A layered view of Google's full-stack AI strategy — custom silicon, foundation models, cloud infrastructure, consumer apps, and frontier bets.
The first $100B+ quarter in company history, driven by AI-powered growth across every business unit.
Click each layer to explore. Alphabet is the only company operating a complete stack from custom silicon through foundation models to consumer-scale applications.
Expanding to London & Tokyo in 2026. Operating in Phoenix, SF, LA, Austin, Atlanta, Dallas, Nashville, Denver, Seattle. Airports approved in San Jose & SFO. Waymo for Business & Waymo Teens launched.
AlphaFold 3 earned a Nobel Prize in Chemistry (2024). Willow quantum chip: 13,000x faster than best supercomputers. Three Nobels awarded to Googlers in two years.
Operating loss of $1.4B — less than 2% of total revenue. Managed moonshot portfolio: Verily (precision health), Calico (longevity), Intrinsic (robotics). Waymo receiving increased resource allocation.
US ride-hail market ~$50B annually. Global ride-hail + taxi ~$300B+. Autonomous vehicles targeting 10-20% at higher margins (no driver costs). Largest deployed AV fleet with billions of real-world miles.
Rolled out across 40 languages in Q3. 100+ product improvements shipped in one quarter. Monetizing at approximately the same rate as traditional search. Driving query growth, especially among younger users.
$10.3B in Q3 (+15% YoY). #1 in US streaming watch time for 2+ years (Nielsen). Shorts now earn more revenue per watch hour than traditional in-stream. 300M+ paid subscriptions across Google. Veo 3, Speech-to-Song creator tools.
650M+ monthly active users. Queries doubled QoQ. Introduced agentic checkout via PayPal partnership. Already driving incremental total query growth — expansion, not cannibalization.
+21% YoY — fastest growth across Alphabet. Pixel 10 with Tensor G5 (best-reviewed devices ever). Android XR launched with Samsung. 300M+ paid subscriptions globally.
Fastest-growing AI Search Ads product. Hundreds of thousands of advertisers. Unlocked billions of net new queries. Kayak +12% conversion value, SoFi +39% conversion volume. Intent discovery, not keyword matching.
Deep integrations with Gemini and AI Mode. More agentic capabilities coming soon. Controls the search box for billions of users. Defensible independent of distribution payments.
13M+ developers building on Google's generative models. 150+ Cloud customers each processed ~1 trillion tokens in 12 months. Revenue from gen-AI products grew 200%+ YoY. Developer ecosystem lock-in & usage data flywheel.
Launched October 2025 across 700+ companies. Packaged enterprise agents optimized for multiple domains. Positioned as "the new front door for AI in the workplace." Workspace showing double-digit growth.
Over 70% of existing Google Cloud customers now use AI products. 9 of top 10 AI labs choose Google Cloud. 13 product lines each at $1B+ annual run rate. WPP: 70% efficiency gains; Swarovski: 17% email open rate increase.
Full MLOps platform for training, tuning, and deploying models. Supports both Google (Gemini, PaLM) and third-party models. Agent Builder for building AI agents with grounding. Enterprise search and conversation capabilities.
+34% YoY growth. Operating income surged to $3.6B (+85% YoY). Margin expanded from 17.1% to 23.7% — 650 bps improvement. "Deep product differentiation built with our own technology."
~60% servers (TPUs, GPUs, CPUs), ~40% data center construction & networking. "Significant increase" expected in 2026. Still in tight demand/supply environment. CapEx roughly equals free cash flow ($73.6B TTM).
A4X Max instances powered by NVIDIA GB300 shipped to Cloud customers. "The only company providing a wide range of both GPUs and TPUs." Infrastructure Switzerland — customer choice, not lock-in.
Revenue from products built on generative AI models grew more than 200% year-over-year. Infrastructure portion is the growth driver looking ahead. Decade of experience building AI accelerators.
World-leading AI model powering Search, Cloud, and Workspace. Advanced reasoning, multimodal understanding, and 1M+ token context window. Powers AI Overviews, AI Mode, and Gemini Enterprise.
Flash: optimized for speed and efficiency at scale. Pro: balanced capability and cost. Ultra: maximum capability for frontier tasks. Powering 7B tokens/minute via API across 13M+ developers.
Video generation model integrated into YouTube creator tools. Speech-to-Song functionality. Genie 3 for world generation. Nano Banana: "viral sensation" from Google Labs. AI-powered channel insights for creators.
AlphaFold's protein folding breakthrough earned Nobel Prize in Chemistry (2024). DeepMind continues pushing scientific frontiers. Computational biology and drug discovery applications. Step-function improvements in healthcare AI.
Google's seventh-generation TPU entering general availability. Described as "highly sought-after." Designed for both training and inference workloads. Decade-long head start — TPU v1 launched in 2016. TSMC fabrication.
Anthropic plans to access up to one million TPUs — one of the largest AI infrastructure commitments ever disclosed publicly. Validates competitiveness for frontier model training. Directly challenges NVIDIA dominance.
Powers Pixel 10 series — "best reviewed devices ever." First chip designed to run Gemini natively on-device. Showcase for Google's AI capabilities. Enables private, low-latency inference on mobile.
Ran an algorithm 13,000x faster than best supercomputers. Verifiable quantum advantage achieved. Paving the way for future practical applications. Nobel laureate Michel Devoret leads quantum hardware.
$102.3B quarterly revenue with accelerating growth across all segments. AI investments are expanding the pie, not just defending share.
A decade of compounding investment in AI infrastructure, models, and applications.
Google becomes the first hyperscaler to deploy custom AI accelerators at scale. The decade-long head start in custom silicon begins.
Google Brain researchers publish "Attention Is All You Need" — the foundational paper behind every modern LLM including GPT and Gemini.
DeepMind's AlphaFold cracks the 50-year protein folding problem. Opens 200M+ protein structures to researchers worldwide.
Google's first natively multimodal AI model family. Ultra, Pro, and Nano variants for different capability tiers and form factors.
Breakthrough context window enabling processing of entire codebases, books, and video content in a single prompt.
AlphaFold earns Nobel in Chemistry. Willow quantum chip achieves 13,000x speedup. Three Nobels to Googlers in two years.
Flash, Pro, Ultra variants. AI Mode reaches 75M DAU. AI Overviews serve 2B+ users across 40 languages.
$102.3B revenue (+16% YoY). Cloud hits $15.2B (+34%). Gemini 2.5 Pro powers the full stack. AI Max unlocks billions of new queries. TPU v7 Ironwood enters GA.
2M enterprise subscribers across 700 companies. Anthropic commits to 1M TPUs. 13M+ developers on Gemini APIs. 7B tokens processed per minute.
Waymo expands to London & Tokyo — first major international markets. Significant CapEx increase planned. Tight demand/supply in AI infrastructure continues.
Each layer reinforces the others. Better infrastructure enables better models. Better models attract more developers. More developers create more usage data. More data improves models. The flywheel accelerates over time.
Alphabet is executing four simultaneous transitions in interface, intent, monetization, and distribution — while managing three strategic time horizons.
$230B+ annual Search revenue creates institutional bias against necessary cannibalization. AI Mode is growing incrementally, but the test is whether Google can sustain disruption velocity.
AI search costs more per query. Search operates at 38.5% margin vs Cloud's 23.7%. Cost curves must improve faster than query mix shifts toward expensive AI-powered experiences.
DOJ ruling imposes restrictions on search distribution. EC fines ($3.5B in Q3). Ongoing investigations. Required to share search data with competitors. Google plans to appeal.
ChatGPT Search, Perplexity answer engine, Amazon Trainium, Microsoft/AMD silicon, Tesla FSD, Cerebras/Groq alternative architectures. NVIDIA's CUDA software moat remains entrenched.