Amazon reported Q4 2025 revenue of $213.4 billion, up 12% year-over-year excluding foreign exchange (FX), with operating income reaching $25 billion despite $2.4 billion in special charges. AWS was the primary growth driver, accelerating to 24% year-over-year growth (or 19% including FX) with a $35.6 billion quarterly revenue, achieving a $142 billion annualized run rate. The North America segment grew revenue 10% to $127.1 billion with a 9% operating margin, while International revenue grew 11% to $50.7 billion. Advertising revenue surged 22% to $21.3 billion. Management highlighted significant momentum in AI, with their custom chip business (Trainium and Graviton) reaching a multibillion-dollar run rate and Bedrock growing 60% quarter-over-quarter. Looking ahead, the company plans to invest approximately $200 billion in capital expenditures, predominantly in AWS, to meet surging AI demand, while guiding for Q1 2026 net sales between $173.5 billion and $178.5 billion.
| Metric | Value | Change |
|---|---|---|
| Q4 Net Sales | $213.4 billion | +12% year-over-year (excluding FX) |
| Q4 Operating Income | $25 billion | Includes $2.4B in special charges (tax, severance, impairment) |
| AWS Revenue | $35.6 billion | +24% year-over-year (accelerating from 13 quarters ago) |
| AWS Operating Income | $12.5 billion | Operating margin of 35% (up 40 bps YoY) |
| North America Operating Income | $11.5 billion | Operating margin of 9% (up from 8% in 2024) |
| Advertising Revenue | $21.3 billion | +22% year-over-year |
| Free Cash Flow (TTM) | $11.2 billion | Reported for trailing twelve months |
| AWS Backlog | $244 billion | +40% year-over-year |
AWS AI Acceleration and Custom Silicon Dominance: AWS growth accelerated to 24% year-over-year, driven by a unique 'top-to-bottom' AI stack. Management emphasized that their custom silicon, Trainium and Graviton, is now a multibillion-dollar run rate business growing triple digits. Trainium 2 offers 30-40% better price performance than GPUs, and Trainium 3 is already fully subscribed for mid-2026. This vertical integration allows AWS to lower inference costs (a major customer pain point) while improving their own margins, creating a competitive moat against rivals relying solely on Nvidia.
Massive CapEx Deployment for AI Infrastructure: The company signaled an unprecedented $200 billion capital expenditure plan, predominantly for AWS. Unlike previous cycles where CapEx led to underutilized capacity, management stressed that demand is outstripping supply, with AI capacity being monetized 'as fast as we install it.' This signals a strategic shift from 'just in case' to 'just in time' aggressive expansion to capture a generational shift in cloud computing, positioning AWS to maintain its market leadership against Microsoft and Google.
Retail Evolution via Speed and AI Agents: The retail business is leveraging AI to drive frequency and basket size. 'Everyday essentials' grew twice as fast as other categories, and same-day delivery units increased 70%. The launch of 'Rufus' (300M users) and 'Add to Delivery' features demonstrates how AI is solving the 'last mile' friction. Management views 'Amazon Now' (30-minute delivery) as a key growth vector, with data showing customers triple their shopping frequency when using it, indicating a strategic pivot toward ultra-fast convenience to lock in consumer habit.
Kuiper (LEO) Commercial Launch and Cost Ramp: Amazon Project Kuiper is moving from R&D to commercial reality in 2026, with 180 satellites launched and agreements with AT&T and Verizon. While this represents a new multi-billion dollar revenue stream, it introduces a near-term financial headwind, with Q1 operating income expected to absorb $1 billion in launch costs. This signals a long-term strategic bet on global connectivity and hybrid cloud architectures, diversifying Amazon's infrastructure portfolio beyond terrestrial data centers.
Advertising as a High-Margin Growth Engine: Advertising revenue grew 22% to $21.3 billion, outpacing retail revenue growth. The integration of 'Prime Video Ads' (315M viewers) and new AI-driven 'Ad Agents' allows Amazon to monetize its high-intent shopping data more effectively. This segment is increasingly crucial for funding the company's heavy infrastructure investments, providing high-margin cash flow to support the AWS CapEx cycle.
International Segment Margin Pressure: Despite an 11% revenue increase in International, operating income was only $1 billion with a slim 2.1% margin. While management noted margin expansion excluding special charges (tax/lawsuit settlements), the structural profitability of international markets remains a weak point compared to North America. Heavy investments in 'Amazon Now' and international fulfillment expansion could continue to pressure margins in 2026.
Kuiper Satellite Drag on Earnings: Management explicitly warned that Amazon LEO (Kuiper) will increase North America segment costs by $1 billion year-over-year in Q1 due to 20+ planned launches in 2026. While these costs will eventually be capitalized, the immediate expensing of launch services creates a significant earnings headwind. This raises concerns about the timing of ROI for this high-risk, high-capital project, especially as investors are already sensitive to the massive AWS CapEx.
Supply Constraints vs. Demand Volatility: While management claims they are 'monetizing as fast as they install,' they also admitted that supply constraints (power, chips) are limiting growth. Jassy noted, 'every provider would tell you... we could actually grow faster if we had all the supply.' This creates a risk that if the AI boom pauses or competitors (like custom ASICs from Google/Microsoft) ramp faster, Amazon's current capacity constraints could cede market share during a critical window.
Guidance Ambiguity and Transcript Error: The Q1 operating income guidance in the transcript contained a likely typo ('$616.5 billion' instead of '$16.5 billion'), highlighting potential communication or control issues. More substantively, the guidance range ($16.5B - $21.5B) implies a potential sequential drop from Q4's $25B, driven by LEO costs and seasonality. This wide range and the heavy reliance on AI monetization to offset depreciation introduce execution risk for the coming year.
Overall: Management exhibited a highly confident and aggressive tone, particularly regarding the AI opportunity and AWS's competitive positioning. Andrew Jassy was emphatic about the 'unprecedented' demand for AI infrastructure and the company's ability to generate strong returns on the planned $200 billion CapEx. There was a notable lack of hesitation when discussing heavy investment spending, framing it as a necessary and immediately accretive move rather than a speculative risk.
Confidence: HIGH - Management displayed high conviction, citing specific metrics like 'monetizing capacity as fast as we install it' and 'fully subscribed' chip supply to justify the massive capital outlay. Jassy's language was definitive ('We are confident this will be the case here as well'), and the team provided detailed technical and financial evidence to support their AI strategy.
Expected to be between $173.5 billion and $178.5 billion. This assumes a favorable foreign exchange impact of approximately 180 basis points.
Expected to be between $16.5 billion and $21.5 billion (Note: Transcript text contained a typo of '$616.5 billion', corrected here based on context). Guidance includes a ~$1 billion year-over-year cost increase in North America related to Amazon LEO satellite launches.
Management expects to invest approximately $200 billion, predominantly in AWS. This includes significant spending on AI-capable servers, data centers, and Project Kuiper launches (20+ in 2026).
Hedging & Uncertainty: Management used minimal hedging regarding the AI opportunity, employing strong, definitive language like 'extraordinarily unusual opportunity' and 'confident this will be the case.' However, they did hedge on the *timing* of broader enterprise adoption, describing the market as 'barbelled' with a gap between AI labs and general enterprise workloads. Phrases like 'I would say that the enterprises are in various stages' and 'very well may end up being the largest' suggest confidence in the outcome but less certainty on the exact timeline of the revenue inflection.
I think this is an extraordinarily unusual opportunity to forever change the size of AWS and Amazon.com, Inc. as a whole. - Andrew Jassy, CEO
We are monetizing capacity as fast as we can install it. - Andrew Jassy, CEO
The lion's share of that demand is still yet to come in the middle of that barbell. - Andrew Jassy, CEO
We have very high demand... Customers really want AWS for core and AI workloads. - Andrew Jassy, CEO
We are investing aggressively here, and we are going to invest to be the leader in this space. - Andrew Jassy, CEO
It's very different having 24% year-over-year growth on a $142 billion annualized run rate than to have a higher percentage growth on a meaningfully smaller base. - Andrew Jassy, CEO
Analyst Sentiment: Analysts were highly engaged and focused on the financial mechanics of the AI boom, specifically the return on invested capital (ROIC) for the massive $200B CapEx plan. Questions from Mark Mahaney (Evercore) and Doug Anmuth (JPMorgan) were direct, probing for 'guardrails' and proof of monetization. There was skepticism masked as inquiry regarding the duration of the cycle and the risk of overbuilding. Analysts also sought clarity on the 'barbelled' nature of AI demand (Ross Sandler, Barclays) and the impact of AI agents on the retail funnel (Michael Morton, MoffettNathanson).
Management Responses: Management responses were detailed and defensive in a constructive way, using specific metrics (e.g., 'fully subscribed' Trainium 3, 'monetizing as fast as we install') to counter skepticism. Jassy was particularly assertive in correcting the narrative around AWS growth, emphasizing that 24% growth on a $142B base is superior to higher percentage growth on smaller bases. Olsavsky provided clear financial guardrails, explaining the mechanics of LEO cost capitalization versus expensing to reassure investors about margin volatility.
AI CapEx and ROIC: The dominant topic was the $200B investment plan. Analysts asked for proof of demand and duration. Management responded by highlighting the 'unusual opportunity' and citing immediate monetization rates and backlog growth ($244B, +40% YoY) as evidence of low risk.
Trainium and Silicon Strategy: There was significant focus on Amazon's custom chips. Jassy revealed that Trainium is a 'multibillion-dollar' business and 'fully subscribed,' arguing that custom silicon provides better price performance (30-40% better than GPUs) and better unit economics for Amazon compared to reselling Nvidia.
Retail Efficiency vs. Investment: Analysts asked how to balance heavy investment (LEO, Amazon Now) with efficiency. Management pointed to 'regionalization' and 'robotics' as efficiency drivers that allow them to invest in speed (same-day, 30-minute delivery) while maintaining or expanding margins.
AI Agents and Retail Disruption: Analysts inquired if AI agents (like Rufus or third-party bots) would disrupt Amazon's search/ad funnel. Jassy argued that Amazon's proprietary data gives it an advantage, predicting customers will prefer agents with full shopping history and inventory access over generic 'horizontal' agents.
Supply Constraints: The topic of power and chip supply constraints arose. Management acknowledged they are 'scrappy' and adding capacity as fast as possible (adding 1.2 gigawatts in Q4 alone), but admitted supply is the only limiter to faster growth.
Amazon is firing on all cylinders, with AWS re-accelerating (24% growth) just as the AI infrastructure boom enters its capital-intensive deployment phase. The investment thesis rests on three pillars: (1) AWS is gaining share in AI compute due to its superior custom silicon (Trainium/Graviton) and full-stack approach, which allows it to monetize massive CapEx faster than peers; (2) The Retail segment is becoming more efficient and sticky through 'everyday essentials' and ultra-fast delivery (Amazon Now), driving higher purchase frequency; and (3) Advertising remains a high-margin growth engine that subsidizes the broader ecosystem. Key risks include the execution of the $200B CapEx plan and the near-term profitability drag from Project Kuiper. However, management's confidence, backed by a $244B backlog and 35% AWS margins, suggests the market is underestimating the durability of this growth cycle. The 'barbelled' AI demand will eventually normalize, and Amazon is best positioned to capture the enterprise migration wave. We recommend buying the stock as the setup for 2026 earnings growth remains strong despite the heavy investment profile.
The AI infrastructure build-out is entering a phase of massive capital intensity, with Amazon alone planning $200B in CapEx. This signals a sustained, multi-year boom in data center construction, power consumption, and semiconductor demand. Management noted that 'every provider' is supply-constrained, indicating that the bottleneck for AI adoption has shifted from software/model capability to physical infrastructure (power and chips). Furthermore, the 'barbelled' nature of demand—dominated by AI labs today but expected to broaden to enterprises—suggests we are still in the early innings of the AI capex cycle, with the bulk of enterprise workload migration yet to come.
Consumer expectations for delivery speed are compressing rapidly, moving from 'same-day' to '30 minutes' (Amazon Now). Amazon's data showing customers 'tripling their shopping frequency' when using ultra-fast delivery suggests a structural shift in consumer behavior where convenience is winning over price/selection curation. This implies that logistics networks must fundamentally reconfigure to support higher density, faster delivery loops, increasing the moat for players like Amazon who can afford the CapEx to build such networks.