Q1 FY2026 Earnings Recap

Amazon.com, Inc. (AMZN)

Reported Apr 29th 2026

News Summary

  • Amazon's custom silicon business (Graviton, Trainium, Nitro) crossed a $20B annualized revenue run rate growing triple digits YoY, with Jassy claiming it is one of the top-3 data center chip businesses in the world; he added that if sold externally at market rates, the implied run rate would be $50B — a direct competitive benchmark for AI chip players including Cerebras. Trainium2 is largely sold out, Trainium3 nearly fully subscribed, and much of Trainium4 already reserved, with over $225B in total Trainium revenue commitments locked in ahead of supply.
  • Jassy explicitly named Cerebras as a new AWS partner alongside OpenAI, Anthropic, Meta, and NVIDIA, indicating a distribution or infrastructure relationship that could materially shape Cerebras's enterprise reach at scale. Separately, OpenAI committed to consume 2 GW of Trainium capacity via AWS starting 2027, and Anthropic secured up to 5 GW of current and future Trainium generations, with a backlog of $364B (excluding Anthropic's $100B+ commitment) confirming that hyperscaler AI compute demand is structurally locked in years ahead.
  • AWS Bedrock processed more tokens in Q1 2026 than all prior years combined, with customer spend up 170% QoQ and 125,000 customers including nearly 80% of the Fortune 100 — metrics that define the current velocity of the inference market Cerebras competes in. Jassy also described a new stateful agentic runtime co-built with OpenAI on Bedrock, with AgentCore deploying an agent every 10 seconds, signaling that inference workload profiles are shifting from stateless token throughput toward persistent multi-step agentic sessions with distinct chip architecture implications.
  • AWS revenue grew 28% YoY to $37.6B (fastest in 15 quarters, largest-ever Q4-to-Q1 sequential increase of $2B), generating $14.2B operating income at a 37.7% margin; total company revenue was $181.5B (+17% YoY), operating income $23.9B at a record 13.1% margin, both beating consensus. Q1 capex hit $43.2B (cash) / $44.2B (P&E), up ~77% YoY, compressing TTM free cash flow 95% to $1.2B, with Amazon reaffirming a $200B full-year capex target; Q2 guidance of $194-199B revenue and $20-24B operating income beat on the top line but the operating income midpoint of $22B was slightly below the $22.9B consensus.
  • AMZN initially fell as much as 7% after hours on capex and free cash flow concerns before recovering to close up roughly 2% in premarket the next day; Evercore's Mark Mahaney called it the 'strongest top-line and consolidated margin print in several quarters,' Bloomberg Intelligence said AWS growth 'confirms' its position to capture enterprise AI adoption, while Citi and Jefferies noted AWS's 28% growth was 'relatively in line' and likely below buyside expectations of 28-30%, leaving sentiment constructive but not euphoric given the scale of ongoing infrastructure spend.

Financial Highlights

MetricQ2 '24Q3 '24Q4 '24Q1 '25Q2 '25Q3 '25Q4 '25Q1 '26
Revenue$148.0B$158.9B$187.8B$155.7B$167.7B$180.2B$213.4B$181.5B
Rev Growth (YoY)10.1%11.0%10.5%8.6%13.3%13.4%13.6%▲ 16.6%
Gross Margin50.1%49.0%47.3%50.6%51.8%50.8%48.5%▲ 51.8%
Operating Margin9.9%11.0%11.3%11.8%11.4%9.7%11.7%▲ 13.1%
EV/Sales14.3x12.7x12.6x13.4x14.4x13.4x11.9x12.9x

Earnings Call Highlights

Amazon's Q1 2026 earnings call was dominated by an unmistakably bullish tone on AI infrastructure, with Andy Jassy projecting confidence bordering on triumphalism about AWS's position in the AI stack. Management's posture was assertive and expansive, framing Amazon's custom silicon business as a top-3 data center chip player and positioning Trainium as a structural cost and margin advantage rather than merely a competitive chip offering. Analysts probed capital intensity, model strategy, and agentic commerce, and management met every question with forward-leaning conviction rather than caution. The dominant theme was that AWS is in a compounding flywheel — AI spend drives core cloud growth, Trainium improves economics, and agentic workloads are pulling CPU demand alongside GPU — and Amazon believes it is uniquely positioned across the full stack to capture this wave.

  • Amazon's custom silicon business now has an annualized revenue run rate of over $20 billion growing triple digits year-over-year, and Jassy argued that if sold externally like other chip companies, the implied run rate would be $50 billion, making it one of the top 3 data center chip businesses in the world — a scale disclosure that will recalibrate competitive benchmarking for AI chip players including Cerebras.
  • Trainium2 is largely sold out with ~30% better price performance than comparable GPUs, Trainium3 just started shipping and is nearly fully subscribed, and much of Trainium4 — still 18 months from broad availability — is already reserved; Jassy also disclosed over $225 billion in revenue commitments for Trainium, signaling that demand is structurally locked in well ahead of supply.
  • Jassy explicitly named Cerebras as a new AWS partner in the same breath as OpenAI, Anthropic, Meta, and NVIDIA, indicating a distribution or infrastructure relationship that could shape how Cerebras reaches enterprise customers at scale.
  • AWS AI revenue is growing triple digits year-over-year, Bedrock processed more tokens in Q1 than all prior years combined, and Bedrock customer spend grew 170% quarter-over-quarter — metrics that define the current velocity of the inference market that Cerebras is directly competing in.
  • Jassy described a new stateful agentic runtime built jointly with OpenAI on Bedrock — 'Amazon Bedrock managed agents powered by OpenAI' — as the future architecture for production AI agents, with AgentCore deploying an agent every 10 seconds; this signals that the inference workload profile is shifting from stateless token throughput toward persistent, multi-step agentic sessions, which has direct implications for chip architecture requirements.

Sources

Generated by Earnings Bot on 2026-04-30 14:02 UTC.