Alphabet Q2 2026 key numbers at a glance
On July 22, 2026, Alphabet reported results for the quarter ended June 30. Revenue and Cloud growth both beat Wall Street consensus, but GAAP diluted EPS was inflated by unrealized equity gains and should not be compared directly to adjusted EPS.
| Metric | Q2 2026 | YoY change |
|---|---|---|
| Total revenue | $119.8B | +24% |
| GAAP diluted EPS | $9.11 | Includes ~$99B unrealized equity gains |
| Adjusted EPS | ~$2.88 | Slightly below consensus $2.89 |
| Google Search & other | $63.3B | +17% |
| YouTube ads | $11.1B | +13% |
| Google Cloud | $24.8B | +82% |
| Operating income | $40.8B | Operating margin 34% |
| Cloud operating margin | 35.6% | Up from 20.7% a year ago |
| Cloud backlog | $514B | Up >$50B quarter over quarter |
| Q2 capex | $44.9B | ~+100% YoY |
| Free cash flow (FCF) | -$5.9B | First negative quarterly FCF since IPO |
Friction points AI developers face right now
The headline beat does not automatically smooth your workflow. These constraints got sharper during the Q2 earnings window:
- Flagship model gap: Gemini 3.5 Pro was slated for June; Bloomberg reported on July 16 that code-generation quality missed internal bars and the launch slipped by months. Agentic coding still lacks Google’s top-tier official option.
- Token cost uncertainty: Higher capex means Google must balance in-house TPUs with third-party compute under tight supply. The Q3 third-party capacity bridge may pressure margins short term, creating risk around API pricing and rate limits.
- Accelerating release cadence: Pichai pledged near-monthly model drops. Flash lines iterate fast (including live Gemini 3.5 Flash Cyber), but the missing Pro flagship makes enterprise POCs hard to lock onto a long-term baseline.
- Competitive pressure: OpenAI and Anthropic keep pushing enterprise coding; open-weight models like Kimi K3 compete on long context and pricing. Multi-vendor evaluation beats betting everything on Gemini.
- Infrastructure bottlenecks: A $514B Cloud backlog signals demand outpacing owned capacity. Large contracts may face scheduling delays, making hybrid and self-hosted options pragmatic.
Google Cloud up 82%: enterprise AI infrastructure as the growth engine
Cloud was the standout line item: $24.8B revenue, 82% YoY growth, operating margin expanding from 20.7% to 35.6%, and backlog reaching $514B. Alphabet expects more than half of that backlog to convert to revenue within 24 months.
Growth drivers span several vectors:
- Enterprise AI infrastructure: GCP is scaling both AI Infrastructure and AI Solutions. Q2 marked the first revenue contribution from TPU system sales into customer data centers.
- Gemini Enterprise: Nearly 90% of the Fortune 100 are adopters, per management. DeepMind CEO Demis Hassabis said post-earnings that enterprise uptake exceeded expectations.
- External TPU sales: Multi-billion-dollar TPU deals with Meta, Anthropic, and other hyperscale customers continue to lift hardware revenue.
- Wiz integration: After the $29.5B cloud-security acquisition, security cross-sell is amplifying Cloud; Gemini 3.5 Flash Cyber targets security workloads.
CEO Sundar Pichai said in the official release: “Q2 was an amazing quarter, with Alphabet revenues growing 24% year-over-year and Google Cloud revenues accelerating to 82% growth, driven by demand for AI infrastructure and AI solutions. Gemini Enterprise is seeing broad adoption, with nearly 90% of the Fortune 100 using it.”
Search and YouTube: AI features lift query volume without killing ads
Search at $63.3B (+17%) and YouTube ads at $11.1B (+13%) show AI features have not yet eroded core ad monetization the way some analysts feared. On the earnings call, Pichai highlighted AI-driven incremental query growth and billions of weekly clicks from AI features to websites.
- AI Mode: Since global expansion in October 2025, monthly active users (MAU) passed 1 billion.
- Gemini app: 950 million MAU; API throughput hit 22 billion tokens per minute.
- Antigravity: 2.4 million weekly active users (WAU). An internal Chrome case study showed model-driven refactors compressing a two-year timeline to three months.
- Site traffic: AI Overviews and AI Mode send billions of clicks to publishers weekly, pushing back on the “AI summaries kill SEO traffic” narrative.
Why GOOGL fell ~4% after hours despite the earnings beat
Revenue of $119.8B beat consensus $116.9B, and Cloud’s 82% growth blew past expectations. After-hours trading still sent the stock down ~4%. The sell-off was not about weak operations — it was about the speed of AI capital expenditure and cash burn:
- Three-step capex guide hike: 2026 full-year guidance moved from $175–185B, to $180–190B in April, then to $195–205B after Q2. Each step added $15–20B, nearly matching the bottom of the original range in cumulative uplift.
- Historic FCF turn negative: Q2 operating cash flow was $39.1B, but capex hit $44.9B, yielding FCF of -$5.9B — Alphabet’s first quarterly FCF deficit as a public company despite high-margin ads.
- Gemini 3.5 Pro overhang: The flagship delay raises whether massive capex converts into model leadership. Raymond James analyst Josh Beck flagged before earnings that missing the June window shifted perception from “leading edge” to “half a beat behind.”
- Hawkish CFO tone: Anat Ashkenazi said free cash flow will “remain under pressure, driven by infrastructure investment,” and previewed 2027 capex rising “significantly” with no easing signal.
Bottom line: An operations beat paired with a capital-spending scare. For developers, expanding Cloud supply and API throughput is bullish; for investors, a $205B capex ceiling plus hints of even higher 2027 spending pushes out the timeline for FCF normalization.
Gemini 3.5 Pro delay: Bloomberg reporting and Google’s response
On July 16, Bloomberg cited 10 current and former employees reporting that Gemini 3.5 Pro slipped by months because code-generation performance missed internal targets. Pichai promised a June launch at Google I/O in May; that timeline has slipped repeatedly.
- Partner testing: A Google spokesperson said 3.5 Pro, an upgraded Flash, and other models remain in partner testing with no new public date.
- Raymond James view: Josh Beck framed the gap as a “perception lag” in an AI lab that ships 20+ X.0/X.5 releases per year — but missing the June flagship still changed how the Street scores Google’s model cadence.
- Gemini 4 underway: Pichai confirmed heavy compute is already allocated to Gemini 4 pre-training, calling it a “very ambitious effort” aimed at frontier competitiveness at launch.
- Near-monthly releases: Pichai said Google plans to ship new models almost every month as Gemini scales; earnings week brought Gemini 3.6 Flash and other lower-cost options.
- Flash Cyber live: Gemini 3.5 Flash Cyber is available now, optimized for security workloads at strong price-performance, paired with the CodeMender agent for automated patching.
Frozen v2 chip: The Information report and the TPU complement strategy
Ahead of earnings, The Information reported Google is developing an AI server chip codenamed Frozen v2, hardening key Gemini compute paths in silicon with a target of 6–10× better tokens per watt versus the latest TPUs.
- Gemini-specific: Frozen v2 applies only to the Gemini stack, embedding portions of the model graph in hardware to cut data movement.
- Complement, not TPU replacement: Google treats it as a specialized add-on, not a wholesale swap; initial volumes will not match TPU scale.
- Capacity-driven: Internal AI compute shortages already forced Cloud to turn away some large enterprise deals; Frozen v2 is an experimental path to relieve bottlenecks.
- 2027 revenue contribution: Analysts expect meaningful Cloud hardware revenue from Frozen v2 earliest in 2027 if production succeeds, with large-scale deployment targeted around 2028.
Where Alphabet’s capital expenditure goes: 60% servers, 40% data centers
Of Q2 capex of $44.9B, the vast majority funded AI technical infrastructure. Ashkenazi broke down the split and funding sources on the call:
| Item | Value / note |
|---|---|
| Technical infrastructure — servers | ~60% |
| Technical infrastructure — data centers & networking | ~40% |
| June 2026 equity raise | $49.6B (Class A/C common + mandatory convertible preferred) |
| Q2 2026 debt issuance | $20.3B senior unsecured notes |
| Cash & marketable securities | $242.5B (Q2 end) |
| Q3 third-party compute bridge | Expanding external cloud capacity as a transition strategy under supply constraints; may compress Cloud margins near term |
Ashkenazi said capex will be funded primarily from FCF and debt, with equity issuance mainly offsetting tax obligations from employee stock compensation. She reiterated 2027 capex will rise “significantly”; Raymond James estimates it could exceed $300B.
Q1 vs Q2 2026: key metrics side by side
Quarter-over-quarter comparison highlights Cloud acceleration and the capex step-up:
| Metric | Q1 2026 | Q2 2026 | Change |
|---|---|---|---|
| Total revenue | $109.9B | $119.8B | +9% QoQ |
| Google Cloud revenue | $20.0B (+63% YoY) | $24.8B (+82% YoY) | +19 pp growth acceleration |
| Cloud backlog | >$460B | $514B | +$54B |
| Cloud operating margin | ~24% | 35.6% | +11.6 pp |
| Capex | $35.7B | $44.9B | +26% QoQ |
| FCF | $10.1B | -$5.9B | Turned negative |
| 2026 full-year capex guide | $180–190B | $195–205B | +$15B raise |
| API token throughput | 16B tokens/min | 22B tokens/min | +38% |
Q3 2026 watchlist: seven items to track
Into the next earnings and product window, AI developers and technical decision-makers should monitor:
- 2027 capex preliminary guide: Raymond James estimates could exceed $300B, directly shaping API pricing and Cloud contract terms.
- Gemini 3.5 Pro timeline: Whether Q3 brings a firm date or Flash lines keep filling the flagship gap.
- Third-party compute bridge vs Cloud margin: Ashkenazi flagged Q3 expansion of external capacity; watch whether gross margin pressure stays “modest” as guided.
- Search Q3 comp: The CFO warned Q3 laps last year’s Search acceleration; sustaining growth off a $63.3B Q2 base is the test.
- FCF recovery: Back-to-back negative FCF quarters would fuel “AI overspend” narratives; whether Q3 capex stays above $40B is key.
- Frozen v2 and TPU roadmap updates: Any hardware detail ahead of Google Cloud Next or I/O.
- Antigravity and Gemini Enterprise adoption: Whether 2.4M WAU converts to paid Cloud contracts and Fortune 100 penetration clears 90%.
Six steps for AI developers responding to Q2 earnings signals
Earnings analysis only matters if it drives action. These six steps balance Gemini ecosystem bets with multi-vendor resilience:
- Lock a baseline on models available today: While 3.5 Pro is absent, benchmark Gemini 3.5 Flash, 3.6 Flash, or Flash Cyber for cost and latency. Log token usage now instead of waiting for the flagship.
- Build a multi-vendor API abstraction: Wrap Gemini, Anthropic, Moonshot, and others behind an OpenAI-compatible schema so delays or rate limits on one provider do not stall production.
- Evaluate Antigravity and local agent workflows: Antigravity’s 2.4M WAU suits agent-first development, but you need a clean, resettable macOS environment for iterative tests without polluting your primary machine.
- Monitor Cloud contracts and API rate limits: A $514B backlog means supply is still tight. Large deployments should budget third-party compute bridges or hybrid cloud, with price-protection clauses in contracts.
- Track capex vs pricing linkage: Set quarterly reminders to compare Alphabet 10-Q capex with Cloud gross margin moves and stress-test API budget buffers.
- Reserve Q3 model-release windows: With near-monthly releases promised, treat August–September as fixed re-evaluation points to compare 3.5 Pro (if shipped) against rivals on agentic coding benchmarks.
Citable hard data and official sources
- Cloud backlog $514B: Up more than $50B quarter over quarter; Alphabet expects more than half to convert within 24 months — source: Q2 2026 earnings call and Exhibit 99.1.
- API 22B tokens/minute: Up 38% from Q1’s 16B tokens/minute, reflecting expanding Gemini API throughput — source: Pichai Q2 call.
- Three-step capex guide: $175–185B → $180–190B → $195–205B — source: CNBC live coverage and Q2 call.
- FCF -$5.9B: Alphabet’s first negative quarterly free cash flow since IPO — source: Q2 2026 supplemental materials.
Earnings figures and model timelines may change with subsequent 10-Q filings or official announcements. Treat the primary sources below as authoritative.
Official Alphabet Q2 2026 earnings release (PDF):
Alphabet Q2 2026 Earnings Release (PDF)
SEC 8-K Exhibit 99.1 full text:
SEC EDGAR — Alphabet Exhibit 99.1 Q2 2026
CNBC intraday and after-hours live coverage:
CNBC — Google (GOOG) Q2 2026 Earnings Live Updates
Why Did GOOGL Stock Fall After Beating Earnings?
What were Google's Q2 2026 earnings results?
Alphabet reported Q2 2026 revenue of $119.8 billion (up 24% YoY), diluted EPS of $9.11 (adjusted ~$2.88 excluding a $99B unrealized equity gain), and Google Cloud revenue of $24.8 billion (up 82% YoY). Results beat Wall Street revenue estimates of $116.9B.
Why did GOOGL stock fall after beating earnings?
Despite the earnings beat, Alphabet shares fell ~4% in after-hours trading because the company raised its 2026 capital expenditure (capex) guidance to $195–205 billion (from $180–190 billion). Q2 capex hit a record $44.9 billion, and free cash flow turned negative at -$5.9 billion.
Is Gemini 3.5 Pro delayed?
Yes. Google confirmed Gemini 3.5 Pro is in testing with partners but has not set a public launch date. Bloomberg reported the model is months behind schedule because coding performance fell short of internal goals.
What is Google's Frozen v2 chip?
Frozen v2 is a Google-designed AI server chip that embeds parts of Gemini's architecture directly into silicon. The Information reported it could be 6–10× more power-efficient than current TPUs for running Gemini models.
Testing Antigravity, the Gemini API, and native Xcode toolchains requires Apple Silicon hardware you can wipe clean between model releases — especially with Google shipping new checkpoints almost monthly. Shared laptops accumulate conflicting SDK versions and background agent processes. KVMFLUX cloud Mac mini rentals from daily terms give you root access, SSH + VNC, and dedicated physical M4 hardware so you can stand up a clean evaluation environment before the Q3 model window and tear it down when done. If you are choosing a rental period, daily billing fits these earnings-driven sprint cycles best.
Test Gemini workflows on real Apple Silicon
Rent a Mac mini M4 by the day to run Antigravity agents or Gemini API integration tests — validate during the earnings window and return the machine when you are done, without touching your primary dev box.