Key insights
- Hyperscalers (GOOGL, AMZN, MSFT, META) show strong cloud and ad revenue growth, but face challenges in free cash flow due to increased CapEx. Revenue backlog is significantly up YoY, indicating future demand. NVDA is repeatedly mentioned in earnings calls, highlighting its crucial role in AI infrastructure for these companies. This signals continued investment in AI and cloud services, which is bullish for the tech sector.
Hyperscalers Quarterly Growth * GOOGL: Cloud revenue $20.03 billion (up 63.38% YoY) and Ads revenue $77.25 billion (up 15.49% YoY) * AMZN: Cloud revenue $37.59 billion (up 28.43% YoY) and Ads revenue $17.24 billion (up 23.85% YoY) * MSFT: Intelligent Cloud revenue $34.68 billion (up 29.64% YoY). Azure specific growth 40%. * META: Ads revenue $55.02 billion (up 32.93% YoY)
Quarterly CapEx * GOOGL: $35.7 billion (up 197.5% YoY) * AMZN: $44.1 billion (up 67% YoY) * MSFT: $31.9 billion (up 49.1% YoY) * META: $19.8 billion (up 44.5% YoY)
Quarterly Free Cash Flow * GOOGL: $10.1 billion (down 46.8% YoY) * AMZN: $1.2 billion (down 95.4% YoY) * MSFT: $14.8 billion (down 17.3% YoY) * META: $12.4 billion (up 19.2% YoY)
Revenue Backlog * GOOGL: $461 billion (up 185% YoY) * AMZN: $268 billion (up 54% YoY) * MSFT: $633 billion (up 94% YoY) * META: N/A as short term contracts
NVDA mentions in the latest earnings calls by hyperscalers' CEOs
GOOGL: "NVIDIA GPUs are a core part of our AI accelerator portfolio and will be among the first to offer NVIDIA Vera Rubin NVL 72 in addition to the Blackwell and Hopper-based instances already available. Our goal is to provide the most comprehensive AI infrastructure, whether through our 8th-generation TPUs or the latest silicon from our partners."
MSFT: "We are building out Azure for the heterogeneous and distributed nature of AI workloads. Partners like NVIDIA, SAP, and ServiceNow are already integrating with Agent 365, our new control plane for agentic AI. Our infrastructure leadership is defined by the depth of our Blackwell clusters and the early integration of the Rubin architecture."
AMZN: "AWS growth is accelerating as we meet the backlog for high-performance compute. We have secured significant inventory and supply commitments for NVIDIA Blackwell and the upcoming Vera Rubin instances. This ensures AWS remains the preferred destination for AI startups and model makers requiring the highest throughput per megawatt."
META: "Our AI roadmap remains aggressive. We’ve seen significant returns on investment in ad-targeting efficiency. To support the training of Llama 4, we are continuing to scale our compute capacity with hundreds of thousands of NVIDIA Blackwell-class GPUs, while simultaneously preparing for the transition to the Rubin architecture late this year."
AI CapEx Supply Chain Beneficiaries * Nvidia: Designs GPUs (H200, Blackwell Ultra and Rubin), CPUs (Grace and Vera) and Networking (NVLink, Ethernet-X and BlueField). * Broadcom: Designs Custom AI ASICs for Google (TPU) and Meta (MTIA). * Marvell: Designs Amazon Trainium and Inferentia. * TSMC: Manufactures all the logic chips from Nvidia, Google, Meta, Amazon and Microsoft as well as the Nvidia networking equipment. * Samsung, SKHynix and Micron: Design and manufacture memory chips. * Vertiv, Schneider Electric and Eaton: Power and cooling.
Long positions: NVDA, GOOGL, META, MSFT, AMZN, TSM and MU. Not financial advice.