Alphabet Q2 2026: CapEx & Cloud — Consolidated Investor Notes
July 22, 2026.
Google Cloud beat every estimate and margin nearly doubled — but the two biggest bullish data points (the $514B backlog, the raised capex guidance) live only in unaudited call commentary, not the GAAP filing. Below is what matters, not a transcript.
CapEx
Q2 capex: $44.9B, +100% YoY — literally doubled. Four straight quarters of acceleration: $23.95B → $27.85B → $35.67B → $44.92B.
FY26 guidance raised to $195–205B, up from $180–190B — a ~$15B raise at the midpoint. This number is call-only — it’s nowhere in the earnings release.
Implied H2 2026 capex: ~$114–124B (guidance minus the $80.6B already spent through H1) — nearly 3x Q2’s run rate. That’s the real scale of “accelerating investment.”
2027 capex flagged to rise “significantly” — no number given yet.
Spend mix (call-only): ~60% servers, ~40% data centers/networking.
Funded by ~$70B raised this quarter: $49.6B equity (stock + mandatory convertible preferred, explicitly earmarked partly for capex) + $20.3B senior notes. Long-term debt roughly doubled to $98.2B.
Why the raise was needed: operating cash flow was only $39.1B against a headline $112.1B net income — $99.0B of that net income was a non-cash equity-securities gain, not cash. This is a real funding gap, not just optics.
FCF: −$5.9B, the first negative print in this data. Driven mostly by capex, but also by a $7.55B inventory build (consistent with the TPU-hardware ramp — see Cloud below), so it’s not purely a capex-timing story.
Depreciation is already climbing: $5.0B → $7.1B YoY (+42%) — the early edge of the bill for this capex cycle hitting the P&L. Management explicitly flagged more of this ahead (depreciation + data-center opex).
Cloud
Revenue: $24.8B, +82% YoY — beat the Street’s most bullish pre-print estimate.
Operating margin: 35.6%, up from 20.7% — nearly doubled in a year. Operating income $8.8B, more than tripled.
Backlog: $514B, +$50B+ sequentially. This figure does not appear anywhere in the GAAP filing — it’s a call-only, unaudited disclosure. ~50%+ of it is expected to convert to revenue over the next 24 months.
No dollar breakdown of what’s driving the 82% — GCP core, AI Solutions, AI Infrastructure, and TPU hardware sales are all bundled into one number. Management says growth “accelerated meaningfully even after excluding TPU system sales,” but gives no figure.
TPU hardware sales started recognizing revenue this quarter for the first time (small amount). Management says the “vast majority” of that revenue lands in 2027, not this year.
Two margin headwinds flagged by management for Q3/H2: (1) leaning on third-party/rented compute capacity as a bridge — explicit “modest margin pressure”; (2) Wiz integration — continued headwind through 2026.
Whether TPU hardware sales help or hurt Cloud margin is an open question — two different analysts asked directly, and the CFO never answered it either time (“we don’t break out margins for any specific products or infrastructure component”).
What the 35.6% margin doesn’t carry: shared AI R&D/infrastructure costs sit in a separate “Alphabet-level activities” bucket, which widened to a $5.8B loss (from $3.4B) — Cloud’s margin benefits from that shared spend without absorbing its cost.
Demand signals: new-customer acquisition velocity more than doubled YoY; existing customers are exceeding their commitments by 50%+ (an acceleration vs. last quarter); Marketplace transactions grew over 7x YoY; ~90% of the Fortune 100 are on Gemini Enterprise (one of the few demand stats that’s also in the filing itself).
Bottom line
The growth and margin numbers are real and beat expectations — but the single most bullish data point (the $514B backlog) and the raised capex guidance both live outside the audited-adjacent financial statements, in commentary that isn’t independently verifiable the way the filed numbers are.
Margin expansion is genuine, but management itself named two near-term headwinds and wouldn’t say whether the fast-growing TPU hardware business helps or hurts it — “is 35.6% the new normal” is still an open question, not a resolved one.
This capex cycle is not self-funded right now. Cash conversion is weak because this quarter’s GAAP profit was mostly a paper securities gain, and depreciation from the current spend is only starting to show up. FY26 guidance implies H2 capex almost triples Q2’s pace.
📊 Read this on SignalDeck.live.
Not financial advice. GAAP/unaudited figures sourced from Alphabet’s Q2 2026 earnings release; management commentary sourced from the Q2 2026 earnings call transcript. Both read in full.