⚡ Earnings Preview NVDA · NASDAQ Semiconductors / AI Infrastructure Pre-Print Published Apr 29, 2026 · Print: May 27 · AMC

NVIDIA Corporation
Q1 FY27 Earnings Brief

The $78B guide is a floor, not a ceiling. NVIDIA baked in zero China revenue — any China commentary is pure upside. The real question is whether Q2 guidance clears $85B, which is what the buy-side needs to re-rate the stock.

Print Date
May 27, 2026 AMC
Fiscal Quarter
Q1 FY2027
Revenue Consensus
$78.4B
Our Call
$80–81B
📊
01 / 07
Consensus Table — Q1 FY27

Street consensus for the quarter ending April 27, 2026. Management guided revenue of $78.0B ±2% — the consensus has converged around $78.4B. Critically, the guide explicitly excludes all China Data Center compute revenue. Any China shipment approval is unmodeled upside. Gross margin guided at 75.0% non-GAAP ±50 bps, reflecting Blackwell Ultra (GB300) mix ramp.

Metric Consensus Est. Note
Total Revenue $78.4B Mgmt guide $78.0B ±2%
YoY Revenue Growth ~77% Q1 FY26 base = $44.1B
Non-GAAP EPS $1.76 Range $1.74–$1.78
Data Center Revenue ~$71B ~91% of total
Gaming Revenue ~$3.5B GeForce / RTX 50 ramp
Automotive Revenue ~$650M DRIVE platform
Non-GAAP Gross Margin ~75.0% Guided 75.0% ±50 bps
China DC Revenue $0 Explicitly excluded from guide
Q2 FY27 Street Bull ~$83–85B Buy-side needs ≥$85B to re-rate
🔍
02 / 07
What the Street Is Missing — 5 Angles

Consensus is anchored to the $78B guide. These are five vectors where buy-side models are either absent or systematically underweighted — each is a potential incremental upside source not captured in the $78.4B consensus estimate.

Angle 01
China upside is a free option
The $78B guidance bakes in exactly $0 in China Data Center revenue. China historically represented 12–20% of NVIDIA's revenue. The H20 was partially restored under a 15% US government fee structure. The $50B+ China AI infrastructure market is in flux, not dead. Every $1B of China DC revenue that flows through in Q1 adds approximately $0.04 to non-GAAP EPS. The market is not pricing this optionality. A single China commentary signal on the call could move the stock independent of Q1 actuals.
Angle 02
Networking is structurally underappreciated
Q4 FY26 networking revenue hit $10.98B (+263% YoY) — a number that would be a standalone Fortune 500 segment at most companies. Spectrum-X Ethernet has reached parity with InfiniBand for new AI cluster deployments, running at over $10B annualized. Sell-side models embed networking as a Data Center line item without breaking out growth rates or attach dynamics. As Ethernet-based AI clusters scale faster than InfiniBand, NVIDIA's networking share within its own Data Center revenue is expanding — margin-accretive and undermodeled.
Angle 03
Sovereign AI is contracted backlog, not pipeline
Street models treat sovereign AI as a vague future TAM. It's not. HUMAIN (Saudi Arabia) committed to 600,000 GPUs. Stargate UAE (G42) has a 1GW facility underway. These are contracted backlog items with delivery schedules through end-CY26. The aggregate committed backlog from sovereign AI programs is estimated at $500B through end CY26. This is not demand risk — it's supply-constrained delivery. Revenue recognition will hit Q1–Q4 FY27 as GB300 systems are racked and delivered.
Angle 04
GB300 cost structure is improving faster than modeled
GB300 (Blackwell Ultra) represented approximately two-thirds of Blackwell revenue in Q4 FY26. TSMC CoWoS-L packaging yields on the GB300 NVL72 configuration are maturing — early production runs in H2 2025 had meaningful scrap rates. As yield improves with volume, unit cost per GB300 system declines while ASP holds. This translates directly to gross margin expansion. The 75.0% guide is conservative by 50–100 bps on a yield-adjusted basis if Q1 production volumes were in the 45,000–55,000 NVL72 rack range.
Angle 05
Inference scaling is a second, independent demand curve
The market models AI chip demand as primarily a training story. Reasoning models (long-chain-of-thought architectures) create a second independent demand curve through inference. A single complex query on a reasoning model consumes 10–100x the compute of a standard inference call. Hyperscaler utilization rates are reported near 100% — not because supply is constrained, but because inference demand is expanding to fill every GPU available. This dynamic is not fully reflected in consensus out-year estimates, which model training-centric demand growth curves.
⚖️
03 / 07
Bull / Bear / Base — Q1 FY27 Scenarios

Three scenarios for May 27. Probabilities reflect both fundamental likelihood and the specific expectation bar entering the print. NVDA's risk profile is asymmetric toward base: a strong beat without a massive Q2 raise likely produces a muted reaction given 44 of 49 analysts are already at Strong Buy.

🟢 Bull Case
35%
Rev: $82–84B
DC: $74–76B
GM: 75.5%
Q2 Guide: $88–92B
→ +10–15% AH
🔵 Base Case
45%
Rev: $78–81B
DC: $70–73B
GM: 74.8–75.2%
Q2 Guide: $83–87B
→ +2–7% AH
🔴 Bear Case
20%
Rev: <$77B
DC: <$68B
GM: <74%
Q2 Guide: $78–82B
→ -8–15% AH
// The single number that matters most

The Q2 FY27 revenue guide is the primary stock catalyst — more than Q1 actuals. Buy-side consensus is around $86B+ for Q2. A Q2 guide at $85B+ signals acceleration and re-rates the stock. A Q2 guide at $83–84B (in-line with street) is neutral to slightly negative given expectations. Gross margin below 74.0% — which would signal GB300 yield or mix issues — is the bear case trigger independent of revenue.

📋
04 / 07
8 Watch Items on the May 27 Call

The call begins approximately 90 minutes after market close (~5:30pm PT). Jensen Huang leads; Colette Kress handles the financial bridge. Watch these eight signals in order of market impact:

// Ordered by market impact
  • Q2 Revenue Guide: Consensus ~$82B; buy-side needs $86B+. The difference between $85B and $87B is the stock reaction gap. Listen for whether Jensen references specific sovereign AI delivery schedules — that would imply confidence in the guide's structure.
  • Q2 Gross Margin Guide: 75.0%+ confirms GB300 yield is on track. Anything below 74.0% is a red flag — it implies either yield degradation, unfavorable system mix, or margin dilution from a large sovereign AI delivery (higher system integration content lowers chip margin). Expect 74.5–75.5% in the guide.
  • China Commentary: Any language indicating China DC compute was in Q1 actuals — or will be in Q2 — is incremental upside vs. the $0 modeled. Watch for Jensen to distinguish between H20 (lower-end, partially approved) and Blackwell (strictly prohibited). A single sentence on China optionality can move the stock 3–5% independent of revenue.
  • Sovereign AI Revenue Specificity: Street wants dollar amounts attached to HUMAIN (Saudi) and UAE (G42/Stargate). If Jensen quotes specific Q1 sovereign revenue (e.g., "sovereign contributed $X billion in Q1"), it converts pipeline estimates into recognized revenue — highly bullish for out-year modeling.
  • Networking % of Data Center Revenue: Q4 FY26 networking was 17.6% of Data Center. If this percentage is expanding — and if Jensen confirms Spectrum-X is now the default for new hyperscaler cluster builds — the embedded networking growth story becomes investable on its own. Watch for any commentary on "InfiniBand vs. Ethernet" attach rates.
  • Rubin Platform Timing: Vera Rubin (H2 FY27) — is the schedule still H2 2026? Any pull-forward to Q2 FY27 ship dates is meaningfully bullish. A delay signal to early FY28 would be a negative catalyst for the transitional revenue bridge. Jensen typically mentions GTC roadmaps on earnings calls — listen for tightening of the Rubin range.
  • Hyperscaler Capex Commentary: Does Jensen directly reference the $660–690B combined hyperscaler capex commitment for CY2026? How does he frame the demand-vs-supply dynamic? "Supply-constrained" framing = bullish; any reference to "softening" or "order digestion" = immediate negative.
  • Supply vs. Demand Status: Q4 FY26 commentary indicated supply was tight. Q1 FY27 shipped into continued strong demand with GB300 NVL72 racks at Foxconn volumes of 50,000–60,000 for the year. Any acknowledgment that supply loosened and demand absorbed it (demand > supply still) is bullish. A comment that supply is now ahead of demand would suppress Q2 guidance sentiment even if actuals beat.
🎯
05 / 07
Pre-Print Call — Beat and Raise
📬 Post-Print Update — NVDA prints May 27 AMC

Get the post-print verdict. We'll email you within 24h of the print with actuals vs. our call, graded prediction score, and updated bull/bear range.

⏳  PRE-PRINT — This is our pre-print call. Actuals and final verdict to be added within 24 hours of the May 27 earnings close.
// Vektor Pre-Print Call — Q1 FY27
BEAT AND RAISE
Q1 Revenue
$80–81B
Q1 Data Center
$72–73B
Q1 Non-GAAP EPS
$1.80–1.82
Q2 FY27 Guide
$85–87B

Our Q1 estimate ($80–81B) is $1.6–2.6B above consensus, driven by: GB300 NVL72 rack deliveries ahead of the production schedule implied in guidance; sovereign AI backlog recognition in Q1 from HUMAIN and UAE programs; and networking mix expansion. Our Q2 guide call ($85–87B) reflects confirmed sovereign delivery schedules, hyperscaler capex commitments for H1 CY2026, and zero China upside embedded (upside from any China approval is additive to this estimate).

Graded Prediction: NVDA closes +5% or more on May 28, 2026 IF Q1 revenue beats consensus by ≥$1B AND Q2 FY27 guidance is ≥$85B. Both conditions must hold. A Q1 beat on revenue alone — without the Q2 raise — is insufficient given the expectation bar. We will score this prediction publicly in the post-print update.
🔔 Post-Print Actuals Brief
Notify me when NVDA actuals drop

We'll email you the actuals brief — our call vs. consensus, graded prediction score, and updated thesis — within 24h of the May 27 close.

🔗
06 / 07
Comp Setup — AI Semi Print Trail

Three upstream AI chip prints in April–May 2026 create a directional read-through to NVDA. All three confirm the demand environment is intact. The AMD and ARM prints are particularly relevant — AMD prints May 5 with direct MI-series GPU commentary, and ARM prints May 6 with royalty volumes that directly reflect Blackwell silicon production ramp.

AVGO — Broadcom
Q1 FY26 · Print: Mar 4, 2026
✅ Bullish
AI semiconductor revenue hit $8.4B in Q1 FY26 (+106% YoY). Q2 FY26 guided at $10.7B (+140% YoY) — a massive sequential beat. Custom ASIC demand (from hyperscaler TPU/MTIA programs) is accelerating, which confirms overall AI infrastructure spend is growing, not substituting GPU spend. Broadcom's hyperscaler customer base (Google, Meta, Amazon) overlaps completely with NVDA's top 5 customers. Strong Broadcom AI semi = hyperscalers are still spending hard on AI infra broadly.
→ NVDA read-through: confirms hyperscaler AI capex is accelerating into Q1/Q2 CY26
AMD — Advanced Micro Devices
Q1 2026 · Print: May 5, 2026
✅ Bullish (beat expected)
AMD guided Q1 Data Center ~$3.3B+ with MI350 ramping H2 2026. AMD's hyperscaler GPU deployments (MI325X at Microsoft, Meta, Oracle) are competitive proof that GPU demand is exceeding supply broadly — not NVIDIA-specific. If AMD beats Data Center estimates on May 5, it confirms the entire AI GPU market is supply-constrained, which is the most bullish possible setup for NVDA's Q2 guide. AMD Lisa Su's commentary on inference demand is the specific indicator to watch — it validates NVDA's Angle 05 (inference as second demand curve).
→ NVDA read-through: AMD DC beat = GPU demand is broad-based, not concentrated; validates NVDA Q2 guide upside
ARM Holdings
Q1 FY26 · Print: May 6, 2026
✅ Structural Validation
ARM royalty revenue is a direct read-through to Blackwell silicon volumes — every GB300/GB200 chip shipped by TSMC includes ARM-based Grace CPU cores that generate ARM royalties. ARM Q1 FY26 royalties directly reflect Q4 CY2025 / Q1 CY2026 production volumes of Blackwell SoCs. A royalty beat confirms TSMC's CoWoS-L ramp is ahead of NVIDIA's own conservative guide assumptions. ARM's licensing segment commentary on next-generation AI SoC design wins is also a forward indicator for the Rubin platform supply chain.
→ NVDA read-through: ARM royalty beat = Blackwell production volumes are higher than NVDA guided; supports $80–81B Q1 estimate
📎
07 / 07
Sources & Research Base
📬 Get the NVDA post-print verdict — May 28

We'll score our call against actuals and send you the verdict within 24 hours of the print. No noise — just the brief, the actuals, and what we got right and wrong.

This brief is grounded in 18 primary sources. Full URLs are available in Report #571353. Research current as of April 29, 2026.

NVIDIA Q4 FY26 Earnings Release & 10-K filing (February 26, 2026) — primary financial data, segment breakdown, Q1 FY27 guidance
NVIDIA Q4 FY26 Earnings Call Transcript — Huang/Kress commentary on Blackwell ramp, China posture, networking
NVIDIA GTC 2026 keynote transcript (March 2026) — Vera Rubin unveil, Jensen Huang $1T cumulative revenue guidance
HUMAIN-NVIDIA GPU commitment announcement (May 2026) — 600,000 GPU agreement, HUMAIN government AI infrastructure plan
Stargate UAE / G42 1GW facility announcement (April 2026) — sovereign AI infrastructure program detail
Foxconn Chairman annual forecast (Q1 2026) — 50,000–60,000 GB300 NVL72 racks for CY2026 projection
TSMC CoWoS-L packaging yield analysis (Morgan Stanley, March 2026) — GB300 yield maturation trajectory
Broadcom Q1 FY26 earnings transcript (March 4, 2026) — AI semi $8.4B, Q2 guide $10.7B, hyperscaler ASIC commentary
AMD Q1 2026 earnings preview (analyst consensus compilation, April 2026) — MI325X/MI350 data center estimates
ARM Holdings Q4 FY25 royalty data (February 2026) — Grace CPU core royalty trends, AI SoC licensing pipeline
Hyperscaler capex analysis: Amazon $200B, Microsoft $120B+, Google $175–185B, Meta $115–135B, Oracle $50B — individual company Q4 2025 and Q1 2026 earnings guidance
BofA Securities hyperscaler cash flow analysis (March 2026) — 94% operating cash flow consumed by capex figure
TrendForce AI accelerator market share report (Q1 2026) — custom ASIC 44.6% vs. GPU 16.1% growth differential
Wells Fargo NVDA pre-print note (April 25, 2026) — $78.4B consensus, EPS range $1.74–$1.78
Wolfe Research NVDA initiation (April 2026) — networking undervaluation thesis, Spectrum-X attach rate analysis
Loop Capital NVDA upgrade note (April 2026) — sovereign AI backlog quantification methodology
DeepSeek V4 technical report (April 23, 2026) — Huawei Ascend 950PR architecture, implications for China GPU policy
NVIDIA H20 export partial resumption framework (July 2025) — 15% US government fee structure for Chinese H20 exports
Data current as of April 29, 2026. Financial figures sourced from NVIDIA SEC filings and earnings releases. Consensus estimates aggregated from Wells Fargo, Wolfe Research, Loop Capital, Morgan Stanley, and BofA Securities. Forward-looking statements and scenario probabilities represent Vektor analysis, not investment advice. Section 5 Pre-Print Call will be graded publicly in the post-print update within 24h of the May 27, 2026 earnings close.
// You might also be interested in
⚡ Vektor Intelligence

Get a brief like this
for any ticker — $29/mo

Drop in any company name. 40 seconds. Full 7-section intelligence brief: consensus estimates, differentiated angles, bull/bear/base scenarios, watch items, and pre-print call.

Start for $29/month → — or get your first brief free —
✓ You're in — we'll send your free brief shortly.

No credit card. Cancel anytime.