☁️ Amazon: Outspending Everyone

AWS is doubling down on AI to meet soaring demand

App Economy Insights

Feb 14, 2025

Welcome to the Free edition of How They Make Money.

Over 180,000 subscribers turn to us for business and investment insights.

1. Amazon Q4 FY24

Income statement:

Revenue breakdown:

💻 Online stores (40% of overall revenue): Amazon.com +7% Y/Y.
🏪 Physical store (3%): Primarily Whole Foods Market +8% Y/Y.
🧾 3rd party (25%): Commissions, fulfillment, shipping +9% Y/Y.
📢 Advertising (9%): Ad services to sellers, Twitch +18% Y/Y.
📱 Subscription (6%): Amazon Prime, Audible +10% Y/Y.
☁️ AWS (15%): Compute, storage, database, & other +19% Y/Y.
Other (1%): Various offerings, small individually +17% Y/Y.

Revenue grew +10% Y/Y to $188 billion ($0.6 billion beat).
- Excluding AWS:
  - North America grew +10% Y/Y to $106 billion.
  - International grew +8% Y/Y to $40 billion.
Gross margin was 47% (+2pp Y/Y).
Operating margin was 11% (+4pp Y/Y).
- AWS: 37% margin (+7pp Y/Y).
- North America: 8% margin (+2pp Y/Y).
- International: 3% margin (+4pp Y/Y).
EPS $1.86 ($0.38 beat).

Cash flow (trailing 12 months or TTM):

Operating cash flow TTM was $116 billion (+36% Y/Y).
Free cash flow TTM was $38 billion (+4% Y/Y), driven by the operating cash flow growth partially offset by a 57% rise in Capex to $83 billion.

Balance sheet:

Cash, cash equivalent, and marketable securities: $101 billion.
Long-term debt: $53 billion.

Q1 FY25 Guidance:

Revenue ~$153 billion or +7% Y/Y ($158 billion expected).
Operating income $14 to $18 billion ($18 billion expected).

So what to make of all this?

AWS growth was steady despite constraints: AWS revenue grew 19% Y/Y for the third consecutive quarter, but supply chain and power limitations are slowing potential growth. Management expects these constraints to ease in H2 2025.

AI-driven investment boom: Amazon plans to invest over $100 billion in 2025, a 20% increase from 2024, primarily in AI-related infrastructure. This is the largest capex investment in Big Tech, outpacing Microsoft, Google, and Meta.
Advertising remains a bright spot: Ad revenue grew +18% Y/Y to $17.3 billion, gaining share from Google Search and YouTube. Sponsored ads and Prime Video monetization are driving growth.
E-commerce resilience amid price-conscious shoppers: Revenue from online stores grew +8% Y/Y, while revenue from third-party seller services grew +9% Y/Y. Consumers are spending more cautiously, leading to a lower ASP (Average Selling Price), but unit volume was up.
2025 guidance spooked investors: Q1 revenue guidance of $151 to $155.5 billion was below the $158 billion consensus, driven by a $2.1 billion foreign exchange headwind and a $1.5 billion impact from the leap year effect.
Long-term AI strategy takes center stage: CEO Andy Jassy doubled down on the AI-first approach, emphasizing custom silicon (Trainium 2 and future iterations) as a competitive advantage to reduce costs and improve AI efficiency.

2. AI tech stack update.

☁️ AWS market share and margins

Total cloud infrastructure market spending grew by 22% Y/Y to $91 billion1 in Q4 2024, steady throughout the year. Roughly half of the growth is attributed to AI through a mix of GenAI, GPU as a service, and other cloud services.

AWS had a commanding 30% market share, compared to 21% for Microsoft Azure and 12% for Google Cloud. Some of the fastest-growing tier-two cloud infrastructure providers were CoreWeave, Oracle, Snowflake, Cloudflare, and Databricks.

For context, in the December quarter:

Google Cloud grew 30% Y/Y (a slowdown from 35% Y/Y in Q3).
Microsoft Azure grew 31% Y/Y (a slowdown from 34% Y/Y in Q3).
AWS grew 19% Y/Y (no slowdown sequentially) from a much larger base.

I wouldn’t read too much into Azure and Google Cloud’s declaration relative to AWS since they are all capacity-contained.

What about margins? AWS maintained an elevated operating margin. Management warned that margins could fluctuate as AI investments flow through the P&L. AI is a potential short-term headwind on margin, but management expects the AI margin to match that of the non-AI business over time.

While AWS’s operating margin improved by 7 percentage points to 37%, 2 percentage points were due to an accounting adjustment (the useful life of servers was extended). In FY24, AWS generated $40 billion in operating income (58% of the total).

🤖 The three layers of the AI stack

Andy Jassy once again provided an update on Amazon’s position within the three layers of the AI stack, each representing a massive opportunity:

⚙️ 1) Infrastructure
🧠 2) Models
📱 3) Applications

Let’s break them down:

1️⃣ Infrastructure Layer

Amazon is doubling down on AI infrastructure with custom silicon to improve price performance. The company already uses custom chips like Nitro and Graviton in its core business and has extended its use to AI chips.

Trainium for training and Inferentia for inference are key pillars. Jassy highlighted the Trainium 2 rollout, emphasizing its 30-40% better price performance than comparable GPU instances.

“Several technically capable companies like Adobe, Databricks, Poolside, and Qualcomm have seen impressive results in early testing of Trainium 2.”

Amazon continues its deep partnership with NVIDIA but also focuses on reducing AI costs at scale with its in-house silicon strategy. The company is also collaborating with Anthropic to build Project Rainier, a massive AI cluster containing hundreds of thousands of Trainium 2 chips. Meanwhile, Trainium 3 is already in development and set to preview in late 2025.

2️⃣ Models Layer

Bedrock is Amazon’s LLM-as-a-Service platform. It offers customers a selection of foundation models from third-party providers and Amazon itself. SageMaker AI is also a notable successful service with AI model builders.

Recently added models include:

Luma AI, Poolside, and over 100 emerging models.
DeepSeek R1, the Chinese open-source model that sparked discussions about AI efficiency.
Amazon’s own model family, Nova, with Jassy claiming they offer lower latency and 75% lower costs than other models in Bedrock. Nova models already include prestigious customers like Palantir, Deloitte, and SAP.

Critically, Amazon wants to ensure frontier models are available on AWS and enable customers to do their best work. This approach reinforces the idea that LLMs will be commoditized.

Bedrock is also rolling out cost-saving features to make AI adoption more affordable, including prompt caching, intelligent prompt routing, and model distillation.

3️⃣ Applications Layer

Amazon is building AI-powered applications to drive business productivity and consumer engagement.

Amazon Q: A Gen AI assistant for software development, similar to GitHub Copilot. Companies are already using it to accelerate migrations from:
- Java JDK updates (saving $260 million).
- Windows to Linux, VMware to EC2, and mainframe migrations (cutting migration times by 50%+).
Rufus: AI-powered shopping assistant to enhance product discovery and customer engagement.
Alexa AI Overhaul: Jassy reiterated that Amazon is redeveloping Alexa with foundational models and aiming for more sophisticated AI-driven interactions.
Enterprise AI Adoption: AWS signed new AI partnerships with Intuit, PayPal, Reddit, Japan Airlines, and others—showing strong enterprise demand for AI-powered cloud solutions.

3. Key quotes from the earnings call

CEO Andy Jassy on advertising:

“Sponsored products, the largest portion of ad revenue are doing well and we see runway for even more growth. We also have a number of newer streaming offerings that are starting to become significant new revenue sources. On the streaming video side, we wrapped up our first year of prime video ads and we're quite pleased with the early progress and head into this year with momentum.“

Sponsored products remain a critical growth driver. Amazon’s high-intent audiences provide excellent returns on ad spend (ROAS), positioning advertising as a crucial growth avenue with a high-margin profile.

On Capex:

“The way the AWS business works and the way the cash cycle works is that the faster we grow, the more CapEx we end up spending because we have to procure data center and hardware and chips and networking gear ahead of when we're able to monetize it. We don't procure it unless we see significant signals of demand.”

In short, expanding Capex is a bullish sign for AWS in the coming quarters.

On AWS capacity:

“We could be growing faster if not for some of the constraints on capacity. And they come in the form of chips from our third-party partners coming in a little bit slower than before with a lot of midstream changes, to taking a little bit of time to get the hardware actually yielding the percentage healthy and high-quality servers we expect.”

Jassy predicted the supply constraints will relax in the second half of 2025.

On DeepSeek:

“I believe the cost of inference will meaningfully come down. I think it will make it much easier for companies to be able to infuse all their applications with inference and with generative AI.”

AI adoption is a clear tailwind for a business like AWS, which helps build customer experiences on top of infrastructure services. As discussed in our review of Microsoft’s quarter, lower inference costs are good news for hyperscalers.

4. What to watch looking forward

📦 Amazon vs. Shein & Temu

Tariffs are back in the spotlight—and a decades-old loophole could reshape e-commerce. So, what’s happening, and how does it impact Amazon?

The loophole: The de minimis exemption allows imports under $800 to bypass US tariffs, a significant advantage for Chinese e-commerce giants Shein and Temu. In 2023, the US imported $55 billion worth of goods this way, with China dominating these shipments.
The crackdown: Last week, Trump initially eliminated the de minimis exemption, only to reinstate it days later, causing chaos for e-commerce retailers. While Shein and Temu have been diversifying, the impact remains uncertain.
Amazon's move: Last fall, the company launched Amazon Haul, a direct response to Shein and Temu. It offers ultra-low-priced products ($20 or less) with longer shipping times.

If the de minimis loophole disappears, Amazon’s logistics advantage could help it outcompete rivals on fulfillment and reliability. Haul is a defensive play—but if the exemption stays, it could become a major business of its own.

🛒 Amazon ads

Advertising revenue reached $17.3 billion in Q4, representing:

32% of Google Search advertising revenue (+1pp Y/Y).
37% of Meta's advertising revenue (-1pp Y/Y).
165% of YouTube ads revenue (+5pp Y/Y).

The rise of Amazon’s ad business relative to YouTube since 2021 is striking.

In Q1 2024, the company flipped the switch, converting all Prime members to an ad-supported tier—unless they pay $2.99/month to go ad-free.

According to Nielsen, Prime Video captured 4.0% of US TV Time in December (+0.7pp Y/Y), nearly half of Netflix’s US market share. Prime Video has quietly become a dominant force in streaming.

🔴 Netflix has 90 million US members (+12% Y/Y).
🔵 Prime has 194 million 2 US members (+10% Y/Y).

Prime Video's engagement trails Netflix, but that's understandable. Prime members subscribe for shopping, shipping, and more, yet the platform's sheer reach still delivers impressive numbers.

A 2022 National Research Group study found that Prime has one of the lowest churn rates, trailing only cloud storage and music streaming. For advertisers, Amazon holds a unique edge: it knows what people actually buy, making it a goldmine for hyper-targeted CTV ads with industry-leading ROAS.

Amazon’s multi-purpose membership is the ultimate churn mitigator. If Prime saves you more than it costs, why would you ever cancel? That’s lock-in at scale.

That’s it for today.

Stay healthy and invest on!

Use our visuals for your organization

Interested in licensing our visuals for a report or presentation? Are you looking for custom visuals for your business or brand? Let's collaborate! Complete the form here, and we'll get in touch.

Disclosure: I am long AMZN, GOOG, META, NET, and SNOW in the App Economy Portfolio. I share my ratings (BUY, SELL, or HOLD) with App Economy Portfolio members.

Author's Note (Bertrand here 👋🏼): The views and opinions expressed in this newsletter are solely my own and should not be considered financial advice or any other organization's views.

Synergy Research Group estimates.

Consumer Intelligence Research Partners estimates.