Spokes.wiki Search About
Blog Posting source ↗ source url updated Thu Jul 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time)

Alphabet Q2 2026 earnings — Pichai’s remarks (roundup)

Sundar Pichai’s first-party CEO message on Alphabet’s Q2 2026 results (google). Like the monthly roundups, it spans models, cloud, search, hardware and Other Bets; this page extracts the model/provider-landscape signal and notes the rest as cross-spoke context. T1 for provenance (the company’s own figures), but every number is company-reported and unaudited-here — treat as a dated snapshot.

Model releases (the in-scope core)

The gemini lineup got its widest spread yet, sorted by cost tier and a testing/pre-training pipeline:

  • Gemini 3.6 Flash and Gemini 3.5 Flash-Lite — the new cost-efficient tiers.
  • Gemini 3.5 Flash Cyber — a cybersecurity-optimized variant pitched at “capabilities comparable to larger models.” First domain-specialized Gemini this wiki has logged.
  • Gemini 3.5 Pro — in testing.
  • Gemini 4pre-training has begun (the next generation, not yet a product).
  • Omni — video creation inside the Gemini app; a claimed +40% DAU creating videos (extends the Omni/generative-media line into the consumer app).

Scale metrics (the market signal)

The numbers that matter for this spoke’s market-size thread, not any single model:

  • ~22 billion tokens/minute across Google’s model APIs — up from 16B the prior quarter (+38% Q/Q).
  • Gemini app at 950M monthly active users, DAU tripled year-over-year.
  • 9M+ developers building on Google’s APIs monthly.
  • Gemma open-weight downloads ~900M total, with Gemma 4 at ~300M since its April launch — the open-track adoption number behind google‘s dual-track strategy.
  • Enterprise consumption concentration: ~500 Cloud customers each processed >1 trillion tokens in the last year; 2,000+ enterprises consumed 100B+ tokens.

Token-throughput and MAU are the clearest evidence yet that the provider market is expanding at the usage layer, not just the model-capability layer — the volume story under the llm-api-pricing price story.

Cross-spoke context (routed here, noted for the runners-up)

Recorded so the facets are visible; not paged here.

  • cloud-wiki — Google Cloud revenue +82% Y/Y, backlog $514B; new TPU 8t / 8i accelerators and the Axion CPU (“30% better performance per dollar”); the Virgo Network linking “a million AI accelerators.” The infrastructure substrate under the token numbers above; cloud-wiki’s territory.
  • search-marketing-wikiAI Overviews and AI Mode merged into one Search experience; AI Mode passed 1B MAU since its October global expansion; Google’s claim it “sends billions of clicks to websites every week” speaks directly to that spoke’s AI-search-shift thread.
  • agentic-tooling-wiki — the Agent Development Kit at ~70M downloads and Gemini Enterprise in ~90% of the Fortune 100; the agent-platform adoption thread.
  • Out of scope entirely: Waymo (Ojai vehicle), Wing (1M deliveries), Isomorphic Labs ($2B raise), YouTube World Cup viewership. Noted, not filed.

google · gemini · gemma-4 · google-ai-updates-may-2026 · gemini-omni-flash-nano-banana-2-lite · llm-api-pricing · llm-provider · synthesis