Amazon vs Google Q3 2025 Earnings: Cloud & AI Business Insights

Comparative Analysis of Amazon and Alphabet (Google) on Cloud and AI from Q3 2025 Earnings Transcripts

---

1. Cloud Business Performance and Growth

Amazon (AWS)

• AWS revenue reached $33 billion in Q3, growing 20.2% year-over-year, marking the fastest growth rate in 11 quarters.
• AWS annualized revenue run rate is $132 billion.
• Operating income for AWS was $11.4 billion, with margins fluctuating due to ongoing investments, especially in AI infrastructure.
• AWS backlog grew to $200 billion by Q3 end, excluding unannounced deals.
• AWS is expanding its data center footprint aggressively, adding 3.8 gigawatts of power capacity in the past year and expecting to double capacity by 2027.
• AWS custom silicon (Trainium chips) is a key differentiator, with Trainium2 fully subscribed and Trainium3 preview expected end of 2025.
• AWS is heavily investing in AI services like SageMaker, Bedrock (inference engine), and AgentCore (infrastructure for AI agents), enabling customers to build and deploy AI agents securely and at scale.
• AWS is monetizing capacity rapidly and sees AI workloads as a significant growth driver alongside core cloud services.

Alphabet (Google Cloud)

• Google Cloud revenue grew 34% year-over-year to $15.2 billion, outpacing overall cloud market growth.
• Cloud operating income increased 85% to $3.6 billion, with operating margin expanding from 17.1% to 23.7%.
• Cloud backlog surged 82% year-over-year to $155 billion, driven primarily by enterprise AI demand.
• Google Cloud signed more billion-dollar deals in the first 9 months of 2025 than in the previous 2 years combined.
• Google Cloud offers a full-stack AI product portfolio, including proprietary AI models (Gemini, Imagen, Veo, Chirp, Lyria) and custom AI accelerators (TPUs).
• Nearly 150 Google Cloud customers processed ~1 trillion tokens each over the past 12 months using Google’s generative AI models.
• Google is investing heavily in TPU capacity and infrastructure to meet strong demand, with Anthropic planning to access up to 1 million TPUs.

2. Artificial Intelligence (AI) Initiatives and Products

Amazon

• AWS AI offerings include SageMaker (foundation model building/deployment), Bedrock (inference engine), and AgentCore (secure, scalable AI agent infrastructure).
• AgentCore SDK has been downloaded over 1 million times; customers like Ericsson, Sony, and Cohere Health use it to deploy AI agents.
• AWS launched AI-powered coding IDE (Kiro) and migration agent (Transform) saving significant developer hours.
• Amazon Connect (AI-powered contact center) surpassed $1 billion annualized revenue, handling 12 billion minutes of AI interactions.
• Amazon’s AI custom silicon (Trainium) offers 30-40% better price performance than competitors; Trainium2 is fully subscribed, and Trainium3 is expected to expand customer base.
• Project Rainier is a massive AI compute cluster with nearly 500,000 Trainium2 chips, used by Anthropic.
• Amazon is integrating AI into its retail operations with AI-powered shopping assistant Rufus, generative AI for seller listings, and agentic commerce features.

Alphabet

• Google’s AI stack includes proprietary generative AI models like Gemini (2.5 and upcoming 3.0), Imagen, Veo, Chirp, and Lyria.
• Gemini app has over 650 million monthly active users; queries tripled from Q2 to Q3.
• Google Cloud’s AI revenue grew over 200% year-over-year.
• Google Cloud launched Gemini Enterprise platform with 2 million subscribers across 700 companies, offering packaged enterprise AI agents.
• Google processes over 1.3 quadrillion tokens monthly across surfaces, a 20x increase year-over-year.
• Google integrates AI deeply into Search (AI Overviews, AI Mode), YouTube (AI-powered content creation and monetization), and Ads (AI Max for search ads).
• Google is advancing AI infrastructure with 7th generation TPUs (Ironwood) and NVIDIA GPUs, scaling capacity to meet demand.
• AI-powered shopping experiences and agentic commerce are being developed, including partnerships with PayPal and open protocols for agent-to-agent transactions.

3. Infrastructure and Capital Expenditures

Amazon

• Q3 CapEx was $34.2 billion, $89.9 billion year-to-date, primarily for AWS AI and core services infrastructure and custom silicon.
• Full-year 2025 CapEx expected around $125 billion, with increases anticipated in 2026.
• AWS power capacity doubled since 2022, with plans to double again by 2027.

Alphabet

• Q3 CapEx was $24 billion, with ~60% in servers and 40% in data centers and networking.
• 2025 CapEx guidance raised to $91-$93 billion, with a significant increase expected in 2026.
• Depreciation expenses increased significantly due to infrastructure investments.

4. Strategic Positioning and Market Implications

• Both companies emphasize AI as a massive growth opportunity and are investing heavily in AI infrastructure and services.
• Amazon leverages its custom silicon (Trainium) and AI agent infrastructure (AgentCore) to differentiate AWS.
• Google’s full-stack AI approach combines proprietary models, custom TPUs, and integration across search, cloud, and consumer products.
• AWS leads in scale with a larger cloud revenue base and aggressive capacity expansion.
• Google Cloud shows faster revenue growth and margin expansion, driven by AI product adoption and large enterprise deals.
• Both companies are innovating in AI-powered commerce and advertising, with Amazon focusing on agentic commerce and AI-powered shopping assistants, and Google enhancing search monetization with AI-driven ad products and AI-enhanced shopping experiences.

---

Summary: Amazon’s AWS remains the largest cloud provider with strong growth fueled by AI workloads, custom silicon, and AI agent infrastructure. Alphabet’s Google Cloud is growing faster with a highly differentiated AI product portfolio, proprietary models, and custom TPUs, driving strong enterprise adoption and margin expansion. Both companies are deeply integrating AI into their cloud offerings and consumer-facing products, positioning themselves as leaders in the AI-driven cloud era.