As OpenAI's 'Midas touch' sweeps through the capital markets, the ecosystem layouts of two major AI camps—Google Gemini and Anthropic—are quietly taking shape. The listed companies involved in the underlying computing chips, cloud infrastructure, and enterprise application layers are becoming new focal points for investors seeking to capitalize on the AI wave.
From 'Midas Touch' to 'Circle of Influence': Two AI Camps Outline Key Investment Themes Across the Industry Chain
The Google Gemini camp centers around Google Cloud's AI infrastructure, relying on its self-developed TPU and the NVIDIA Blackwell platform to build a computing power foundation. It also integrates Broadcom's custom chips and Arm-based server CPUs, forming end-to-end AI service capabilities.
The Anthropic camp showcases a 'multi-cloud collaboration' feature: after securing $8 billion in total investment from Amazon, it has designated AWS as its primary cloud service provider, using AWS’s proprietary Trainium chips to train models and delivering Claude services to enterprise clients. In October 2025, the company further signed a cooperation agreement worth tens of billions of dollars with Google Cloud, highlighting its diversified computing power strategy.
Chip Layer: NVIDIA, Broadcom, ARM
NVIDIA: Serving as the core computing carrier for Gemini inference tasks, NVIDIA's Blackwell platform (B200/GB200) has been deployed on Google Cloud. Its data center revenue reached $41.1 billion in the second quarter of fiscal year 2026 (a 56% year-over-year increase), demonstrating the explosive demand for AI.
Broadcom: Long-term provider of customized networking chips and Ethernet switching solutions for Google TPUs. In 2025, its next-generation AI network chip entered mass production, enabling hyperscale clouds to achieve efficient parallel computing. Revenue for the third quarter of fiscal year 2025 reached $15.95 billion (a 22% year-over-year increase), with AI semiconductor business contributing significantly to growth momentum.
Arm Holdings: With the ecosystem penetration of Google's self-developed Arm-based server CPU 'Axion' and NVIDIA's Grace CPU, Arm Holdings occupies a key position on the CPU side of AI data centers. Revenue for the second quarter of fiscal year 2026 reached $1.14 billion (a 34% year-over-year increase), with royalty income reaching $620 million (a 21% year-over-year increase), reflecting the ongoing expansion of the Arm architecture in the computing field driven by AI.
AI Partners: Salesforce, ServiceNow
Salesforce (CRM): Integrating both Claude and Gemini into its Einstein platform, Salesforce provides Claude services via Amazon Bedrock and has established a seven-year, $2.5 billion cloud partnership with Google to support direct invocation of Gemini models within Agentforce.
ServiceNow (NOW): Integrated Gemini and Claude models into its Now Assist platform, enabling enterprises to flexibly select AI engines based on specific scenarios. Total revenue reached $3.407 billion in Q3 2025 (a year-over-year increase of 22%), with subscription revenue accounting for over 96%. Management cited AI-driven workflow automation demand as the primary growth driver.
Overall, the core investment logic of the current AI industry chain is as follows:
The computing power layer benefits first from the AI capital expenditure cycle. NVIDIA’s Blackwell platform achieving volume delivery, Google expanding TPU capacity to a million units, and increased demand for Broadcom's Ethernet chips form a high-certainty main track.
Cloud providers secure ecosystem entry points through equity partnerships and model distribution rights. The case of Amazon and Anthropic validates the synergistic effect of the 'investment + cloud services' model on revenue growth.
Enterprise application platforms leverage workflow integration capabilities to convert AI into measurable efficiency improvements and revenue growth. Salesforce and ServiceNow’s financial reports have already demonstrated the commercialization potential.
Investors can position themselves across three key layers: computing hardware (NVIDIA, Broadcom, ARM), cloud platforms (Amazon AWS, Google Cloud), and enterprise applications (Salesforce, ServiceNow). In the next phase, penetration rates of large model enterprises, API call volumes, and partner financial guidance will become critical high-frequency indicators to validate the investment thesis.
Upgraded performance insights deliver a competitive edge in seizing opportunities!Come and experience it now >>
Editor/Ray
