Alphabet_Earnings_Q2_FY2026
| Alphabet Inc. Nasdaq: GOOGL / GOOG · Q2 FY2026 (quarter ended June 30, 2026) Reported July 22, 2026 · Prepared September 8, 2026 | EARNINGS UPDATE |
SECTION 1 — EXECUTIVE SNAPSHOT
| THESIS STATUS MONITORING CONVICTION CHANGE Unchanged on the Search engine; lower on capital discipline. The operating business did what the thesis expects. The obligation side of the balance sheet did not. BIGGEST CHANGES THIS QUARTER • Purchase commitments and other contractual obligations rose from $332.4B (Mar 31) to 478.6B in one quarter, against a 5.9B (Q2 2025: +$5.3B); FY2026 capex guidance raised a second time, to $195–205B from $175–185B in February. • Cloud backlog growth decelerated hard: +93% q/q in Q1 2026 to +11.2% q/q in Q2 2026, even as the obligation side accelerated. • Two risk factors appear for the first time: TPU-system supply (“excess capacity that we cannot easily redeploy”) and access to the capital markets. INVESTOR TAKEAWAY Nothing this quarter weakens Alphabet’s earnings power — Search compounded 17%, Cloud 82%, operating margin 34% — but the company committed roughly half a trillion dollars of incremental future obligations against $52B of incremental contracted revenue, so what changed is the risk profile behind the compounding, not the compounding itself. WHAT TO MONITOR NEXT QUARTER • Purchase commitments vs. Cloud backlog: does the obligation line grow faster than contracted revenue for a second consecutive quarter? • Cost-per-click direction in Q3 as Search laps the Q3 2025 acceleration — a fourth consecutive deceleration would be the first hard evidence of answer substitution. • Cloud operating margin after third-party bridging capacity and Wiz: does it hold above ~33%? |
Sources retrieved: FY2026 Q2 Form 10-Q (filed 2026-07-23), FY2026 Q1 Form 10-Q, FY2025 Q3 Form 10-Q, FY2025 Form 10-K, Q2 2026 earnings release (Exhibit 99.1, SEC), and earnings-call transcripts for Q3 2025, Q4 2025, Q1 2026 and Q2 2026. Knowledge base holds 10-Ks FY2016–FY2025, 10-Qs through Q2 FY2026 and four quarters of transcripts — financial comparison depth is full; verbatim management-language comparison is limited to four quarters. Confidence that Q2 FY2026 is the latest reported period: high (Q3 FY2026 not yet reported as of September 8, 2026).
SECTION 2 — KEY CHANGES THIS QUARTER
What management emphasised MORE than before
| Change | Why it matters |
|---|---|
| TPU systems as a merchant product (28 transcript mentions vs. 8 in Q3 2025) | Reframes Cloud from a rental business to a hardware business with different margin, inventory and working-capital economics. Inventory rose $2.4B → $10.0B in six months. |
| Supply constraint stated repeatedly and explicitly, plus a new “third-party bridging capacity” strategy for Q3 | Management is now buying someone else’s compute at negative near-term margin to hold customers. Scarcity is being converted into a cost, not just a price. |
| Free cash flow re-entered the CFO script (three mentions; absent from Q1 2026) | Pre-managing expectations for a negative print. “FCF will remain under pressure” is a new explicit forward statement. |
| Capital structure and balance-sheet resilience (debt ~100B in twelve months) | Funding regime has shifted from internally generated cash to external capital — a change in the character of the business, not just the financing. |
What management emphasised LESS or stopped discussing
| Change | Why it matters |
|---|---|
| Paid subscription count — 300M (Q3 2025), 325M (Q4 2025), 350M (Q1 2026), then not disclosed | A headline consumer KPI given three quarters running, dropped in the quarter it would have lapped its toughest comparison. Subscriptions revenue growth also slowed from 19% to 15%. |
| Gemini Enterprise paid seats (8M, Q4 2025) and paid MAU growth (+40% q/q, Q1 2026) | Replaced by “nearly 90% of the Fortune 100 using it” — a penetration statistic, not a paid-usage one. Adoption breadth is now being shown instead of monetised depth. |
| AI Mode / AI Overviews as named products (30 and 15 mentions in Q3 2025, 7 and 5 now) | Merged into “one seamless Search experience.” Defensible product logic, but it also removes the ability to track the monetisation of AI surfaces separately. |
| Billion-dollar Cloud deal counts (given in Q3 2025 and Q4 2025), and backlog emphasis (15 mentions → 8) | Both disclosures faded in the quarter backlog growth decelerated to +11.2% q/q. Management de-emphasises the metric as it stops flattering. |
KPIs that became more or less important
| KPI | Direction | Signal |
|---|---|---|
| Purchase commitments and off-balance-sheet backstops | More important | The fastest-moving number in the filing and the least discussed on the call. |
| Paid-click vs. cost-per-click split | More important | The only quantified test of whether AI queries monetise as well as the queries they replace. |
| Cloud backlog q/q growth | Less emphasised | Decelerated from +93% to +11.2% q/q; management gave it half the airtime of last quarter. |
| Paid subscriptions (350M) | Withdrawn | Removed after three consecutive quarters of disclosure. |
| Tokens per minute (22B) | Maintained | Still given, but sequential growth decelerating: +43%, +60%, +38% over the last three quarters. |
Risks that increased or decreased
| Risk | Direction | Implication |
|---|---|---|
| TPU-system supply obligations | New | First-time risk factor; management concedes it “may increase our costs and operational complexity.” |
| Access to capital markets / indebtedness | New | A first-time risk factor for a company holding $242.5B of cash — the tell that the funding model has changed. |
| Third-party data-centre and energy backstops | Increased | Credit-derivative notional $16.9B → $43.8B; equity derivatives $0 → $20.0B; financial guarantees $7.6B. |
| Cloud margin | Increased | Guided down near-term for the first time: third-party bridging capacity plus a low-single-digit-point Wiz headwind. |
| Search answer substitution | Unresolved, first bearish datapoint | No new management evidence; the CPC deceleration is the first quantified figure that leans against the expansion narrative. |
→ The pattern is consistent: disclosure expanded where results are strong and contracted where growth is lapping.
SECTION 3 — KPI DASHBOARD
| Metric | Q2 FY2026 | Q1 FY2026 | Q2 FY2025 | YoY | Trend |
|---|---|---|---|---|---|
| Total revenues | $119.8B | $109.9B | $96.4B | [YoY↑] | +24% y/y |
| Google Search & other | $63.3B | $60.4B | $54.2B | [YoY↑] | +17% y/y (Q1: +19%) |
| YouTube ads | $11.1B | $9.9B | $9.8B | [YoY↑] | +13% y/y (Q1: +11%) |
| Google Network | $7.3B | $7.0B | $7.4B | [YoY~] | -1% y/y |
| Subscriptions, platforms, devices | $12.9B | $12.4B | $11.2B | [YoY↑] | +15% y/y (Q1: +19%) |
| Google Cloud | $24.8B | $20.0B | $13.6B | [YoY↑] | +82% y/y (Q1: +63%) |
| Paid clicks change (y/y) | +13% | +13% | +7% | [YoY↑] | Volume accelerating |
| Cost-per-click change (y/y) | +3% | +5% | +7% | [YoY↓] | Third consecutive deceleration |
| Total operating income | $40.8B | $39.7B | $31.3B | [YoY↑] | +30% y/y |
| Operating margin | 34.0% | 36.1% | 32.4% | [YoY↑] | — |
| Google Services operating margin | 41.8% | 45.3% | 40.1% | [YoY↑] | 43.4% ex-$1.5B legal charge |
| Google Cloud operating margin | 35.6% | 32.9% | 20.7% | [YoY↑] | — |
| Alphabet-level activities (loss) | -$5.8B | -$5.4B | -$3.4B | [YoY↓] | Shared AI R&D, unallocated |
| Other Bets operating loss | -$1.8B | -$2.1B | -$1.2B | [YoY↓] | — |
| Cloud revenue backlog | $513.9B | $462.3B | n/d | [YoY?] | +11.2% q/q (Q1: +93%) |
| Capital expenditures | $44.9B | $35.7B | $22.4B | [YoY↑] | +100% y/y |
| Operating cash flow | $39.1B | $45.8B | $27.7B | [YoY↑] | TTM $185.7B |
| Free cash flow | -$5.9B | +$10.1B | +$5.3B | [YoY↓] | TTM $53.3B |
| Depreciation of P&E | $7.1B | $6.5B | $5.0B | [YoY↑] | +42% y/y |
| Purchase commitments (total) | $811.0B | $332.4B | n/d | [YoY?] | $93.7B at Sep 30, 2025 |
| Diluted shares | 12,309M | 12,238M | 12,198M | [YoY↓] | +0.9% y/y |
| Diluted EPS | $9.11 | $5.11 | $2.31 | [YoY↑] | Distorted by equity gains |
Segment and cash-flow figures are per the Q2 FY2026 and Q1 FY2026 Forms 10-Q; quarterly cash-flow items are derived by differencing the six-month and three-month statements. “n/d” = not disclosed for that period. Cost-per-click and paid-click changes are the 10-Q monetisation metrics (year-over-year, Google Search & other).
SECTION 4 — GUIDANCE ANALYSIS
| Metric | Guidance | vs. Prior | vs. Consensus | Management explanation | Analyst read |
|---|---|---|---|---|---|
| FY2026 capex | $195–205B | Raised from $180–190B (Q1) and $175–185B (Q4 2025) | Above consensus at each raise | “primarily due to an acceleration in the delivery of capacity to meet growing demand” | Second raise in six months, +11% at the midpoint. Q1’s raise was attributable to an acquisition; this one is pure demand-chasing. |
| FY2027 capex | “increase significantly” | Maintained, third consecutive quarter | Unquantified | “we’ll provide more details at a later date” | Three quarters of an unquantified commitment to a larger number. Visibility is being asserted, not demonstrated. |
| Cloud operating margin | Modest near-term pressure in Q3 | New qualification (Q1 flagged Wiz only) | n/a | third-party capacity is “a bridging strategy… will create modest margin pressure in the near-term” | First time margin has been guided down. The 35.6% print is a scarcity margin that management is telling you not to extrapolate. |
| Search & other revenue | No numeric guide; lapping flagged | New qualification | n/a | “we will begin lapping an acceleration in Search performance that began in the third quarter last year” | A pre-emptive deceleration warning. Combined with the CPC trend, Q3 is the quarter the Search debate gets tested. |
| TPU system revenue | Small in 2026, ramping into 2027 | Maintained | n/a | “the vast majority of the revenues from these agreements will be realized in 2027” | Consistent across three calls. The inventory build ($7.7B use of cash in H1) is arriving well ahead of the revenue. |
| Free cash flow | “will remain under pressure” | New explicit statement | n/a | “driven by our investments in technical infrastructure” | Management has stopped implying FCF recovers on a near-term schedule. |
| Q3 FX | Slight headwind | From +1pp tailwind in Q2 | n/a | “primarily in Search and YouTube Ads” | Removes roughly one point of optical growth from the two lines under the most scrutiny. |
The guidance picture is one of rising commitment against falling specificity. Every number that governs spending was raised or reaffirmed upward; every number that governs return — 2027 capex, Cloud margin, TPU economics, FCF recovery — became more qualified. That asymmetry is the quarter’s clearest signal about management’s own visibility.
Accounting and non-recurring clarifications
• A $2.1B charge for the Stockholm Patent and Market Court decision in the PriceRunner (Klarna) action: $1.5B of principal damages in Google Services G&A and $581M of interest and costs in OI&E. Google Services operating margin was 41.8% reported, approximately 43.4% excluding the charge.
• OI&E of $98.0B included $99.0B of net gains on equity securities, primarily unrealised marks on SpaceX and one private company — equivalent to roughly $8.04 per diluted share before tax, against reported diluted EPS of $9.11. Reported EPS and net income are not a read on operating performance this quarter; operating income of $40.8B is.
• A $7.7B inventory build in H1 2026, principally TPU systems held for sale, is a use of operating cash that precedes the associated revenue by roughly a year.
• Q2 2026 revenue carried an approximately one-point FX tailwind; Q3 is guided to a slight headwind.
→ Underlying economic performance: strong and improving at the operating line, materially weaker at the cash line — and the gap between the two is the investment case.
SECTION 5 — STRATEGIC / MOAT EVOLUTION
| Dimension | Condition | Direction | Key observations | Investment implication |
|---|---|---|---|---|
| Pricing power | Strong but no longer expanding | Weakening at the margin | CPC growth decelerated from +7% to +5% to +3% y/y while paid clicks held at +13%. Network cost-per-impression +13% on impressions -12%. | Search growth is now volume-led. Incremental AI-generated queries appear to clear at a lower price than the queries they supplement — durable if volume keeps compounding, fragile if it does not. |
| Competitive positioning | Strong | Stable to improving | Cloud +82% with new-customer acquisition velocity more than doubling y/y; existing customers exceeding commitments by >50%. Gemini API at 22B tokens/minute. | Alphabet is taking enterprise AI share. The position is being bought with capital rather than won with margin, which is a fair trade only if the capital earns. |
| Switching costs | Improving | Improving | $513.9B Cloud backlog, just over half recognisable within 24 months; Gemini Enterprise at ~90% of the Fortune 100; Agent Development Kit near 70M downloads. | Multi-year contracted commitments are the most genuinely improved element of the moat this quarter, and the one least reflected in near-term cash flow. |
| Operating leverage | Present but cyclical, not structural | Deteriorating in quality | Cloud margin 20.7% → 35.6% y/y, but guided down for Q3 on third-party bridging and Wiz. Depreciation +42% y/y. Unallocated AI R&D of $5.8B sits outside every segment margin. | Reported segment margins overstate underlying economics: $5.8B of shared AI cost is charged to no segment, and the Cloud margin is partly scarcity rent. |
| Capital intensity | Severe and rising | Deteriorating | Capex $44.9B in the quarter (+100% y/y); purchase commitments $811.0B versus $93.7B nine months earlier; credit-derivative backstops $43.8B notional. | The obligation base is now large enough that demand normalisation, not competitive loss, is the primary path to permanent value destruction. |
| Economic moat durability | Intact at the franchise level | Unchanged this quarter | Thirteen billion-user products, owned silicon, owned models, and default distribution remain in place; the December 2025 Search judgment and the pending ad-tech remedy are unresolved. | The moat is not what changed. What changed is how much capital must be committed each year to keep standing inside it. |
Over a five-to-ten year horizon the question is no longer whether Alphabet keeps its position, but what return on capital that position now requires. A business that converted roughly 44% of operating cash flow into free cash flow in FY2025 converts under 30% today. If AI compute earns what the last three years of compute earned, this is the cheapest moat purchase in the company’s history; if it earns half, Alphabet has bought a lower-return version of the business it already owned.
SECTION 6 — MAJOR RISKS & CHANGES
| Risk | Prior status | Current status | Direction | Investor implication |
|---|---|---|---|---|
| Capital-cycle risk — commitments outrunning contracted revenue | High, watched | High, materially worse | Worsening | Purchase commitments rose $478.6B in a quarter versus a $52B backlog increase. The obligation side is now compounding roughly nine times faster than the contracted revenue behind it. |
| TPU-system supply obligations (regulatory: new) | Not disclosed | New risk factor | New | Management now discloses it may “have excess capacity that we cannot easily redeploy” — the bear case written in the company’s own words. |
| Capital-structure and financing access (capital structure: new) | Not disclosed | New risk factor | New | First-time disclosure for a company with $242.5B of cash, naming the ATM programme and the dilutive impact of equity issuance. |
| Counterparty and circularity exposure (customer concentration) | Medium | Medium, growing | Worsening | Credit derivatives $16.9B → $43.8B notional; a new $20.0B milestone-contingent equity derivative; non-marketable securities $68.7B → $131.5B. Alphabet increasingly funds the customers it books revenue from. |
| Cloud margin durability (pricing competition) | Improving | Guided down near-term | Worsening | First explicit margin-pressure guide. Third-party capacity is bought at a cost that makes the bridge dilutive by design. |
| Answer substitution in Search (technological disruption) | Unresolved | Unresolved, first hard datapoint | Slightly worse | CPC +3% versus +7% a year ago, on +13% clicks. Consistent with AI generating incremental but lower-value queries. Management discloses no monetisation-per-query figure. |
| Antitrust remedies (regulatory) | High | High, unchanged | Stable | No new development this quarter; $17.4B of accrued fines and settlements remains. A separate $2.1B PriceRunner judgment landed in Sweden. |
Debated topics — concern, response, evidence
| Is the capex programme still disciplined, or is this an arms race? Investor concern: The market sees a second guidance raise in six months and a negative FCF quarter. Management response: “We are working off a disciplined ROIC framework… the dynamics look healthier than where we were about a year ago.” Demand outpaces capacity; existing deals are renewing with exceptional demand. Supporting evidence: Supports management: Cloud +82%, backlog $513.9B, customers exceeding commitments by >50%. Against: purchase commitments grew nine times faster than backlog, and no ROIC figure, hurdle rate or payback period has ever been disclosed. |
| Is AI expanding or eroding the Search profit pool? Investor concern: That AI answers displace monetisable clicks. Management response: “AI is fundamentally changing how the world searches… queries at an all-time high”; AI Mode past 1 billion MAU; monetisation on AI Overview queries “encouraging.” Supporting evidence: Supports: paid clicks +13%, Search & other +17%, AI Max unlocking “billions of net new searches.” Against: CPC decelerating three quarters running to +3%. Both facts are true — volume is being won at a lower price. |
| Is the 35.6% Cloud margin the new normal? Investor concern: That AI-era Cloud margins are structurally higher. Management response: “Outstanding strong operational discipline… leveraging growth in the top line,” but Q3 will carry third-party capacity and Wiz pressure. Supporting evidence: Against extrapolation: management itself guided the margin down, and $5.8B of shared AI R&D never touches the segment. The scarcity component of the margin is unquantified. |
SECTION 7 — CAPITAL ALLOCATION & PER-SHARE IMPACT
Free cash flow per share evolution
| Period | Free cash flow | Diluted shares | FCF / share | Change |
|---|---|---|---|---|
| Q2 2025 | $5.3B | 12,198M | $0.43 | — |
| Q3 2025 | $24.5B | ~12,230M | $2.00 | +365% |
| Q4 2025 | $24.5B | ~12,230M | $2.01 | Flat |
| Q1 2026 | $10.1B | 12,238M | $0.83 | -59% |
| Q2 2026 | -$5.9B | 12,309M | -$0.48 | n.m. |
| TTM to Q2 2026 | $53.3B | 12,309M | $4.33 | vs. $5.26 TTM at Q1 2026 |
Buybacks, issuance and dilution
• Buybacks: $11.5B (Q3 2025), $5.5B (Q4 2025), $0 (Q1 2026), $0 (Q2 2026). The programme is suspended, not tapered.
• Issuance: $49.6B of common and 6.25% mandatory convertible preferred, plus a $40.0B ATM authorisation (undrawn at June 30) to fund SBC tax obligations.
• Debt: $20.0B of US-dollar and $31.8B of foreign-currency senior notes issued in 2026; long-term debt $46.5B → $98.2B in six months, against roughly $16B twelve months ago.
• Net effect on share count: diluted shares 12,198M → 12,309M y/y, a 0.9% increase, before the mandatory preferred converts. Preferred dividends now appear in the EPS bridge ($86M in Q2).
• Acquisitions: 29.5B, into Cloud) and Intersect ($5.9B, data-centre energy). Goodwill $33.4B → $57.8B; intangibles $1.3B → $9.1B.
• Dividend: raised 5% to $0.22 per quarter — the one shareholder return still growing, and the smallest.
Long-term compounding implications
For a long-term owner the arithmetic has inverted. Through FY2025 Alphabet returned capital while growing; in H1 2026 it raised capital while growing, and per-share economics moved against the holder for the first time in a decade of buybacks. Whether that is good capital allocation depends entirely on the return earned on $195–205B of 2026 capex and a larger 2027 number — a return management describes qualitatively and has never quantified. Free cash flow per share of $4.33 on a trailing basis, against a share price in the $330s, means the market is paying for deferred earnings, not for current cash. That is a defensible position, but it is a position, not a fact.
INVESTOR QUESTIONS — WHAT A LONG-TERM OWNER MUST NOW REASSESS
| Question | Why it matters |
|---|---|
| What return does management actually underwrite on the $195–205B of 2026 capex, and over what payback period? | Three calls have invoked a “disciplined ROIC framework” without a single figure. If the hurdle is materially below the ~30% returns Alphabet has historically earned on capital, the whole programme is value-dilutive even if it is revenue-accretive. |
| Why did purchase commitments rise $478.6B in a quarter when Cloud backlog rose $52B? | If the commitments are chips, land and power bought ahead of contracted demand, Alphabet has taken merchant risk on AI compute for the first time. That is a different business than renting capacity against signed contracts. |
| Is the cost-per-click deceleration a mix effect or a value effect? | If AI-driven incremental queries are structurally worth less per click, Search revenue growth becomes dependent on perpetual volume expansion — a far more fragile engine than the price-and-volume compounding of the last decade. |
| What is Cloud’s operating margin excluding TPU system sales, Wiz, and third-party bridging capacity? | Management says growth accelerated ex-TPU but has never quantified either growth or margin ex-hardware. Without it, the 35.6% figure cannot be used to underwrite anything. |
| How concentrated is the $513.9B backlog among counterparties Alphabet also funds or backstops? | Revenue booked from customers whose data centres Alphabet backstops and whose equity Alphabet holds is lower quality than the headline implies, and correlates on the way down. |
| Why was the 350M paid-subscription disclosure dropped, and will Gemini Enterprise paid seats return? | Subscriptions are the one consumer line where Alphabet charges directly for AI. If disclosure is being withdrawn as growth slows, the consumer AI monetisation story is weaker than the product narrative suggests. |
| What would cause management to reduce capex rather than raise it? | No downside condition has ever been articulated. An investment programme with no stated stopping rule is a commitment, not an allocation decision. |
| Does the six-year server life survive another generation of accelerators? | With depreciation up 42% y/y and the 10-Q newly flagging that technology advances could change the benefit period, a life extension would flatter reported margins precisely when underlying economics are deteriorating. |
SECTION 8 — MONITORING CHECKLIST
| Category | Items |
|---|---|
| Five signals to track | • Purchase commitments at September 30, 2026 versus the $811.0B disclosed at June 30 — and whether Cloud backlog keeps pace. • Cost-per-click change in the Q3 10-Q: +3% or better arrests the trend; below +3% makes it a four-quarter deceleration. • Cloud operating margin after third-party bridging and Wiz — the test of whether 35.6% was structural or scarcity. • Depreciation of property and equipment as a share of revenue (5.9% in Q2 2026, up from 5.2% a year ago). • Whether the 350M paid-subscription disclosure returns, and whether Gemini Enterprise paid seats are given again. |
| Key risks building | • Off-balance-sheet exposure: credit derivatives $43.8B, equity derivatives $20.0B, financial guarantees $7.6B — all rising, none in the headline leverage. • Concentration: Alphabet is now simultaneously supplier, lender and shareholder to several AI counterparties. • Useful-life sensitivity: the 10-Q now flags that asset performance and technology advances could change the depreciation period. • Search laps its toughest comparison in Q3 with an FX headwind and decelerating CPC. |
| Confirmation signals | • Cloud backlog growing at or above the growth rate of purchase commitments. • TPU system revenue recognised on schedule as 2026 exits, with inventory drawing down. • Cloud margin holding above 33% through the bridging period. • Free cash flow returning to positive on a quarterly basis during 2027 without a capex cut. • Search & other growth holding at or above 15% through the Q3 and Q4 lapping period. |
| Conviction tripwires | • A third consecutive capex raise without a commensurate step-up in contracted backlog. • Any return to the equity market beyond the announced ATM — management has explicitly said there will be none. • An extension of server useful lives, which would flatter margins without improving economics. • Search & other growth into single digits while queries are still reported as growing — the combination that would falsify the expansion thesis. • A structural (divestiture) ad-tech remedy that survives appeal. |
