Built by Grok 4.5

AMZN | Earnings Review — 2026 Q2-Grok

Verdict: ACCELERATING — and the acceleration is concentrated in the highest-margin segment. Consolidated revenue growth climbed five straight quarters from an +8.6% trough (Q1'25) to +19.6% in Q2'26, with two consecutive ~+300 bps steps in the YoY rate. AWS accelerated for a fifth consecutive quarter to +36.7% — fastest in 18 quarters — on a base more than 2× the last time growth ran this hot. AWS operating margin flipped from multi-quarter compression to +645 bps YoY expansion to 39.4% (+520 bps ex energy-derivative gain). Growth accelerating with margins expanding is the strongest combination in the framework.

Hold the clean numbers. GAAP diluted EPS of $5.75 (+242% YoY) embeds $53.4B non-operating pre-tax other income (primarily Anthropic). Analyst-derived clean EPS is ~$1.95, a ~+7% beat vs $1.82 street — not +216%. Operating income $27.5B (+43%) includes $1.2B of disclosed one-time benefits (~$600M tariff refunds in North America + ~$600M energy-derivative FV, mostly AWS). Clean OI is ~$26.3B, +37%. Roughly ~400 bps of the revenue acceleration is Prime Day landing in Q2 this year vs entirely in Q3 last year — clean underlying revenue growth near ~15–16%. Every adjustment still leaves a strong quarter; the difference between narrative and underwriting is the clean bridge.

The beat is an AWS beat. Strip the two one-offs and upside sits in AWS revenue acceleration and AWS margin. Retail did not drive the beat: North America operating margin is ~−18 bps YoY excluding the tariff refund; International margin is flat at 4.1%. Advertising is the second accelerant — $19.8B, +26%, breaking a four-quarter ~+22% plateau, with agentic ad surfaces converting 48% more often and spending 21% more.

Guidance reads as deceleration and is not. Q3 net sales $197–202B (+9–12%) embed ~400 bps Prime Day timing drag and ~80 bps FX headwind. Underlying guided growth is ~13–16% (~15.5% mid) — essentially flat vs Q2 underlying. Operating income $22.5–26.5B (mid $24.5B) is the conservative line. Nine months of FY2026 OI at the guide LOW is $73.8B — already $7.9B above the $65.9B full-year street EBIT in amzn_consensus.json, with Q4 still ahead. Street EBIT must reprice.

Constraint flipped from demand to supply. Q1 language stressed memory supply under control; Q2: “we will still not have enough capacity to meet all the demand we have in 2026, and I believe this dynamic will also be true in 2027.” RPO backlog $496B (triple-digit YoY; ~2.5× Q3'25). Capex raised to ~$220B from ~$200B on memory cost inflation — still capacity-short. TTM FCF −$7.6B (from +$18.2B) is the framework red flag (rising revenue, falling FCF), offset by TTM OCF +$161.4B (+33%) and contracted backlog.

Tone and contradictions. Management shifted from asking for patience (Q1 “confident”) to publishing ROIC unit economics unprompted (servers break even <3 years; useful life ≥5–6 years; AI capacity contracted ≥5 years). 9 contradictions found (4 high-severity) — all load on the capital/ROIC narrative (Nova “frontier model” already launched vs still “pursuing”; memory “right spot” vs +$20B raise; useful-life cut for AI vs ROIC assumes extension; server purchase lead-time). None touch the AWS operating facts.

Product-line drivers (above consolidated)

Trajectory of YoY rates (not levels):

Segments

Consolidated 8-quarter view

Do not trend GAAP EPS through Q2'26. Use operating income (+43% / +37% clean) as the earnings trajectory.

Revenue and AWS growth trajectory

One-line verdict: Accelerating — multi-quarter AWS-led re-acceleration with expanding consolidated and AWS margins; FCF deliberately sacrificed to still-insufficient AI capacity.

This quarter vs consensus

Clean EPS bridge: pre-tax $80.9B − other income $53.4B ≈ $27.4B; tax @ ~22.5% ETR → clean net ~$21.3B ÷ 10,903M diluted shares ≈ $1.95.

Heatmap — last 8 quarters (THIS quarter highlighted)

Management variance drivers: AWS acceleration + margin expansion (real/recurring); Prime Day timing (~400 bps optical); $1.2B one-offs (non-recurring — CFO volunteered unprompted); retail not the beat source.

Q3 2026 guide (issued 2026-07-30)

FQ+1 vs prior vs consensus

Amazon guides two lines, one quarter forward only. No EPS/segment/AWS guide.

Optical guide waterfall (sales)

Read: reported guided deceleration is almost entirely Prime Day + FX. Underlying sales growth guided flat, not down.

The number that has to move

With Q4 still ahead (seasonally large), street FY26 EBIT needs a large upward revision. Revenue consensus is closer to fair.

Tone vs Q1 call

Inflection points

1. Q1'25 trough — consolidated rev YoY +8.6%; AWS +16.9%.
2. Q2'25 re-acceleration — rev +470 bps step; multi-quarter climb begins.
3. Q1–Q2'26 second wave — two +300 bps consecutive steps; AWS +830 bps to +36.7%.
4. AWS margin flip — compression → +645 bps expansion (capacity optimization + mix + efficiency; +520 clean).
5. Cash inflection (negative) — TTM FCF from +$18.2B → −$7.6B on AI capex (OCF still +33%).

Plain English: Amazon is mid-reacceleration led by AWS AI + core pull-through, with advertising as a second engine. Management is supply-constrained, not demand-constrained. The operating story is excellent; the cash story is a deliberate multi-year investment cycle with incomplete funding disclosure.

Pattern: demand/product questions answered expansively; capital, capacity timing, and returns questions deflected. No analyst asked about the $53.4B Anthropic gain or TTM FCF −$7.6B.

9 found (4 high / 3 medium / 2 low). All load on capital/ROIC framing; none invalidate AWS +36.7% / $496B RPO / 39.4% margin.

Not forced as contradictions: pure capex raise (disclosed revision); “margins will fluctuate” vs good print; FCF headwinds while arguing medium-term ROIC (explicit tension, not silent flip). Q1 “not capacity constrained there” = memory supply scope, not total AWS capacity.

Macro

Companies / entities mentioned

Amazon printed an AWS-quality quarter: fifth straight acceleration, record sequential revenue add, margin expansion in the high-margin segment, advertising re-acceleration, and a contracted $496B backlog that makes the capacity story credible. Use clean revenue (~15–16% underlying), OI (~$26.3B / +37%), and EPS (~$1.95) — not GAAP EPS. Treat Q3 +9–12% guide as optical; underwrite ~15% underlying and ~$24.5B OI mid with historical beat bias. The single red flag is cash (TTM FCF −$7.6B, capex ~$220B, incomplete funding answer). Underwrite ROIC on Amazon’s older, more conservative disclosures (5-year life, longer lead times), not only the Q2 ROIC speech. BUY on operations with a deliberate discount for capital-cycle disclosure risk.