Faircurve Equity Pulse — NVDA — 2026-05-16

NVIDIA Corporation closes 2026-05-16 at $225.32, +67.1% over the past year from the May-2025 low of $129.16. The Feb-2026 print delivered $68.1B in Q4 revenue (+72% YoY) and FY26 free cash flow of $96.7B at a 44.8% margin. Sell-side consensus models FY27 revenue at $369B (+71%) and EPS at $8.31. The paradox: on those consensus FY27 numbers NVDA trades at 27× forward P/E — a 35% discount to peer cluster AMD/AVGO/TSM/MRVL at 42× despite +26pp higher growth. Base case $241 (+6.8%), bull $329, bear $136, with effectively zero net debt and a $500B+ Blackwell/Rubin pipeline through CY26.

By Faircurve Research

NVIDIA & the Discount

Faircurve Equity Pulse · Single-Name Research
SATURDAY, MAY 16, 2026 NVDA · NASDAQ · Semiconductors / AI Accelerated Compute
Spot $225.32 · 52W $129.16 — $236.54
HOLD · PT $241

NVIDIA Corporation closes 2026-05-16 at $225.32, +67.1% over the past year from the May-2025 low of $129.16. The Feb-2026 print delivered $68.1B in Q4 revenue (+72% YoY) and FY26 free cash flow of $96.7B at a 44.8% margin. Sell-side consensus models FY27 revenue at $369B (+71%) and EPS at $8.31. The paradox: on those consensus FY27 numbers NVDA trades at 27× forward P/E — a 35% discount to peer cluster AMD/AVGO/TSM/MRVL at 42× despite +26pp higher growth. Base case $241 (+6.8%), bull $329, bear $136, with effectively zero net debt and a $500B+ Blackwell/Rubin pipeline through CY26.

Price
$225.32
+67.1% 1Y · -4.4% Fri
Market Cap
$5.48T
~24.4B diluted shares
12-Month Target
$241
+6.8% · vs Street $277
Fwd P/E (FY27)
27.1×
peer mean 41.8×
NVDANVIDIA Corporation1-YEAR PRICE HISTORY$225.32PRICE50DMA200DMA$100$150$200$250$30052W HIGH $236.542026-05-1452W LOW $129.162025-05-23Jul 25Oct 25Jan 26Apr 26SPOT1Y RETURN+67.1%MAX DRAWDOWN-20.2%30D VOL (ann.)38.1%BETA (5Y)2.24
§ 01 — Tape & Setup

i.The Snapshot

MetricValueContextYoY / vs Peer
FY26 Revenue$215.9Bjust-closed FY26+65.4% YoY
FY27 Revenue (consensus)$369.4B39 analysts+71.1%
FY26 Free Cash Flow$96.7B44.8% FCF margin+59% YoY
FY26 Capital Returned$41.0B$40B buybacks + $1B div~0.7% of mcap
Non-GAAP Gross Margin (Q4 FY26)75.2%mid-70s FY27 guide+1.7pp QoQ
52W Range$129.16 — $236.54spot $225.32+67.1% 1Y
The 27× forward P/E is the headline finding. NVIDIA — the highest-growth, highest-margin, highest-share name in AI silicon — trades at a 35% discount to peer-average forward P/E and is the only one of the five at sub-30×. "Cheaper than AMD and Intel" is mathematically correct on forward consensus. The market is paying for execution certainty on a $369B FY27 revenue number, not optimism. Base case sits just +6.8% above spot because consensus is already in the price; bull case (+45.9%) requires FY28 capacity disclosure on the Q1 FY27 call (May 20, 2026).
§ 02 — Thesis · Bull vs Bear

ii.The Debate at $225

Half-trillion-dollar pipeline through CY26.

Q3FY26 disclosed $500B Blackwell + Rubin revenue visibility through end-CY26. By Q4FY26 mgmt said the figure had been 'exceeded' with line-of-sight into CY27. This is the single most concrete pipeline ever disclosed by a hardware company.

Forward multiple discount to peers despite +71% FY27 growth.

NVDA at 27x forward P/E vs peer-avg 42x and AMD 57x. NVDA's revenue growth premium over peers is +26pp — historically reasoned multiples should compress this gap, not widen it. Multiple expansion alone (to 33x) is +22% upside.

Free-cash-flow machine, fortress balance sheet.

FY26 FCF $96.7B on $216B revenue = 44.8% FCF margin (and Q1 FY26 included a $4.5B H20 write-down). $74.2B cash and short/long-term investments vs $11.4B debt = effectively zero net debt at $5.5T market cap. Bought back $40B of stock in FY26.

Sovereign + Rubin + inference reasoning = three new pillars.

Sovereign AI run-rate scaled from high-single-digit billions (FY25) to ~$30B in FY26 with EU, Saudi, UAE, India anchors. Rubin samples have shipped — first-pass yield will determine whether the CY27 ramp matches Blackwell's. Inference reasoning models drove a 10x compute step-up post-DeepSeek R1.

Beta 2.24 — fair value collapses under any cycle scare.

On CAPM (Beta 2.24 × 5% MRP + 4.2% Rf = 15.4% cost of equity), DCF intrinsic falls to ~$90 even with 50% FCF margin assumptions. The stock is uniquely sensitive to AI-narrative reversal — a 1pp WACC move shifts intrinsic value 15-20%.

Hyperscaler capex digestion is the consensus single-point-of-failure.

Top-4 hyperscaler CapEx 'doubled in two years' to ~$700B. A 20% pullback in CY27 capex on any of: post-Stargate digestion, regulatory action, or a single hyperscaler 'pause' headline — flips FY28 consensus revenue by 15-20% and shifts NVDA into multiple-compression territory.

China remains zero-ed; geopolitical re-escalation is open-ended downside.

Q1FY26 lost $50B of TAM on H20 export controls; subsequent licenses arrived with 15% USG revenue-share, $0 shipped. Trump-Xi H200 discussions (May 2026 news) reopened the question. Worst-case: an EU/India/Japan equivalent restriction would compound the loss.

Margin sustainability under Rubin and Blackwell Ultra transitions.

Non-GAAP gross margin trough was 61% (Q1 FY26, H20 charge); ex-charge 71.3%. Each next-gen transition (Rubin in CY27, Vera Rubin CPX/Ultra in CY28) compresses margins 200-400bps during ramp. Two compressions in two years could keep GMs from sustainably returning above 75%.

Three of the five most-asked-about questions on NVDA are already answered. Consensus FY27 (+71% revenue, +70% EPS) is priced into the 27x forward multiple. The CY26 pipeline is anchored at $500B+. Rubin samples are out. What remains genuinely uncertain — and what the stock has not yet priced — is whether the FY28-29 capacity buildout (already implicit in the $3-4T AI-infrastructure TAM Jensen now cites quarterly) sustains FY+2 revenue growth above +30% post the FY27 step-function. That is the bull-bear axis. We are HOLDers waiting for either (a) a digestion-quarter sell-off to $170-180 that resets the forward multiple, or (b) explicit FY28 capacity disclosure on the next call that converts our +$280 bull case into a tradeable target.

Three things to watch.
  1. Q1 FY27 print on May 20: gross margin progression (>75% non-GAAP confirms the Blackwell-Ultra ramp is margin-accretive); next-Q guide midpoint vs the $78B implied; explicit FY28 capacity commentary if offered.
  2. China policy headline tape: H200 license clarification, USG revenue-share negotiations, and whether any meaningful $ ships in 2H CY26. Each $5B of recovered China revenue is worth $0.15-0.20 of EPS.
  3. Hyperscaler CapEx guidance from Microsoft, Meta, Google, Amazon, Oracle — if FY27 capex prints flat-to-down vs the +50% trend, that is the dominant signal regardless of NVDA's own guide.
§ 03 — The Latest Print

iii.Q4 FY26 — the ramp continues

Q4 FY26 (reported Feb 25, 2026) delivered $68.1B revenue (+72% YoY, +20% QoQ), Data Center $62.0B, diluted EPS $1.76 (+98% YoY), and non-GAAP gross margin 75.2% — the first quarter ever to print >$50B in Data Center and >$1.70 in quarterly EPS. Management disclosed $500B+ Blackwell + Rubin revenue visibility through end-CY26 had been 'exceeded' with line-of-sight extending into CY27, guided Q1 FY27 to $78B (+15% QoQ), and confirmed Rubin samples have shipped to lead customers.

Operational dashboard, Q4 FY26

MetricValueContext
Total revenue (Q4 FY26)$68.1B+72% YoY, +20% QoQ
Data Center revenue$62.0BBlackwell ~80% of compute
Gaming revenue$3.7B−14% QoQ; supply-shifted to DC
Automotive + Robotics$0.60B+6% QoQ; physical-AI ramp
Networking revenue$11.0B+34% QoQ; Spectrum-X + NVLink
Non-GAAP gross margin75.2%+1.6pp QoQ; recovering toward FY27 mid-70s guide
FCF (Q4)$35.0BFY26 total $96.7B; 44.8% FCF margin
Capital returned to shareholders$41B FY26$40B buybacks, $1B dividends

Trailing 8 quarters — revenue and diluted EPS

QuarterRevenueDiluted EPS
Q1 FY25$26.0B$0.60
Q2 FY25$30.0B$0.67
Q3 FY25$35.1B$0.78
Q4 FY25$39.3B$0.89
Q1 FY26$44.1B$0.76
Q2 FY26$46.7B$1.08
Q3 FY26$57.0B$1.30
Q4 FY26$68.1B$1.76
§ 04 — Five-Year Trends

iv.From $27B to $216B in five years

MetricFY22FY23FY24FY25FY26
Revenue ($B)26.927.060.9130.5215.9
Revenue YoY growth+61%+0%+126%+114%+65%
Gross margin (GAAP)64.9%56.9%72.7%75.0%71.1%
Operating margin (GAAP)37.3%15.7%54.1%62.4%60.4%
Net income ($B)9.84.429.872.9120.1
Diluted EPS ($)0.380.171.192.944.90
Operating cash flow ($B)9.15.628.164.1102.7
Capex ($B)1.01.81.13.26.0
Free cash flow ($B)8.13.827.060.996.7
FCF margin30.2%14.1%44.4%46.6%44.8%
Capex / revenue3.6%6.8%1.8%2.5%2.8%
Total debt ($B)11.812.011.110.311.4
Cash + ST investments ($B)21.213.326.043.262.6
Net debt ($B)9.88.63.81.70.8
Buybacks ($B)10.09.533.740.1
Diluted shares (B)25.425.124.924.824.4

NVDA's five-year arc is one of the cleanest growth-and-profitability prints in modern equity history. Revenue 8x'd from $26.9B (FY22) to $215.9B (FY26) while operating margin expanded from 37% to 60% and FCF margin from 30% to 45%. Capex stays sub-3% of revenue — the TSMC-fabbed, software-moated economics give NVDA the rare combination of hyperscaler-grade growth with asset-light returns. Net debt is effectively zero ($0.8B against $5.5T market cap) and FY26 alone returned $41B to shareholders — $40B in buybacks shrinking the diluted share count from 25.4B to 24.4B. The single discontinuity in the series is FY23 (crypto unwind, channel destocking) — context worth remembering when evaluating bear-case scenarios on hyperscaler digestion.

§ 05 — Earnings Call Signal

v.What management said, eight quarters of nuance

Theme frequency (mentions per quarter, oldest → newest)

ThemeQ1FY25Q2FY25Q3FY25Q4FY25Q1FY26Q2FY26Q3FY26Q4FY26
Data Center / AI4238556058526570
Hopper / H100 / H2002218211411954
Blackwell / B100 / B200 / GB2001824324542383028
Rubin / R1000014271218
Sovereign AI107659846
China / export ctrls / H2012756221489
Networking / NVLink / Spectrum-X2018162228322630
Inference1514223028253035
Software / CUDA / Enterprise2528302832303538
Automotive / Robotics129142224222025

Recurring operational metrics — verbatim from prepared remarks

MetricQ1FY25Q2FY25Q3FY25Q4FY25Q1FY26Q2FY26Q3FY26Q4FY26
Total revenue$26.0B$30.0B$35.1B$39.3B$44.0B$46.7B$57.0B$68.0B
Data Center revenue$22.6B$26.3B$30.8B$35.6B$39.1B$41.1B$51.0B$62.0B
Gaming revenue$2.65B$2.88B$3.3B$2.5B$3.8B$4.3B$4.3B$3.7B
Automotive revenue$329M$346M$449M$570M$567M$586M$592M$604M
Networking revenue$5.0B$7.3B$8.2B$11.0B
Non-GAAP gross margin78.9%75.7%75.0%73.5%61% / 71.3% ex-H2072.7%73.6%75.2%
Capital returned$7.8B$7.4B$11.2B$8.1B$14.3B$10.0B$41B FY
What this tells us.
  1. Product-cycle pivot is real and trackable: Hopper mentions collapsed from 22→4 across the eight quarters while Rubin went from 0→18. Blackwell peaked at 45 (Q4 FY25) then settled at 28-38 as it matured from 'announcement' to 'shipping'.
  2. Data Center revenue went from $22.6B → $62.0B in 8 quarters, a 2.7x absolute step-up. Networking emerged as a distinct disclosed line at $5.0B → $11.0B (Q1 FY26 onward) — the 'systems sale' story is showing up in the financials.
  3. Gross margin traced a clean U: 78.9% → 61% (Q1 FY26 H20 hit) → 75.2% (Q4 FY26). Mgmt's mid-70s guide has held through Blackwell ramp; the Rubin transition in CY27 will be the next stress test of this guide.
  4. China mentions spiked in Q1 FY26 (22x — H20 export controls + $4.5B write-down) then receded to single digits as the segment was zero-ed out of forward guidance. This is now option-value, not embedded revenue.

FY guide track — what management told us each quarter

QuarterNext-Q rev guideGM guideAction / framing
Q1 FY25 (May 2024)$28.0B (Q2 FY25)75.5%; mid-70s FYRaised — Blackwell announced full production; H200 ramp
Q2 FY25 (Aug 2024)$32.5B (Q3 FY25)75.0%; mid-70s FYRaised — Blackwell mask change; Hopper still strong
Q3 FY25 (Nov 2024)$37.5B (Q4 FY25)73.5%; low-70s rampRaised — 'staggering' Blackwell demand
Q4 FY25 (Feb 2025)$43.0B (Q1 FY26)71.0%; mid-70s exitRaised — $11B Blackwell in Q4 (fastest ramp ever)
Q1 FY26 (May 2025)$45.0B (Q2 FY26); $8B H20 loss72.0%; mid-70sMaintained despite shock — $4.5B H20 write-down
Q2 FY26 (Aug 2025)$54.0B (Q3 FY26); no China DC73.5%; mid-70s exitRaised — $3-4T AI infra TAM first cited; sovereign $20B+
Q3 FY26 (Nov 2025)$65.0B (Q4 FY26); +14% QoQ74.8%Raised significantly — $500B Blackwell+Rubin visibility CY26
Q4 FY26 (Feb 2026)$78.0B (Q1 FY27); no China DC75.0%; mid-70s FYRaised — $500B 'exceeded'; Rubin samples shipped; hyperscaler CapEx ~$700B

Verbatim quotes — anchoring the arc

FUTURE-VISION · Jensen Huang · CEO · Q4 FY25 (Feb 2025)
“Going forward, data centers will dedicate most of CapEx to accelerated computing and AI. Data centers will increasingly become AI factories... No technology has ever had the opportunity to address a larger part of the world's GDP than AI. No software tool ever has.”
CANDID-DOWNSIDE · Colette Kress · CFO · Q1 FY26 (May 2025)
“On April 9, the US government issued new export controls on H20... we recognized $4.6B in H20 revenue, but also recognized a $4.5B charge as we wrote down inventory and purchase obligations... Losing access to the China AI accelerator market, which we believe will grow to nearly $50 billion, would have a material adverse impact on our business and benefit our foreign competitors.”
CONDITIONAL-BULL · Jensen Huang · CEO · Q3 FY26 (Nov 2025)
“Currently, we have visibility to a half a trillion dollars in Blackwell and Rubin revenue from the start of this year through the end of calendar year 2026... Over the next five years, we are going to scale with Blackwell, with Rubin, and follow-ons to scale into effectively a $3 to $4 trillion AI infrastructure opportunity.”
Reading the arc.
  1. Every guide print except Q1 FY26 (the H20 shock) was a RAISE. The cadence is now: each quarter mgmt raises forward $-pipeline disclosure by an order of magnitude — '$7B Q4 FY25 Blackwell' → '$500B Blackwell+Rubin through CY26' → 'exceeded $500B, visibility into CY27' in five quarters.
  2. The Q1 FY26 CFO quote is the bear-case anchor: NVDA management itself sized the China loss at '$50B TAM,' giving the bear case a quantified downside number. The bull thesis only stands if non-China growth continues to compound through CY27.
  3. The Q3 FY26 CEO quote ('half a trillion dollars... $3-4T over five years') is the most concrete forward-revenue framing ever offered by a hardware company. Even haircut 50%, it implies CY27 revenue meaningfully above current FY28 consensus ($489B).
  4. Software / CUDA / Enterprise mentions rose monotonically (25→38) — management is building the post-hardware-cycle narrative explicitly. If Rubin or successor faces yield issues, the software-moat argument is the second-tier valuation defense.
§ 06 — Peer Set · Forward Multiples

vi.Where it Trades

Forward P/E (FY+1) · Sorted high to low

AMD
57.3×
MRVL
46.2×
AVGO
37.5×
Peer Mean (ex-NVDA)
41.8×
TSM
26.2×
NVDA (subject)
27.1×

Faircurve-curated peer set: AMD, AVGO, TSM, MRVL. Selected to triangulate three narrative engines: direct GPU/accelerator competition (AMD), custom AI-ASIC + networking economics (AVGO, MRVL), and the foundry that physically fabricates the basket (TSM). All forward-multiple comparisons use FY+1 consensus revenue, EBITDA, and EPS, pulled live from FMP analyst consensus. Forward anchor years: NVDA FY27 (Jan 2027); AMD FY26 (Dec 2026); AVGO FY26 (Nov 2026); TSM FY26 (Dec 2026, NT$→USD at NT$32.4); MRVL FY27 (Jan 2027). Fwd EV/EBITDA for AMD and MRVL reflects GAAP-basis EBITDA (depressed by acquisition amortization); the cleaner comparable for those names is Fwd P/E.

TickerMkt cap ($B)Fwd P/EFwd EV/EBITDAFwd P/SFwd EBITDA marginFwd rev growth
NVDA$5,47527.1x29.0x14.8x51.1%+71.1%
AMD$69157.3x64.0x*13.9x21.7%*+45.7%
AVGO$2,01337.5x37.1x19.4x53.9%+63.3%
TSM$2,09726.2x15.4x12.9x70.0%+38.0%
MRVL$15546.2x67.5x*14.3x21.4%*+32.7%
Peer avg (ex-NVDA)41.8x46.0x15.1x41.8%+44.9%

* AMD and MRVL Fwd EV/EBITDA elevated by GAAP-basis amortization drag — non-GAAP EBITDA materially higher; treat as not-meaningful comparables.

The counterintuitive headline. NVDA — the highest-growth, highest-margin, highest-share AI silicon name — trades at a 35% discount to peer-average forward P/E. Three explanations are not mutually exclusive: (1) the market discounts FY27 consensus more heavily because $369B revenue embeds aggressive ramp execution; (2) law-of-large-numbers — a 27× multiple at $5.5T mcap implies $200B incremental market cap per turn; (3) AMD/MRVL multiples are partly artifacts of their lower GAAP EBITDA base. The base-case 33× forward P/E sits between current spot and the peer-clean median of 36×.
§ 07 — Valuation

vii.The Frame

Key assumptions — base case inputs and basis

VariableBase valueRationale & basis
FY27 Revenue$369.4BFMP 39-analyst consensus (+71.1% YoY vs FY26 $215.9B). Implies hyperscaler CapEx sustains.
FY27 EBITDA (non-GAAP)$188.9BConsensus 51.1% EBITDA margin; reflects Blackwell-Ultra ramp + Rubin sample contribution.
FY27 EPS$8.31FMP 30-analyst consensus; ~$202B net income on 24.4B diluted shares. +70% YoY.
Forward P/E multiple33.0xAnchor: peer-clean median of AVGO 37.5x + TSM 26.2x = 31.85x; +4% premium for NVDA growth gap.
Forward EV/EBITDA33.0xSame anchor logic; AMD/MRVL excluded from anchor due to GAAP-EBITDA artifact.
DCF WACC12.7%CAPM: through-cycle Beta 1.7 × 5.0% MRP + 4.2% Rf = 12.7% CoE; D/V 0.2% (essentially all-equity).
DCF terminal growth3.0%US long-run GDP 2.0% + 1.0% AI-infrastructure durability premium.
Net debt$0.8BFY26 BS: $11.4B total debt − $62.6B cash & ST investments + $22.3B LT investments = effectively zero.
Diluted shares24,432 MQ4 FY26 weighted-average diluted share count; net of $40B FY26 buyback.

Bull / Bear flex bridge — what changes from consensus base

VariableBearBear: whyBaseBullBull: why
FY27 Revenue$310BHyperscaler CapEx digests 15%; China $0; one-quarter sell-through delay$369B$420BHyperscaler CapEx +60%; sovereign accelerates; modest China rebound
FY27 EBITDA margin47.0%GM compresses on Blackwell Ultra → Rubin transition double-step51.1%53.5%Software/services mix rises; Networking $50B+ FY27 run-rate
FY27 EPS$6.80Lower revenue × lower margin × buyback yield drops on lower FCF$8.31$9.80Higher revenue × margin expansion + accelerated buybacks
Fwd P/E multiple24.0xAI-narrative pause; multiples compress to mega-cap-tech average33.0x38.0xMultiple expansion as CY28 capacity disclosed; full peer-avg credit
Fwd EV/EBITDA22.0xSympathy with semis cycle bear; risk-off rotation33.0x38.0xAVGO-equivalent multiple credit
WACC15.4%Spot Beta 2.24 used; risk premium unchanged12.7%11.0%Through-cycle Beta 1.4; AI durability premium
Terminal growth2.0%GDP-only; no AI-tailwind premium3.0%3.5%Sector durability premium + secular shift to accelerated compute

Blended methodology — 40% Fwd P/E + 40% Fwd EV/EBITDA + 20% DCF

MethodWeightBearBaseBull
Forward P/E (FY27)40%$163.20$274.23$372.40
Forward EV/EBITDA (FY27)40%$131.16$255.11$349.45
DCF (10Y, terminal)20%$90.00$145.00$200.00
Blended fair value100%$135.75$240.74$328.74
§ 08 — The Range · 12-Month View

viii.Three scenarios

BEAR · fair value $136
$136
-39.6% vs spot

Hyperscaler CapEx pauses (–15%), China remains zero, Rubin slips two quarters, and Blackwell Ultra → Rubin transition compresses EBITDA margin to 47%. FY27 revenue $310B, EPS $6.80. Forward P/E compresses to 24x = $163 (40% = $65). Forward EV/EBITDA 22x on $146B EBITDA → $131 (40% = $52). DCF at $90 with spot Beta 2.24 (20% = $18). Implies the kind of multiple reset seen in the 2018 datacenter pause or 2022 crypto unwind. Probability we attach ~25%; trigger is two consecutive hyperscaler capex guide-downs or an EU/India export-control extension.

BASE · fair value $241
$241
+7.0% vs spot

Consensus FY27 (+71% revenue $369B, $8.31 EPS) prints cleanly. 33x forward P/E = $274 (40% = $110) plus 33x forward EV/EBITDA on $189B EBITDA = $255 (40% = $102) plus DCF $145 (20% = $29) blends to $240.74. Implies +6.8% to current $225.32 spot — fairly valued on consensus. Multiple discipline anchors here; through-cycle Beta 1.7 keeps DCF at $145. Sell-side consensus PT $276.75 (+22.8%) sits between our base and bull — the gap is multiple-expansion credit we are not yet willing to extend without FY28 capacity disclosure.

BULL · fair value $329
$329
+46.0% vs spot

Q1 FY27 call discloses explicit FY28 capacity; sovereign + inference reasoning drive FY27 revenue to $420B at 53.5% EBITDA margin. EPS $9.80 at 38x P/E = $372 (40% = $149); 38x EV/EBITDA on $225B EBITDA = $349 (40% = $140); DCF $200 (20% = $40). Probability ~30%; the most plausible single trigger is Rubin first-yield news in 2H CY26 enabling Rubin Ultra timeline pull-forward and an explicit CY27 capacity number on the next call.

HOLD — 12-month price target $240.74
Implies +6.8% upside vs $225.32 spot — inside the HOLD band (BUY threshold >+15%; SELL threshold <−10%). The base case is fairly valued on consensus FY27 numbers; bull skew (+45.9%) is meaningful and asymmetric to bear (−39.8%) because of the +$0.5T+ visibility pipeline and the structurally low net-debt position. Our PT sits below sell-side consensus $276.75 (+22.8%) because we apply growth-discount multiples rather than extrapolated forward expansion. DCF intrinsic of $145 anchors the cash-flow floor: even at the through-cycle Beta 1.7 and 50% FCF margins, the AI premium is already discounted into the share price.
§ 09 — Risks & Catalysts

ix.What could change the call

Principal risks

Hyperscaler CapEx digestion
MED
Top-4 hyperscaler CapEx doubled in 24 months to ~$700B. A 15-25% pullback would compress FY28 consensus revenue 15-20%. Cycle-historical precedent: 2018 datacenter pause, 2022 crypto unwind.
China export controls
HIGH
H20 was zero-ed out in Q1 FY26 ($50B TAM loss, $4.5B writedown). H200 / Blackwell-level export remains constrained; USG revenue-share negotiations unresolved.
Product transition margin risk
MED
Blackwell → Rubin transition in CY27 + Rubin Ultra in CY28 = two compressions in two years. Each historically takes 200-400bps off non-GAAP GM during ramp.
Custom ASIC competition
MED
Google TPUv7, Amazon Trainium2, Meta MTIA, Microsoft Maia gain inference share. AVGO custom AI ASIC FY26 revenue $24B+. NVDA-only architecture risk is the long-term moat question.
Geopolitical re-escalation
HIGH
EU, India, or Japan adoption of US-style export controls would compound the China loss. Taiwan strait tail risk on TSMC supply for all NVDA products.
Beta / volatility
HIGH
Spot Beta 2.24 means 1pp WACC move shifts intrinsic ±15-20%. Any AI-narrative reversal (regulation, alignment incident, viable open-source alternative) is amplified.

Catalysts over the next 12 months

  1. Q1 FY27 print (May 20, 2026): FY28 capacity commentary; Q2 guide vs sell-side $78B implied; gross margin recovery toward 75% non-GAAP.
  2. Rubin first-customer ship (mid-CY26 expected) and yield disclosure — defines the CY27 ramp slope and Rubin-Ultra timeline.
  3. Hyperscaler CapEx guide cycle (Q2/Q3 CY26): MSFT, GOOGL, META, AMZN, ORCL — flat or up sustains the bull case; flat-to-down anchors the bear.
  4. China H200/Blackwell license clarity — each $5B of recovered China revenue is worth $0.15-0.20 of FY27 EPS and a 1-2x multiple expansion.

Faircurve Equities · Independent Single-Name Research · NVDA · Saturday, May 16, 2026

Sources: Financial Modeling Prep (price, financials, peer comp, estimates · 39 analysts FY27 rev, 30 analysts FY27 EPS). Earnings call transcripts, trailing 8 quarters (Q1 FY25 — Q4 FY26). Methodology: forward-multiple peer comparison, consensus-anchored base, 40/40/20 P/E · EV/EBITDA · DCF blend, CAPM WACC cross-check.

This document is research, not advice. Faircurve Equity Research does not make a market in the securities discussed. Forward-looking statements involve risks, uncertainties, and assumptions. Past performance is not indicative of future results.