pivot: $49.6B of Equity, Buybacks at Zero, Debt at $98B

The June financing package is the quarter's most consequential disclosure. Alphabet raised $30.5B of common stock and $19.1B of 6.25% mandatory convertible preferred (the first preferred in company history, now trading as depositary shares), issued $24.8B of debt in the quarter, and authorized a $40B ATM program which the CFO tied specifically to offsetting the tax cost of stock-based compensation. Share repurchases, $13.2B in the year-ago quarter and a decade-long constant of the equity story, went to zero with no commentary. Long-term debt stands at $98.2B, up from roughly $16B a year ago per the CFO's own framing. Shares outstanding rose 142M in the first half.

"We expect the free cash flow will remain under pressure, driven by our investments in technical infrastructure which enables us to capitalize on the AI opportunity and continue to drive attractive returns." — Anat Ashkenazi, CFO

Assessment: This is a regime change and should be modeled as one. The Q1 thesis held that Alphabet's buildout was uniquely self-funded; that pillar is now broken, not because cash generation weakened (OCF grew 41%) but because the spend ambition outgrew it. The equity holder's claim is being diluted at roughly 1-2% annually between the raise, the ATM and halted buybacks, and the 6.25% preferred coupon plus a 4.9x-larger interest expense line are permanent P&L costs. We accept the strategic logic (funding a supply-constrained land grab at the cheapest available cost of capital) while marking our fair value down for it. What we will not accept silently is a second surprise raise: the CFO's "not planning to go back to the equity markets" is now a commitment on our scorecard.

3. Cloud at +82% With a $514B Backlog: The Growth Machine Compounds

Cloud's second consecutive sharp acceleration (30% → 63% → 82% across the last three prints) came with breadth: GCP grew faster than the segment, and core GCP, AI solutions and AI infrastructure all contributed materially. The customer-metrics ladder moved up across every rung disclosed: nearly 500 customers over 1T tokens processed (from 330), more than 2,000 enterprises over 100B tokens, new-customer acquisition velocity doubling, existing customers exceeding commitments by over 50%, and marketplace transactions up 7x.

"Google Cloud's backlog increased by more than $50 billion sequentially, reaching $514 billion in the second quarter. The increase was driven by strong demand for our Enterprise AI offerings. The majority of the backlog is related to typical GCP contracts for a broad mix of customers, and we expect to recognize just over 50% of the total backlog as revenue over the next 24 months." — Anat Ashkenazi, CFO

Assessment: At +81.8% on a $25B quarter, Cloud is growing faster than any $100B-run-rate software business ever has. The backlog deceleration versus Q1's doubling is optics, not deterioration: Q1 captured the one-time inclusion of TPU hardware agreements; Q2's $52B add is the underlying bookings cadence, and it exceeds the segment's entire quarterly revenue by 2x. The thesis question has moved from "is the demand real" (answered) to "what does serving it cost" (Topic 4).

4. The Q3 Margin Trade: Renting Third-Party Compute to Bridge the Supply Gap

The most analytically important guidance of the call: Alphabet will expand its use of third-party data-center capacity in Q3 as a bridge while internal capacity builds, explicitly accepting near-term Cloud margin pressure to avoid turning away customers. In Q&A, management framed the economics as a deliberate trade: a "very high" short-term cost over roughly six months against multi-year customer relationships whose lifetime returns are "very, very attractive." One analyst identified the SpaceX compute arrangement by name in a question; management did not engage the specific counterparty but confirmed the strategy.

"Given the supply constrained environment, we plan to expand the use of third-party capacity in Q3 as a bridging strategy while we build up more internal capacity. This strategy allows us to keep growing our customer base and capture greater overall value. However, it will create modest margin pressure in the near term as we utilize this capacity." — Anat Ashkenazi, CFO

Assessment: The bridge is rational and the disclosure is honest, but it stacks three simultaneous margin headwinds into H2: rented capacity, the Wiz integration drag, and a rising TPU-hardware revenue mix that carries inventory cost rather than service margin. We model Cloud margin at 30-33% for Q3-Q4 versus 35.6% printed, and we flag the risk that "modest" is doing heavy lifting in that sentence. The offset: capacity constraints capping growth would be strictly worse than margin dilution funding it.

5. TPU System Sales Go Live: First Revenue, a $10B Inventory Build, and a New Business Model

Alphabet recognized its first revenue from TPU systems delivered into customer data centers, a quarter ahead of the "later this year" framing from Q1, with the ramp still expected to deliver the vast majority of agreement revenue in 2027. The balance-sheet tell is inventory: $10.0B at June 30 versus $2.4B at year-end, a build the CFO tied directly to the hardware pipeline ("we start building inventory to be able to sell those systems"). Management's allocation hierarchy for TPUs was stated plainly: frontier model development first, then serving Google's own consumer and enterprise products, then external infrastructure demand, including placements into customer and partner data centers such as the Blackstone-related project referenced on the call.

"When we signed the agreements... they would be then reflected in the cloud backlog. Now the vast majority of the $514 billion of cloud backlog is the GCP agreements, but the TPU system sales are reflected in that backlog. We start building inventory to be able to sell those systems... This quarter was a small amount of the total. That total agreement will continue to ramp up throughout 2026, but then you'll see the vast majority of the revenue from that agreement comes through in 2027." — Anat Ashkenazi, CFO

Assessment: The 2027 revenue wave is now physically visible on the balance sheet. A $10B inventory position at hardware-typical margins implies a mid-teens-billions revenue pipeline from deliveries alone, before consumption agreements layered on top. Strategically, this is Alphabet monetizing the one asset NVIDIA's customers cannot buy elsewhere at scale, and the margin question (hardware margin dilutes the segment percentage while growing segment dollars) is a high-class problem. We add explicit TPU hardware revenue to the FY27 model for the first time: $12-20B.

6. Search Decelerates to +16.8%: The Streak Ends, the Lapping Begins

Search & other grew 16.8%, ending a four-quarter acceleration streak and landing a rounding error below Street expectations. The engagement data points remain unambiguous: AI Mode passed 1B monthly active users following its global expansion last October, queries hit an all-time high during the World Cup, AI Mode response costs hit their lowest level since launch on serving optimizations, and Alphabet is "sending billions of clicks to websites every week" through AI features (a datapoint aimed at regulators and publishers as much as investors). The CFO then delivered the forward warning with unusual directness: Q3 begins lapping the Search acceleration that started in Q3 2025, meaning YoY comparisons compress mechanically from here.

"In Q3, we will begin lapping an acceleration in Search performance that began in the third quarter last year." — Anat Ashkenazi, CFO

Assessment: The bear reading (AI chat products finally biting) and the mechanical reading (a +12% comp became a +15% comp) are observationally identical this quarter, which is exactly why the next two prints matter. Our base case is mechanical: two-year stacked growth actually improved sequentially, the US +32% regional print shows monetization deepening where AI Mode is most penetrated, and the Gemini app's 950M MAU has coexisted with accelerating Search for five quarters. We set the signpost now: Q3 Search growth of +12-14% is on-model; below +11% reopens the cannibalization file.

7. The $99B Equity Gain: The Investment Book Becomes a Second P&L

Other income of $98.0B, driven by $99.0B of gains on equity securities, dwarfed operating income of $40.8B. The balance sheet motion: non-marketable securities nearly doubled since year-end to $131.5B, marketable equity securities are separately disclosed at $87.1B, and Alphabet purchased $21.1B of new non-marketable positions in the quarter alone (versus $1.4B a year ago), with no disclosure of the recipients. Management's only commentary was a single sentence attributing OI&E to unrealized portfolio gains.

"Other income and expenses was $98 billion representing a substantial increase from the prior year, primarily due to unrealized gains in our equity securities portfolio." — Anat Ashkenazi, CFO

Assessment: Alphabet now runs a $200B+ investment book whose quarterly marks exceed the operating earnings of almost every company on Earth, financed partly by the same capital raises funding capex, and disclosed at a level of granularity appropriate to a book one-tenth the size. Two quarters running, the gains have flattered headline EPS by 2-3x; the symmetric quarter, when private AI valuations mark down, will produce GAAP losses of the same scale. We continue to model on operational EPS only, and we flag disclosure of the investment book's composition as an open governance ask.

8. Gemini 4 in Pretraining; the Cadence Machine Ships Flash 3.6 and a Cyber Model

The model-velocity disclosures: Gemini 3.6 Flash and 3.5 Flash-Lite launched the day before the call, Gemini 3.5 Flash Cyber (paired with the CodeMender agent for finding and fixing vulnerabilities) entered the security stack, Gemini 3.5 Pro is in testing, and the "most ambitious pretraining run yet" for Gemini 4 has begun. Token throughput on the model APIs reached roughly 22B tokens per minute, up from 16B a quarter ago; 9M developers build with the models monthly; the Gemini app reached 950M MAU with daily actives tripling YoY. In Q&A, management was candid that coding and agentic coding remain areas needing improvement, framed Gemini 4 as the larger base model required "to compete at that frontier level," and committed to a near-monthly release cadence built on top of it.

"We have started our most ambitious pretraining run yet for Gemini 4 and are excited by the progress we are seeing at the frontier... We continue to be supply constrained, a sign of momentum and rapid adoption." — Sundar Pichai, CEO

Assessment: The candid acknowledgment of the coding gap is more reassuring than a denial would have been; coding is the highest-monetizing enterprise model workload, and closing it is a roadmap item with a named vehicle (Gemini 4). The strategic posture is unchanged and correct: own the Pareto frontier of cost-performance (Flash), keep a frontier flagship credible, and let the full stack (silicon to agents) monetize models that rivals must monetize standalone. The 37% sequential token-throughput growth is the cleanest single indicator that model demand is compounding ahead of supply.

9. Gemini Enterprise at ~90% of the Fortune 100; Security Becomes the Second Enterprise Franchise

Adoption metrics stacked this quarter: nearly 90% of the Fortune 100 use Gemini Enterprise; the agent development kit approached 70M downloads; named wins spanned PepsiCo, Intel, HSBC, Bell Canada, Macy's and Signal Iduna. The security platform got equal billing for the first time: 90% of the Fortune 100 are Google Cloud security users, AI workloads scanned and protected grew more than 45% QoQ, and the customer list (Morgan Stanley, TELUS, Texas Children's, Atlassian, government agencies) reads like a critical-infrastructure roster. The Wiz integration is the visible engine; the new cyber-tuned model and CodeMender agent extend it up the stack.

"Nearly 500 cloud customers have each processed more than 1 trillion tokens in the last year... Over the last 12 months, more than 2,000 enterprises consumed over 100 billion tokens." — Sundar Pichai, CEO

Assessment: Enterprise AI adoption has crossed from experimentation to standardization, and Alphabet's agent platform is becoming the default at exactly the moment enterprises pick defaults. Security is the strategically underpriced piece: it is the one enterprise category where AI escalation (threats) mechanically drives budget escalation (defense), giving the Wiz-anchored franchise a demand driver independent of the AI capex cycle itself.

10. YouTube's World Cup: 1.7B Viewers and the Living-Room Flywheel

The World Cup was the largest engagement event in YouTube history: 1.7B unique viewers watched tournament-related content, 550M on television screens, with monetization through takeovers, Gemini-powered sponsorship formats and game-day mastheads. Ask YouTube, the conversational layer over video, engaged more than 140M users on the watch page in June. The subscription complex (YouTube Music & Premium plus TV offerings) is growing faster than YouTube ads, and shoppable formats (Buy with Google Pay on CTV, affiliate programs) are converting the living room from a brand surface into a transaction surface.

"Over 1.7 billion unique viewers watched World Cup-related videos and over 550 million watched them on their televisions. This made the FIFA World Cup 2026 the most viewed World Cup in YouTube history." — Philipp Schindler, CBO

Assessment: YouTube's +12.9% carried a one-time sports tailwind that will not repeat in Q3, so we model YouTube ads back to +9-11% for H2. The durable takeaways are structural: the platform's event gravity keeps compounding, and the subscription line growing faster than ads diversifies the P&L away from ad cyclicality.

Guidance & Outlook

Alphabet does not formally guide revenue; the CFO's outlook commentary for Q3 and FY26:

Implied trajectory: Q3 2026 revenue of roughly $122-126B (+19-23% YoY) on continued Cloud compounding against a slight FX headwind and compressing Search comps; FY26 revenue of $485-495B (+20-23%). Street at: FY26 consensus revenue moves up several billion on the Cloud beat while FY26 EPS models absorb higher depreciation and the new preferred/interest lines; expect consensus operational EPS revisions to be roughly flat despite the revenue raise. Guidance style: the pattern is now established: operational metrics are guided conservatively and beaten; spending is guided optimistically and raised. Both halves of that pattern held this quarter, and the market has started pricing the second half.

Analyst Q&A Highlights

How the GenAI Return Math Has Changed a Year Into the Buildout

The opening exchange asked management to mark its own book: with a year more evidence and spend stepping up again, how has the view of the total GenAI return opportunity and its timing changed? The response re-anchored on early-innings framing across consumer, enterprise and developer surfaces, arguing the opportunity set has expanded rather than narrowed, and the CFO's companion answer on budgeting philosophy confirmed that spend is gated on observed returns and demand rather than a fixed envelope.

Q: "With more time in GenAI products and tools in the market and investment continuing to step up, can you just kind of give us some perspective on how your view on the size of the overall GenAI ROIC opportunity and the timing of the ROIC has changed now versus 1 year ago?"
— Brian Nowak, Morgan Stanley

A: "It feels like we are in very early innings of what feels like secular shift across multiple areas... I would say from an ROIC standpoint, I think we are taking a full stack approach. We are seeing momentum across consumers and enterprises and developers and so on. So it feels like, if anything, over the past year, we've gotten more bullish on the opportunities ahead."
— Sundar Pichai, CEO

Assessment: "More bullish than a year ago" from the team that just doubled quarterly capex is internally consistent, but it also confirms there is no natural ceiling in management's frame: the spend stops when demand stops, and demand is not stopping. Investors seeking a self-imposed limit did not get one; the discipline argument rests entirely on the observed ROIC evidence (Cloud growth, backlog), which for now supports it.

The Capital-Structure Defense: Why Equity, Why Now, and Whether It Repeats

The most consequential exchange of the call asked directly how management thinks about optimal capital structure and the cost of debt versus equity after the June raises. The answer laid out the funding hierarchy (operations first, then debt, then equity), quantified the debt journey (roughly $16B to $100B in a year), justified the equity raise as balance-sheet resilience, and, critically, committed that Alphabet is not planning further equity issuance beyond the ATM program tied to stock-compensation tax costs.

Q: "Following the recent equity and debt raises, can you just help us understand how you think about optimal capital structure and how you view the cost of debt versus equity?"
— Doug Anmuth, JPMorgan

A: "We first look at how much we can support based on cash from operations... And then we look at debt and most recently, we did the equity raise. And we have expanded our debt portfolio quite significantly over the past 12 months. If you've seen a year ago, we were at about $16 billion, and we went to about $100 billion across multiple currencies and geographies... we also want to make sure we have a resilient... balance sheet, which is the rationale behind expanding into the equity markets. At this point, we're not planning to go back to the equity markets with the exception of... the ATM or at-the-market offering that we will do to address the... tax on SBC, which we'll do for some period of time."
— Anat Ashkenazi, CFO

Assessment: The "not planning to go back to the equity markets" line is the single most tradeable sentence of the call and we log it as a standing commitment. What the answer did not contain is notable: no cost-of-capital comparison, no framework for when buybacks resume, and no acknowledgment that issuing stock at 17x forward operating income while halting repurchases inverts a decade of capital-return doctrine. The financial logic is defensible; the shareholder communication was thin.

TPU Allocation: Frontier First, Products Second, External Sales Third

A two-part question probed the learnings from scaling TPU external sales and how the backlog, revenue recognition and margin effects should be modeled. The allocation hierarchy came back explicit: frontier model development is the non-negotiable base load, first-party serving comes next, and external placements (including into customer and partner data centers) absorb what remains. The CFO's rev-rec answer put mechanical structure on the 2027 wave.

Q: "Can you talk about some of your key learnings as you scale your TPU efforts in terms of both the demand for TPUs and how you think about balancing the external demands for TPUs versus the internal demand for custom silicon inside the organization as you look over the next couple of years?"
— Eric Sheridan, Goldman Sachs

A: "Our first priority is making sure we are allocating what we need to compete at the frontier in terms of AGI development... To the extent people want it as infrastructure, we are balancing it by increasingly looking at opportunities to put TPUs in their data centers or in other data centers like the project we are doing with Blackstone... so we are using that as an ability to balance and make sure we are allocating as much of the available compute for frontier model development as well as serving both our consumer business and our Enterprise business."
— Sundar Pichai, CEO

Assessment: The Blackstone reference confirms the TPU placement model is expanding beyond direct customer data centers into third-party-operated capacity, a structure that scales faster than either building or shipping alone. The allocation hierarchy also quietly explains the third-party compute bridge: external TPU commitments are being honored even while internal serving is capacity-constrained, because the hardware relationships are strategic. Supply, not demand, is the binding constraint on this business line through at least mid-2027.

Model Release Cadence and the SpaceX Compute Deal

A question on the model war asked how Alphabet can accelerate its release cadence, referencing the third-party compute arrangement with SpaceX as evidence of the lengths being taken, and whether the cost-efficient Flash tier is the right end of the market as the low-cost segment crowds. The answer committed to a faster cadence explicitly built on the Gemini 4 base.

Q: "We all saw the third-party compute deal with SpaceX. What else are we doing to kind of speed up the pace of model releases compared to some of the other folks out there? And... the Flash models... do you view that as the right end of the market to be in, given how crowded the lower cost end of the model space has become?"
— Ross Sandler, Barclays

A: "We really think about the Pareto frontier. We want to make sure for our customers we are offering... best models at various price points... We are really focused, putting a lot of effort into Gemini 4. It's a very ambitious effort. We wanted to compete at the frontier level of where the frontier will be when Gemini 4 comes out... But with that, we are creating a baseline on which... you will see us rapidly iterate... picking up pace and releasing models almost at a monthly cadence is part of our road map."
— Sundar Pichai, CEO

Assessment: The monthly-cadence commitment is aggressive and now measurable. The Pareto-frontier answer to the commoditization question is the right economic frame: Alphabet does not need to win every price tier, it needs the cost-performance frontier to run through its own silicon, which it currently does. The unaddressed SpaceX reference effectively confirms the arrangement without disclosing its terms, leaving the cost side of the Q3 margin bridge unquantified.

What Protects the Franchise If Models Commoditize

The moat question was posed directly: if every lab reaches similar model capability, what sustains Alphabet's growth? The answer argued models are becoming ingredients inside solutions (cybersecurity, data analytics, agent platforms) where distribution, data integration, trust and serving infrastructure decide winners, followed by the CFO addressing whether TPU sales dilute or expand Cloud economics.

Q: "Can you talk a bit about what do you think the moats are... what is your strategic advantage... that even if everyone gets to the same type of model capabilities, what gives you the edge to keep growing?"
— Michael Nathanson, MoffettNathanson

A: "We don't break out margins for any specific products or infrastructure component and certainly, there are benefits from designing and manufacturing our own chips. But the way to think about it is, this is an expansion of our total addressable market... But overall, if you think about the cloud margins, obviously, a really fantastic expansion of margin to 35.6% this quarter."
— Anat Ashkenazi, CFO

Assessment: "TAM expansion" is the honest answer on TPU economics: hardware sales will run below GCP service margins but add operating-income dollars the segment otherwise could not touch. The moat answer (solutions over models) matches where the numbers are: 90% Fortune 100 penetration in both Gemini Enterprise and security is distribution lock-in that survives model-capability convergence. We consider this the strongest version of the bull case available if commoditization accelerates.

The Economics of the Six-Month Bridge

A question pressed on where the capacity constraints bind hardest and how compute is triaged across Search, model training and Cloud. The answer reaffirmed the allocation hierarchy and then quantified the logic of the third-party bridge in the frankest cost language of the call.

Q: "On these third-party deals and kind of the bridge in capacity, is there a particular objective where the constraints are most severe? Or is it really broad-based?... is there a way to contextualize how you're now thinking about the trade-off between allocating compute to search versus model training versus for the GCP business?"
— Mark Shmulik, Bernstein

A: "They are on the margin... very, very large customers of ours on cloud, who we are trying to support... through this extraordinary moment. And the incremental opportunities they are bringing to us, while a short-term cost over a few months may be very high, in the lifetime of the deal as we bring more capacity on, is highly ROI positive... So are you willing to take upfront a 6-month deal to be able to serve that customer in what is a multiyear opportunity where the margins and the returns are very, very attractive over that multiyear horizon."
— Sundar Pichai, CEO

Assessment: "A short-term cost over a few months may be very high" is materially blunter than the CFO's "modest margin pressure," and we weight the CEO's version. The six-month framing at least bounds the duration: this is a bridge with a stated end condition (internal capacity landing through H2), not a structural outsourcing dependency. Q4's Cloud margin print will show which adjective was right.

Under What Conditions Waymo Leaves the Nest

With Waymo scaling and its sixth-generation vehicle launching, the structural question arrived: what factors would management weigh in changing Waymo's corporate structure? The answer engaged the premise not at all, re-anchoring on execution inside the existing bet structure.

Q: "When you think about Waymo, what are the key factors you would consider when evaluating a change in corporate structure for Waymo?... Under what conditions would it make sense for Waymo to live outside of Alphabet?"
— Ken Gawrelski, Wells Fargo

A: "We are really focused on executing and scaling up Waymo. We have set it up with our bet structure and given the team a lot of support... our ability to think and plan long term and invest in the business and give them that long-term road map... is all hugely valuable. We're really focused on scaling the business right now and executing to that extraordinary potential."
— Sundar Pichai, CEO

Assessment: A complete non-answer, which is itself informative: with Isomorphic raising $2B externally this quarter and Verily and GFiber already deconsolidated, the playbook for monetizing bets exists and management chose not to apply its logic to Waymo even hypothetically. Waymo optionality remains real but should carry no near-term catalyst weight in the model.

What They're NOT Saying

1. The buyback halt: Repurchases went from $13.2B a year ago to zero, and neither the prepared remarks nor nine Q&A exchanges addressed it. A decade of capital-return doctrine ended in silence; the omission is the loudest disclosure in the filing.
2. 2027 capex magnitude: "Increase significantly" is now three quarters old as a formulation, with details perpetually "at a later date." On a $200B base, the range of interpretations spans $50B+ of annual FCF impact.
3. What the $21.1B of new private investments bought: Purchases of non-marketable securities ran 15x the year-ago pace with zero attribution, while the $99B markup similarly carries no holding-level disclosure. The investment book is now material to the equity story and effectively unexplained.
4. The SpaceX arrangement's cost: An analyst named the deal; management confirmed the bridging strategy without terms, duration specifics, or the margin math. "Modest pressure" (CFO) versus "may be very high" (CEO) is the disclosure gap, unresolved.
5. Search's miss and AI Mode's monetization rate: No acknowledgment that Search landed below Street expectations, and still no coverage-ratio or monetization-parity metrics for AI Mode at 1B MAU. The engagement metrics are abundant; the economics remain asserted.
6. Apple Foundation Models and the Anthropic TPU commitment: The two anchor relationships that headlined prior quarters went entirely unmentioned. Presumably both are progressing inside the backlog, but the silence on the two largest named commitments is a continuity gap.
7. DOJ remedies: Silent for the sixth consecutive quarter, with the remedies framework now in effect and compliance costs (data sharing, default-agreement changes) presumably running through the P&L unquantified.
8. Preferred and ATM dilution mechanics: No conversion-ratio guidance, no expected ATM utilization pace, no pro-forma share-count trajectory. The 12,309M diluted share count is the first print of a rising series whose slope management declined to sketch.
9. Wiz standalone performance: Second consecutive quarter with no ARR or growth disclosure since the $32B close; the security-adoption metrics impress but cannot be tied to the acquisition price.

Market Reaction

• Pre-print setup (Jul 22 close): $342.09, +9.3% YTD (from $313.00 at year-end), in line with the S&P 500's +9.5% YTD; +78.8% trailing twelve months but -2.2% over the trailing 30 days, with the stock 15% below its 52-week closing high of $402.62. Positioning into the print was noticeably cooler than into Q1: the AI-financing debate (sparked industry-wide by the June capital raises) had already compressed the multiple, and GOOGL entered the print as a market performer on the year rather than a leader.
• After-hours (Jul 22): The stock was roughly flat immediately on the release, with the revenue and Cloud beats absorbing the equity-gain-distorted EPS optics. The sell-off developed as the call progressed: the capex raise to $195-205B, the Q3 third-party-capacity margin commentary, and the free-cash-flow-under-pressure guidance turned the tape decisively negative into the evening.
• T+1 session (Jul 23, in progress at publication): Opened $321.13, down 6.1%, and traded to a morning low of $314.94; at approximately 11:30 AM ET the stock is near $316, down roughly 7.6%, with volume already matching the prior full session. The session had not closed at publication.

The reaction is a mirror image of Q1's +10%: the same demand evidence (backlog, Cloud acceleration) that earned the capex guide a re-rating in April is being discounted against the financing evidence (negative FCF, equity issuance, halted buybacks) in July. The composition of the sell-off tells you what is being repriced: this is not a demand scare (Cloud beat by 11%) but a returns-per-share scare, in which each incremental capex dollar is now visibly funded by dilution rather than retained cash. Entering the print flat to the market YTD, GOOGL has surrendered nearly all of its year-to-date gain intraday, leaving the stock at roughly 17x FY27 operating income on an enterprise-value basis: the cheapest AI-infrastructure franchise in the megacap complex on our numbers.

Street Perspective

Debate: Is the equity-funded buildout a discipline break or a rational cost-of-capital call?

Bull view: Funding a supply-constrained land grab with equity at all-time-high-adjacent prices is textbook corporate finance: the $514B backlog is contracted demand, the marginal ROIC on capacity is demonstrably high (Cloud margin +1,490bps while doubling), and terming out the balance sheet now avoids financing at worse terms later. The dilution is 1-2% annually against a franchise compounding revenue at 24%.

Bear view: Companies that issue equity to fund capex have historically done so at cycle tops, and the sequence (buybacks halted, preferred issued, ATM standing, FCF negative, capex guide raised four straight quarters with no ceiling articulated) is the classic pattern of spend outrunning governance. If AI demand plateaus in 2027-2028, shareholders own more shares of a lower-margin business with $100B of debt and a depreciation wave cresting.

Our take: The bull case holds while two conditions hold: backlog keeps growing faster than revenue recognizes out of it, and the CFO's no-more-equity commitment survives. Both are currently true, and the burden of the bear case rests on a demand reversal for which there is no present evidence. We side with the bulls at this price, having marked our fair value down for the dilution that is already contractual.

Debate: Is the Search deceleration comp mechanics or the first crack in the moat?

Bull view: A 16.8% print against the quarter where the AI-driven acceleration began lapping is arithmetic, not deterioration: two-year stacked growth improved, US revenue grew 32% where AI Mode is deepest, queries hit all-time highs, and AI Mode crossed 1B users while serving costs fell to launch-era lows. The CFO flagging the Q3 comp proactively is expectation management, not warning.

Bear view: The first miss and first deceleration of the AI-Mode era arrived in the same quarter the Gemini app's daily actives tripled, which is precisely the substitution signature the cannibalization thesis predicts: engagement migrating to a surface that monetizes at a fraction of Search's rate, with the revenue effect lagged.

Our take: The mechanical reading fits the data better today: substitution would show in query volumes, which made all-time highs. But this is now a live empirical question with a clean test: Q3 Search growth of +12-14% confirms comp mechanics; below +11% and the cannibalization file reopens. We hold the bull side with a dated checkpoint rather than conviction into perpetuity.

Debate: Can Cloud margins survive the bridge, the hardware mix, and Wiz simultaneously?

Bull view: 20.7% to 35.6% in four quarters demonstrates the structural unit economics of the vertically integrated stack; the three headwinds are all elective, bounded and disclosed (a six-month bridge, an integration year, a deliberately-taken hardware mix), and a segment adding $11B of YoY revenue can fund all three while growing operating income 3x.

Bear view: The exit-rate margin is unknowable because the clean number no longer exists: from Q3 onward, rented capacity, hardware cost-of-goods and acquisition amortization are permanently blended in, and "modest pressure" from management contrasts with the CEO's own "very high" short-term cost framing. The 35.6% may prove the cycle peak rather than the base.

Our take: We model Q3-Q4 Cloud margin at 30-33% and treat anything in that band as on-plan. The distinction that matters is dollars versus percentage: Cloud operating income can grow 50%+ through the bridge quarters even at a 31% margin, and the bridge exists because demand exceeds capacity, the opposite of a margin problem's usual cause. The genuine risk is duration: if "six months" becomes structural, the segment's margin narrative resets lower.

Model Update Needed

Valuation impact: At roughly $316 intraday, market capitalization is approximately $3.88T on 12.31B diluted shares; net of $242.5B cash and marketable securities against $98.2B of long-term debt (and attributing no value to the $131.5B non-marketable book), enterprise value is approximately $3.74T. On our revised FY27 estimates (roughly $575B revenue and $215B operating income at the midpoints), GOOGL trades at 6.5x EV/Sales and 17.4x EV/Operating Income: the lowest operating-income multiple in our megacap AI coverage, for the fastest-growing cloud franchise in the group. Applying 22-26x EV/Operating Income on FY27 (a discount to the 24-28x we used at Q1, reflecting the financing regime and share-count growth) yields $4.7-5.6T of EV, or roughly $390-460 per share on the enlarged share count. Our 12-month fair-value estimate moves to $390-430 (from $400-440): the operational estimates went up, the per-share claim on them went down, and the band's midpoint implies roughly +30% from the intraday quote. Upside to $470+ requires the Q4 Cloud margin confirming the bridge was temporary and the 2027 capex guide arriving with commensurate backlog growth.

Thesis Scorecard Post-Earnings

Overall: The operational thesis strengthened again: Cloud, backlog, TPU hardware, Wiz and UCP all confirmed or exceeded, and the lone operational wobble (Search's deceleration) has a credible mechanical explanation with a dated test. The financial thesis changed shape: the self-funded-buildout pillar that distinguished Alphabet from every AI-capex peer is broken, replaced by an externally financed model whose costs (dilution, preferred coupon, 4.9x interest expense, negative FCF) are now permanent model inputs. We grade the quarter as the strongest operational print of our coverage delivered in the weakest financial packaging of our coverage.

Action: Maintaining Outperform, sixth consecutive call, with fair value trimmed to $390-430 (from $400-440) for dilution and financing costs. The 7%+ de-rate against an 82% Cloud quarter is, on our numbers, the market charging Alphabet twice for the same capex: once through the estimates and again through the multiple. Position sizing 5-7%, unchanged. Downgrade triggers: Q3 Search growth below +11%; Cloud growth below +60% with backlog declining QoQ; any equity issuance beyond the disclosed ATM program; FY27 capex guided above $270B without commensurate backlog growth. Position-expansion triggers: Q3-Q4 Cloud margin holding at or above 31% through the bridge; TPU hardware revenue ramping on the disclosed schedule into the $10B inventory; the stock holding below $330 while the operational pillars stay confirmed.