Will Azure constant-currency growth be at or above 38% in Q3 FY26?

Three-of-three recent quarters where Azure beat the implicit guide by 100-200bps. With Q3 FY26 guide of +37-38%, midpoint is 37.5%; even a 0-100bps beat lands at or above 38%. Hood's 'would have been over 40%' Q2 disclosure means latent demand well above realized rate — supply unlock translates directly to growth. Atlanta + Wisconsin AI super factory linked Q2 + Maya 200 silicon online + ~1 GW added per quarter mean Q3 capacity adds support holding +37-38%. Beyond-12-month RPO +56% YoY shows long-duration backlog accelerating. Lean YES at 68%.

The threshold (+38%) is at the high end of explicit +37-38% guide. Base rate for landing at-or-above the high end of MSFT's guide range historically ~50-55%; another ~30-35% probability of landing above the high end. Combined ~80-85% probability of clearing +38%. But beat magnitude has been compressing (Q3 FY25 +400bps over implicit guide → Q1-Q2 FY26 +200bps over guide); trend says Q3 FY26 beat magnitude trends toward 0-100bps. Capacity-constrained framing has held 4 quarters at +39% — eventually supply rate becomes binding constraint and print converges to guide. Q3 capex sequentially decreasing (finance lease timing) reduces incremental capacity. Net 65%.

Q3 quarter is essentially complete by prediction date. Hood pattern: she guides to supply-feasible floor and lets demand do the rest — Q3 FY25 was guided ~31-32%, printed +35% (large beat). Q4 FY25 guided +37-38%, printed +39%. Beat pattern of 100-200bps over guide midpoint puts central tendency at +38.5-39.5%. The question is whether something specific to Q3 FY26 makes the pattern break — Hood's deceleration acknowledgment is a real signal but the 'would have been over 40%' framing suggests demand-driven softness is not the issue. Tilt toward 70%.

Direct read: Hood explicitly guided +37-38% with documented capacity-constrained demand for 4 quarters; she has beat by 100-200bps three quarters running. Threshold at the high end of guide. Most likely outcome is print at +38-39% range. Tilt against: beat magnitude compressing, capex sequentially decreasing in Q3 means incremental capacity moderates. Net 66% YES.

Base-rate-anchored view. Q3 FY26 is the FIRST sequential deceleration since Q3 FY25 — first-deceleration prints have limited track record. Hood deceleration acknowledgment is a real signal that supply-capped narrative may be transitioning. OpenAI workload accounting opacity could shift toward longer realization horizon dampening reported cc growth. The +39% cc plateau has lasted 3 quarters which is on the long end of hyperscaler plateaus before transition. Probability lower-end of context range, ~58%.

Balanced. The +38% threshold sits at the high end of the +37-38% guide. With beat magnitude compressing (now ~100-200bps from prior 200-400bps), central tendency moves toward +37.5-38.5%, putting roughly 60-70% of the probability mass at-or-above 38%. Anthropic Nov 2025 commitment + Foundry 80K customers + non-OpenAI RPO +28% YoY add demand pipeline. Net 67%.

Hood guides +37-38%; she beat by 100-200bps 3 of 3 quarters. Capacity unlock continuing. Demand documented above realized rate. Lean YES 65%.

Devil's-advocate: first sequential deceleration is regime change; beat magnitude compressing means midpoint convergence; capex sequentially decreasing reduces incremental capacity. Modest tilt YES at 55%.

Optimist read: Hood typically guides to supply-feasible floor and lets demand do the rest. With latent demand at +40%+ per Hood Q2 disclosure, capacity unlock translates directly. 3-of-3 beats. Multiple non-OpenAI demand engines. 68% YES.