Trainium2 [AWS]
Let’s put that in numbers. Amazon’s capex target for the year could potentially exceed $100 billion this year. The news follows a wave of intensified spending commitments from other Big Tech players, including Alphabet, Meta, and Microsoft. While the company doesn’t break out R&D figures, Jassy noted that the “vast majority” of its $26.3 billion in capex in Q4 went to “AI for AWS.”
If you believe that plus altogether new experiences that we’ve only dreamed about are going to actually be available to us with AI, AI represents, for sure, the biggest opportunity since cloud and probably the biggest technology shift and opportunity in business since the Internet.
Other tech giants are also ramping up their AI spending:
- Alphabet (Google): Announced plans to boost capex to $75 billion in 2025, after spending $52.5 billion last year.
- Meta (Facebook parent): CEO Mark Zuckerberg signaled capex of up to $65 billion in 2025, compared with $39.2 billion in 2024.
- Microsoft: Following its recent earnings, analysts project total capex near $94 billion in 2025—an 83% year-over-year increase.
Collectively, Alphabet, Amazon, Apple, Meta, and Microsoft posted $128 billion in net income on $523 billion in revenue—a 26% year-over-year profit jump and a 10% increase in sales, as Barron’s has noted. Meta saw a strong 51% lift in net income, while Alphabet’s YouTube ad revenue surged even as cloud growth slowed. Microsoft expects continued double-digit revenue growth despite Azure’s deceleration, and Apple managed a 10% gain in net income, offsetting a 1% dip in iPhone sales. Overall, Barron’s attributes much of Big Tech’s mixed market reaction to ongoing tensions between massive AI expenditures and near-term earnings.
Amazon’s AI investments are multi-pronged. A key focus is on custom silicon, with the Trainium2 chips boasting 30% to 40% better price-performance than comparable GPU-powered instances, according to the company. These chips aren’t just for internal use; Amazon is positioning them as a competitive offering within AWS, attracting clients like Anthropic for its next-generation models. (Amazon is also backing the company, having invested $8 billion in it.) Beyond hardware, Amazon is bolstering its AI software services. Jassy highlighted Amazon Bedrock, a service providing access to various foundation models, including DeepSeek’s R1 and Amazon’s own Nova models, which reportedly offer lower latency and cost. He also touted Amazon Q, an AI assistant for software development and data analysis, as a “multiyear effort,” citing examples like Q’s capability in accelerating mainframe migrations. Even established services like Alexa are being re-architected with generative AI. A new generative AI-infused chatbot, has shown a notable 500 basis points higher customer satisfaction rate.
And he mentioned that Amazon is using AI in customer service. “For instance, if you look at customer service and you look at the chatbot that we’ve built, we completely rearchitected it with generative AI,” he said. “It’s delivering. It already had pretty high satisfaction. It’s delivering 500 basis points better satisfaction from customers with the new generative AI-infused chatbot.”
Concerning the Q4 capital outlays of $26.3 billion in the latest quarter, if that amount, which Jassy termed a “reasonable” run rate for 2025, is extrapolated over four quarters brings Amazon’s projected spending to about $105 billion—significantly higher than the $86 billion analysts forecasted.
To be clear, Amazon’s preferred definition of capex differs from the traditional measure (purchases of property and equipment). The company factors in both cash expenditures and finance lease acquisitions—while netting out proceeds from any property or equipment sales. As a result, the actual cash flowing out could exceed stated capex in any given quarter.
Wh Amazon’s R&D spending is a bit of an enigma
While Amazon doesn’t explicitly disclose a separate “R&D” budget, it reports a broad “Technology and Content” category in its financials that includes everything from AWS infrastructure to digital media. Over the last five years (2018–2022), these expenses grew from about \$29 billion to over $73 billion—making Amazon the world’s largest tech investor by many estimates.
Much of this outlay goes toward:
- AWS and cloud infrastructure: Funding data centers, custom chips (e.g. Graviton, Inferentia), and new services.
- AI: From Alexa voice assistants to large-scale machine learning models, as well as a $8 billion investment in AI startup Anthropic.
- Robotics and automation: Acquisitions like Kiva Systems, plus in-house development of warehouse robots and self-driving tech.
- Logistics Innovation: Investments in electric delivery vans (via Rivian), drone delivery (Prime Air), and route-optimization software.
Because “Technology and Content” also includes digital content (e.g., Prime Video spending), analysts caution that not all of Amazon’s reported figures strictly qualify as research and development. Yet, the company’s persistent double-digit annual increases in this category indicate an investment that spans hardware, AI, software, and next-generation retail infrastructure.
Sources: Amazon 10-K filings (2018–2023), Amazon press releases, and third-party analyses (e.g., Statista, CapitalOneShopping). For more detail, see Amazon’s SEC filings under “Technology and Content” expenses.
