Most important: Q1 capex cadence vs $175-185B annual guideIn the print: Capex $, Cloud revenue, Cloud backlog, EPSOn the call: H1/H2 capex split, Ironwood TPU ramp, 2027 anchoringDownstream: NVDA, AVGO, TSMC, SK Hynix, Micron, ETN
House view
Street prices GOOGL on Search durability and Cloud growth. The $175-185B 2026 capex guide is the largest single-year infrastructure commitment among hyperscalers. We read this print primarily as a Q1 cadence check on that guide, which tightens NVDA, AVGO, and SK Hynix allocations through year-end.
What is priced in
Street expects Q1 revenue ~$106.8B (+20% YoY), EPS ~$2.62 (+29% YoY), and quarterly capex consistent with the $175-185B FY guide mid-point (~$45B/quarter). Not priced in: whether Q1 capex runs front-loaded above $45B (pulling forward GPU/TPU deliveries), an update to Cloud backlog beyond the $240B disclosed in Q4, or a shift in silicon mix toward Ironwood TPU that would alter the NVDA vs. AVGO allocation split.
What decides the capex readRanked by capex-trajectory impact
Priority 1 · primary capex signal
Q1 capex cadence vs $175-185B annual guide
Q4 2025 capex was $27.9B. Hitting the $175-185B guide requires a step-up to ~$44-46B per quarter. Whether Q1 runs at, above, or below that pace reveals if the buildout is accelerating or if supply bottlenecks are gating deployment.
✓Q1 capex at or above $45B with management reaffirming or raising the $175-185B FY guide, signaling GPU and TPU deliveries are on schedule.
✗Q1 capex below $40B with language citing supply delays, permitting issues, or power availability constraints that push spend into H2.
Priority 2
Custom silicon mix shift: Ironwood TPU vs third-party GPU
Ironwood (7th-gen TPU) is purpose-built for inference at scale. A faster ramp shifts capex dollars from NVDA toward AVGO and TSMC custom silicon lines, changing the downstream allocation read even if total capex holds steady.
✓Management quantifies Ironwood TPU deployment scale (pod count or percentage of inference fleet) and names it as a growing share of 2026 capacity additions.
✗No Ironwood deployment update, with commentary emphasizing continued reliance on third-party GPUs for both training and inference workloads.
Priority 3
Cloud backlog trajectory and 2027 forward anchoring
Cloud backlog jumped from $155B to $240B in Q4 2025 (55% QoQ). Sustained backlog growth above $240B would validate the $175-185B capex guide and signal demand visibility extending into 2027, anchoring multi-year infrastructure spend.
Working the eventRelease drop vs. Q&A
In the release · first 60 seconds
Quarterly Capex
$27.9B (Q4 2025)
Does Q1 capex reach $45B or above, consistent with the $175-185B annual guide mid-point? A print below $40B signals supply or siting delays.
FY 2026 Capex Guide
$175-185B (guided Q4 2025)
Is the $175-185B range reaffirmed, narrowed, or raised? Any tightening of the range toward the high end is a positive signal for downstream names.
Google Cloud Revenue
$17.66B (Q4 2025)
Does Cloud revenue sustain above $17B with growth rate holding near 30% YoY? Deceleration below 25% YoY would question whether backlog converts to near-term revenue.
Total Revenue
Street expects ~$106.8B (+20% YoY)
Does total revenue clear $107B? A miss below $105B would pressure the FCF math against the massive capex ramp.
EPS
Street expects ~$2.62 (+29% YoY)
Does EPS clear $2.65 despite the capex step-up? A miss below $2.55 combined with high capex would compress the multiple.
Cloud Backlog
$240B (Q4 2025, +55% QoQ)
How the stock likely tradesQualitative
What this means downstreamNamed names
Capex at or above $45B + FY guide reaffirmed/raised + Cloud revenue above $18B
Sharp rally. Revenue growth funds the capex ramp without FCF deterioration. Market rewards both the demand signal and the earnings quality.
Capex at or above $45B + FY guide reaffirmed + Cloud revenue below $17B
Flat to down. Massive spend without matching Cloud acceleration raises ROI concerns. FCF compression weighs on the multiple despite the capex commitment.
Mixed. Strong Cloud growth is positive, but the capex shortfall raises questions about supply delays or execution risk on the back-loaded spend profile.
Sharp selloff. Demand deceleration combined with a guide cut signals the AI infrastructure cycle may be peaking. Multiple compresses on both growth and capex narratives.
Capex at or above $45B + FY guide reaffirmed/raised + Cloud revenue above $18B
NVDA GB-series delivery cadence confirmed. AVGO TPU design backlog firms. SK Hynix and Micron high-bandwidth memory allocations tighten through Q3. ETN data center power backlog extends into 2027.
Capex at or above $45B + FY guide reaffirmed + Cloud revenue below $17B
NVDA and AVGO still benefit from the spend commitment regardless of Cloud revenue softness. SK Hynix high-bandwidth memory demand intact. TSMC CoWoS utilization stays full. Downstream reads bullish even as GOOGL stock fades.
2031 signals · 70 high-qualityResearch read-through · not a trade recommendation
✓Cloud backlog above $260B with management offering any directional commentary on 2027 capex staying elevated or accelerating.
✗Backlog growth decelerating below 10% QoQ with no 2027 capex commentary, suggesting demand is plateauing relative to committed spend.
Does backlog grow above $260B? Flat or declining backlog would undercut the demand justification for the $175-185B capex guide.
On the call · where capex read moves
H1/H2 capex split and quarterly cadence guidance
Front-loaded spend confirms GPU/TPU delivery schedules are intact and tightens NVDA and SK Hynix high-bandwidth memory allocations through mid-2026.
Ironwood TPU deployment scale and inference fleet share
Quantified Ironwood ramp shifts the downstream read from NVDA toward AVGO (TPU design partner) and TSMC (fabrication). Silence keeps the GPU-heavy read intact.
Gemini API token throughput update (was 10B tokens/min in Q4)
Continued rapid growth in inference token volume validates the inference-heavy capex thesis and sustains demand for high-bandwidth memory at SK Hynix and Micron.
Power and data center siting constraints
Any mention of MW commitments, grid interconnection timelines, or power procurement difficulties reads directly to ETN and PWR order visibility.
Forward 2027 capex or multi-year infrastructure commentary
Even directional language on 2027 spend staying elevated would extend the demand runway for NVDA GB-series, TSMC CoWoS capacity, and ETN power infrastructure.
Memory supply or component availability commentary
Amazon flagged memory chip supply as a potential risk factor. Any similar language from Alphabet would confirm high-bandwidth memory tightening at SK Hynix and Micron.
Near-term NVDA and SK Hynix delivery schedules may be slipping. TSMC CoWoS orders could shift to H2. ETN and PWR see order timing risk. Back-loaded guide preserves full-year demand but creates a Q2-Q3 air pocket.
NVDA faces order reduction risk from a top-3 customer. AVGO custom silicon pipeline narrows. SK Hynix and Micron high-bandwidth memory pricing softens. TSMC CoWoS utilization drops. ETN power backlog shortens.