NewMCP serverUse SEC data in Claude, ChatGPT, Claude Code, Codex, Gemini.Read the guide →

August 27, 2026·15 min read

NVIDIA Corporation

Five year financials and ratios, sector benchmarking, market history and forward scenarios

NVDA, NASDAQ, CIK 1045810, SIC 3674

FY ends last Sunday of Jan, FY2026 ended 25 Jan 2026

Data as of 27 Aug 2026, last filing 10-Q Q2 FY2027

Source: SEC EDGAR via SEC-API.io

NVIDIA is no longer expensive on earnings. It is expensive on the durability of those earnings. At 18.4× forward EPS the stock trades below AMD (43×), Broadcom (23×) and Texas Instruments (27×); the entire debate is whether $279bn of committed supply is a demand signal or an inventory bomb.

The stock has returned 837% over five years and its multiple fell 58% doing it. Revenue went from $26.9bn (FY2022) to $215.9bn (FY2026) and the trailing twelve months stand at $303.0bn. Operating margin of 60.4% and ROIC of 150% are without precedent at this scale. But three things changed in the last four quarters: gross margin peaked, working capital began absorbing cash, and NVIDIA now finances its own customers. Our base case has revenue reaching $636bn by FY2029 (43% CAGR) and EPS of $14.02, implying a $336 target; the bear case is a 64% drawdown. The asymmetry is the position sizing question, not the multiple.

FY2026 revenue

$215.9bn

+65.5% YoY, TTM $303.0bn

Operating margin

60.4%

peer median 22.2%

Free cash flow

$96.7bn

44.8% of revenue

ROIC (NOPAT/IC)

150%

FY2025: 198%

Commitments

$366bn

$279bn of it supply

Top 2 customers

36%

of FY2026 revenue

1Where the business stands Q2 FY2027, reported 26 Aug 2026

Quarterly revenue — Q4 FY2026 is derived as FY2026 full year less the nine months reported in the Q3 FY2026 10-Q; NVIDIA files no Q4 10-Q. Q3 FY2027 is company guidance of $108.0bn ±2% and is not yet split by platform. Sources: 10-K acc. 0001045810-26-000021; 10-Q acc. 0001045810-26-000075; 8-K acc. 0001045810-26-000073.

Three facts that reframe the fiscal year numbers

The fiscal year is already stale. FY2026 closed at $215.9bn. Q1 and Q2 of FY2027 delivered $177.8bn between them, and Q3 is guided to $108.0bn. A single quarter now exceeds the whole of FY2023 ($27.0bn) by 3.6×.
Growth is decelerating in percentage terms and accelerating in dollars. FY2025 added $69.6bn of revenue; FY2026 added $85.4bn; the first half of FY2027 alone added $87.0bn year on year. Percentage deceleration from +114% to +66% is arithmetic, not weakness.
Reported net income is increasingly not operating income. FY2026 booked $9.0bn of other income, of which $8.9bn was gains on unlisted equity stakes. In the first half of FY2027 those gains reached $23.7bn, or roughly one fifth of half year net income. Stripping them out, trailing twelve month EPS falls from $7.92 to $6.87 and the trailing P/E rises from 26.5× to 30.5×.

Definition change to watch. From Q1 FY2027 NVIDIA's non GAAP measures no longer exclude stock based compensation. Non GAAP Q2 EPS of $2.22 is therefore below GAAP EPS of $2.46. That reverses the historical relationship. Do not splice the non GAAP series across FY2027.

Q2 FY2027 vs prior year ($m)
Line	Q2 FY27	Q2 FY26	Δ
Revenue	96,221	46,743	+106%
Data Center	89,023	41,096	+117%
Hyperscale	48,710	24,168	+102%
AI Clouds, Ind. & Ent.	40,313	16,928	+138%
Edge Computing	7,198	5,647	+27%
Gross profit	72,142	33,853	+113%
Gross margin	75.0%	72.4%	+260bp
R&D	7,054	4,291	+64%
SG&A	1,354	1,122	+21%
Operating income	63,734	28,440	+124%
Equity security gains	7,771	2,247	n/m
Effective tax rate	16.5%	15.3%	+120bp
Net income	59,688	26,422	+126%
Diluted EPS ($)	2.46	1.08	+128%
Non GAAP diluted EPS ($)	2.22	1.01	+120%
Free cash flow	21,341	13,450	+59%

Guidance carries a free option. Q3 FY2027 is guided to revenue of $108.0bn ±2%, gross margin of 74.0% ±50bp, GAAP operating expenses of about $9.2bn and a 16% to 18% tax rate. It assumes zero Data Center compute revenue from China, and Hopper shipments to China were already under 1% of Data Center revenue in Q2. China fell from 20% of revenue in FY2024 to 9% in FY2026. Any re opening is upside not in the number; the H200 licence granted in February 2026 has produced no revenue and carries a 25% US import duty NVIDIA cannot pass through.

Revenue by market platform, $m
Platform	Q2 FY27	Q1 FY27	Q2 FY26	Q/Q	Y/Y
Data Center	89,023	75,246	41,096	+18%	+117%
Hyperscale	48,710	43,050	24,168	+13%	+102%
AI Clouds, Industrial & Enterprise	40,313	32,196	16,928	+25%	+138%
Edge Computing	7,198	6,369	5,647	+13%	+27%
Total revenue	96,221	81,615	46,743	+18%	+106%

Revenue by reportable segment, $m
Segment	Q2 FY27	Q1 FY27	Q2 FY26	Q/Q	Y/Y
Compute & Networking	88,299	74,550	41,331	+18%	+114%
Graphics	7,922	7,065	5,412	+12%	+46%
Total revenue	96,221	81,615	46,743	+18%	+106%

Taxonomy and figures as reported in the Q2 FY2027 CFO commentary. Data Center and Edge Computing are the two market platforms; Compute & Networking and Graphics are the two reportable segments. A company was reclassified from AI Clouds, Industrial & Enterprise to Hyperscale during the quarter and the prior period was recast. Periods before FY2027 in this document predate the platform split and carry Data Center and Edge Computing only.

2Twelve months of 8-K filings 13 current reports, 27 Aug 2025 to 26 Aug 2026, and their exhibits

The most useful thing in twelve months of 8-K exhibits is the guidance record. NVIDIA has beaten its own revenue guidance midpoint by 5.6%, 4.8%, 4.6% and 5.7% in the last four quarters, an average of 5.2%, and has met or beaten its gross margin guide every single time. Apply that average to the $108.0bn guided for Q3 FY2027 and the quarter lands near $113.6bn, which would put FY2027 revenue around $411bn against the $406bn in our base case. On this evidence the base case is mildly conservative, and the guidance number should be read as a floor the company expects to clear, not a forecast.

Guidance given versus delivered, from each quarter's CFO commentary
Quarter	Revenue guided	Delivered	Beat	GAAP GM guided	GM delivered	GAAP opex guided	Opex actual	Tax guided	Tax actual
Q3 FY2026	$54.0bn ±2%	$57.0bn	+5.6%	73.3%	73.4%	$5.9bn	$5.84bn	16.5%	15.9%
Q4 FY2026	$65.0bn ±2%	$68.1bn	+4.8%	74.8%	75.0%	$6.7bn	$6.79bn	17.0%	14.8%
Q1 FY2027	$78.0bn ±2%	$81.6bn	+4.6%	74.9%	74.9%	$7.7bn	$7.62bn	17 to 19%	16.6%
Q2 FY2027	$91.0bn ±2%	$96.2bn	+5.7%	74.9%	75.0%	$8.5bn	$8.41bn	16 to 18%	16.5%
Q3 FY2027	$108.0bn ±2%	guided	avg +5.2%	74.0%	guided	$9.2bn	guided	16 to 18%	guided

The effective tax rate has come in below guidance in every quarter shown. From the Q4 FY2026 commentary NVIDIA dropped the other income and expense line from its outlook entirely and moved the tax guide from a quarterly range to a full year range, then cut that full year range from 17 to 19% down to 16 to 18% one quarter later.

Five quarters of operating detail, from the CFO commentary exhibits
$bn unless noted	Q2 FY26	Q3 FY26	Q4 FY26	Q1 FY27	Q2 FY27
Revenue	46.7	57.0	68.1	81.6	96.2
GAAP gross margin	72.4%	73.4%	75.0%	74.9%	75.0%
GAAP opex	5.41	5.84	6.79	7.62	8.41
GAAP diluted EPS ($)	1.08	1.30	1.76	2.39	2.46
Non GAAP diluted EPS ($)	1.05	1.30	1.62	1.87	2.22
Equity security gains	2.2	1.4	5.5	15.9	7.8
Effective tax rate	15.3%	15.9%	14.8%	16.6%	16.5%
Working capital and commitments
Accounts receivable	27.8	33.4	38.5	40.7	63.1
Days sales outstanding	54	53	51	45	60
Inventory	15.0	19.8	21.4	25.8	31.6
Supply commitments	45.8	50.3	95.2	119.0	279.0
Cash
Operating cash flow	15.4	23.8	36.2	50.3	24.1
Free cash flow	13.5	22.1	34.9	48.6	21.3
Returned to shareholders	9.9	12.7	4.0	20.0	26.0
Cash & securities	56.8	60.6	62.6	50.3	56.6

Networking revenue, disclosed under the old taxonomy, ran $7.3bn, $8.2bn and $11.0bn through Q2 to Q4 of FY2026, the Q3 figure being up 162% year on year. As originally reported, Q1 FY2027 split as Hyperscale $37.9bn and AI Clouds, Industrial & Enterprise $37.4bn. The Q2 recast moved roughly $5bn between them, to $43.1bn and $32.2bn.

The 13 current reports and what each carried
Filed	Items	Exhibits and content
26 Aug 26	2.02, 9.01	EX-99.1 release, EX-99.2 CFO commentary. Q2 FY27 results, Q3 guided to $108.0bn
17 Aug 26	1.01, 2.03, 7.01	EX-99.1 SB Energy release. PORTS-Pike Ohio, 4.25 IT gigawatts plus an option on 3.75 more, $1.5bn equity investment
2 Jul 26	5.02	Officer arrangement, no exhibit
30 Jun 26	5.07	Annual meeting voting results
18 Jun 26	8.01, 9.01	EX-1.1 underwriting, EX-4.2 notes, EX-5.1 opinion. $25bn of senior unsecured notes in seven tranches
20 May 26	2.02, 9.01	Q1 FY27 results. New platform taxonomy introduced, dividend raised from $0.01 to $0.25, $80bn added to the buyback authorisation
8 May 26	5.02	Officer arrangement
27 Apr 26	5.02	Officer arrangement
6 Mar 26	5.02, 9.01	EX-10.1 FY2027 Variable Compensation Plan
25 Feb 26	2.02, 9.01	Q4 and full year FY2026 results
23 Jan 26	5.02	Officer arrangement
19 Nov 25	2.02, 9.01	Q3 FY26 results. Networking a record $8.2bn, up 162%
27 Aug 25	2.02, 9.01	Q2 FY26 results. No H20 sales to China, $180m reserve release on $650m sold outside China

The June 2026 notes offering, seven tranches
Maturity	Coupon	Principal $bn	Maturity	Coupon	Principal $bn
2028	4.250%	3.5	2036	4.950%	4.0
2029	4.350%	3.5	2046	5.550%	3.0
2031	4.500%	4.0	2056	5.625%	3.5
2033	4.750%	3.5	Total	4.83%	25.0

Priced 15 June, closed 18 June 2026, led by Goldman Sachs, J.P. Morgan and Morgan Stanley with Citigroup and HSBC. At a 4.83% weighted average coupon the annual interest bill is about $1.2bn, roughly 4.7 times the entire FY2026 interest expense of $259m. Use of proceeds is not stated in the 8-K or the underwriting agreement.

The governance detail that bears on the central question. The FY2027 Variable Compensation Plan, filed as EX-10.1 on 6 March 2026, allocates senior officer bonuses entirely to one metric, fiscal 2027 revenue. There is no margin, cash flow, return on capital or inventory measure in the plan, and the payout runs from zero below threshold to a hard cap of 200% of the base amount at stretch. In a year when the company added $160bn of supply commitments in a single quarter, it is worth knowing that the people approving those commitments are paid on revenue alone.

3Five year financial record FY2022 to FY2026, $ millions

Income statement, from 10-K acc. 0001045810-26-000021 (FY24 to FY26), 0001045810-24-000029 (FY22 to FY23)
$m	FY2022	FY2023	FY2024	FY2025	FY2026	4 yr CAGR
Revenue	26,914	26,974	60,922	130,497	215,938	68.3%
Cost of revenue	9,439	11,618	16,621	32,639	62,475	60.4%
Gross profit	17,475	15,356	44,301	97,858	153,463	72.1%
Research & development	5,268	7,339	8,675	12,914	18,497	36.9%
Sales, general & admin.	2,166	2,440	2,654	3,491	4,579	20.6%
Acquisition termination (Arm)	—	1,353	—	—	—	n/m
Total operating expenses	7,434	11,132	11,329	16,405	23,076	32.7%
Operating income	10,041	4,224	32,972	81,453	130,387	89.8%
Interest income	29	267	866	1,786	2,300	n/m
Interest expense	(236)	(262)	(257)	(247)	(259)	2.4%
Other income, net mostly equity gains	107	(48)	237	1,034	9,022	n/m
Pretax income	9,941	4,181	33,818	84,026	141,450	94.2%
Income tax expense	189	(187)	4,058	11,146	21,383	n/m
Net income	9,752	4,368	29,760	72,880	120,067	87.3%
Diluted EPS ($, split adjusted)	0.385	0.174	1.19	2.94	4.90	88.9%
Diluted shares (m, split adjusted)	25,350	25,070	24,940	24,804	24,514	(0.8)%

Per share data restated for the 4:1 split (Jul 2021) and 10:1 split (Jun 2024). FY2023 reflects the $1.35bn Arm termination charge and a gaming channel inventory correction.

Balance sheet: selected lines
$m	FY22	FY24	FY26	Q2 FY27
Assets
Cash & equivalents	1,990	7,280	10,605	22,443
Marketable securities	19,218	18,704	51,951	34,143
Accounts receivable	4,650	9,999	38,466	63,059
Inventories	2,605	5,282	21,403	31,575
Total current assets	28,829	44,345	125,605	197,412
Property, plant & equipment	2,778	3,914	10,383	14,285
Goodwill	4,349	4,430	20,832	21,125
Unlisted investments	—	—	22,251	51,157
Total assets	44,187	65,728	206,803	320,272
Liabilities & equity
Accounts payable	1,783	2,699	9,812	15,059
Accrued & other current	2,552	6,682	21,352	26,960
Total debt	10,946	9,709	8,468	33,366
Total liabilities	17,575	22,750	49,510	91,288
Shareholders' equity	26,612	42,978	157,293	228,984
Net cash (marketable)	10,262	16,275	54,088	23,220

Cash flow: selected lines
$m	FY22	FY23	FY24	FY25	FY26
Cash from operations	9,108	5,641	28,090	64,089	102,718
D&A	1,174	1,544	1,508	1,864	2,843
SBC	2,004	2,709	3,549	4,737	6,386
Working capital change	(3,363)	(2,207)	(3,722)	(9,383)	(15,949)
Capital expenditure	(976)	(1,833)	(1,069)	(3,236)	(6,042)
Free cash flow	8,132	3,808	27,021	60,853	96,676
Share repurchases	—	(10,039)	(9,533)	(33,706)	(40,086)
Dividends paid	(399)	(398)	(395)	(834)	(974)
Acquisitions & stakes	(263)	(49)	(945)	(2,493)	(32,037)
FCF less capital return	7,733	(6,629)	17,093	26,313	55,616

The cash flow tell. Working capital absorbed $15.9bn in FY2026, split between receivables of +$15.4bn and inventory of +$11.3bn. Operating cash flow fell to 0.86× net income, from 0.94× in FY2024. At Q2 FY2027 receivables of $63.1bn represent 60 days of sales, up from 45 days in a single quarter, and the company attributes the build to "extended payment terms on large multiquarter agreements with certain investment grade customers." Revenue is increasingly being recognised ahead of cash.

4Ratio analysis FY2022 to FY2026, all ratios computed from filed XBRL

Ratio	FY2022	FY2023	FY2024	FY2025	FY2026	Read
Profitability & operating leverage
Revenue growth	n/a	0.2%	125.9%	114.2%	65.5%	Decelerating in %, yet the absolute dollar addition rose every year
Gross margin	64.9%	56.9%	72.7%	75.0%	71.1%	FY26 dip: $4.5bn H20 charge plus Blackwell rack mix; 75.0% back by Q2 FY27
R&D as % of revenue	19.6%	27.2%	14.2%	9.9%	8.6%	Lowest in the peer set, a denominator effect; $18.5bn is the group’s largest budget
Opex as % of revenue	27.6%	41.3%	18.6%	12.6%	10.7%	The clearest evidence of operating leverage in large cap technology
Operating margin	37.3%	15.7%	54.1%	62.4%	60.4%	Peaked FY2025; Q2 FY27 annualises at 66%
EBITDA margin	41.7%	21.4%	56.6%	63.8%	61.7%	D&A is trivial at 1.3% of revenue, the fabless model at work
Net margin	36.2%	16.2%	48.8%	55.8%	55.6%	FY26 flattered by $8.9bn investment gains; core net margin ≈ 52.1%
Effective tax rate	1.9%	(4.5)%	12.0%	13.3%	15.1%	Rising as FDDEI and SBC benefits shrink vs pretax income; guided 16% to 18%
Returns on capital
Return on equity (avg)	36.6%	17.9%	91.5%	119.2%	101.5%	Falling only because retained earnings are compounding faster than income
Return on assets (avg)	22.1%	10.2%	55.7%	82.2%	75.4%	Same dilution effect from the cash and investment pile
ROIC (NOPAT / IC)	60.2%	24.4%	124.9%	198.2%	149.8%	Only $103bn of invested capital produces $130bn of operating profit
Asset turnover	0.61×	0.63×	1.14×	1.47×	1.36×	Slipping as the balance sheet fills with securities and receivables
Liquidity & leverage
Current ratio	6.65×	3.52×	4.17×	4.44×	3.91×	No liquidity question at any point in the period
Quick ratio	6.05×	2.73×	3.67×	3.88×	3.24×	Declining as inventory grows faster than current assets
Total debt / equity	0.41×	0.50×	0.23×	0.11×	0.05×	Rose to 0.15× at Q2 FY27 after the $25bn June 2026 note issuance
Total debt / EBITDA	0.98×	1.90×	0.28×	0.10×	0.06×	Leverage is immaterial even post issuance (~0.13× annualised)
Interest coverage	42.6×	16.1×	128×	330×	503×	Coverage is not a constraint; the new debt is opportunistic, not needed
Liabilities / assets	39.8%	46.3%	34.6%	28.9%	23.9%	Balance sheet is under levered relative to cash generation
Working capital efficiency
Days sales outstanding	63	57	41	46	52	Fell to 45 days by Q1 FY27, then jumped to 60 in one quarter. The line to watch
Days inventory (DIO)	101	122	115	86	92	Deliberate prebuild ahead of Rubin, given lead times beyond 12 months
Days payable outstanding	69	47	43	50	47	NVIDIA pays suppliers faster than customers pay it, unlike Apple
Cash conversion cycle	95	133	113	82	97	Deteriorated 15 days in FY2026; at FY2027 run rate one day of DSO is ~$1.1bn
AR + inventory / revenue	27.0%	33.3%	25.1%	25.4%	27.7%	Back to FY2022 intensity despite 8× the revenue base
Cash flow & capital allocation
Op. cash flow / net income	0.93×	1.29×	0.94×	0.88×	0.86×	Three consecutive years of decline. Earnings quality is drifting, not broken
Free cash flow margin	30.2%	14.1%	44.4%	46.6%	44.8%	Best in class; H1 FY27 ran at 39.4%
Capex as % of revenue	3.6%	6.8%	1.8%	2.5%	2.8%	vs TSMC at 33%. NVIDIA rents the capital intensity of the industry
SBC as % of revenue	7.4%	10.0%	5.8%	3.6%	3.0%	Diluted away by revenue growth; buybacks more than offset issuance
Buybacks + divs / FCF	4.9%	274%	36.7%	56.8%	42.5%	Share count down 3.3% over five years; $80bn added to the authorisation in May 2026

The honest counter argument on ROIC. A 150% return on invested capital is real but partly definitional: NVIDIA's invested capital base is small precisely because TSMC, SK hynix and the ODMs carry the fixed assets. The $279bn of supply and capacity commitments and the $105bn SB Energy guarantee are economically capital like obligations that sit outside invested capital. Treating half of the near term supply commitment as quasi capital would still leave ROIC above 60%. That is extraordinary, but the headline figure overstates the durability of the asset light advantage.

5Sector analysis Latest reported fiscal year, from each company 10-K or 20-F

Sector scatter — Latest completed fiscal year per company. TSMC converted from NT$ at the 20-F convenience rate of NT$31.37/US$. Market capitalisations as of 27 Aug 2026.

Semiconductor peer group: operating and return metrics
Company	FY end	Revenue $m	Growth	GM %	OM %	NM %	R&D %	ROE	ROIC	FCF $m	Days inv.
NVIDIA	Jan 26	215,938	+65.5%	71.1%	60.4%	55.6%	8.6%	101.5%	149.8%	96,676	92
TSMC	Dec 25	121,424	+31.6%	59.9%	50.8%	44.6%	6.5%	35.1%	49.5%	31,959	69
Broadcom	Nov 25	63,887	+23.9%	67.8%	39.9%	36.2%	17.2%	31.0%	19.9%	26,914	36
Intel	Dec 25	52,853	−0.5%	34.8%	−4.2%	−0.5%	26.1%	−0.3%	n/m	(4,949)	126
Qualcomm	Sep 25	44,284	+13.7%	55.4%	27.9%	12.5%	20.4%	23.3%	20.9%	12,820	120
Micron	Aug 25	37,378	+48.9%	39.8%	26.1%	22.8%	10.2%	17.2%	14.8%	1,668	140
AMD	Dec 25	34,639	+34.3%	49.5%	10.7%	12.5%	23.4%	7.2%	6.8%	6,697	142
Super Micro	Jun 25	21,972	+46.6%	11.1%	5.7%	4.8%	2.9%	17.9%	18.9%	1,532	84
Texas Instruments	Dec 25	17,682	+13.0%	57.0%	34.1%	28.3%	11.8%	30.1%	20.7%	2,603	224
Marvell	Jan 26	8,195	+42.1%	51.0%	16.1%	32.6%	25.3%	19.3%	7.2%	1,396	110
Arm Holdings	Mar 26	4,920	+22.8%	97.5%	18.3%	18.4%	56.4%	12.0%	16.4%	979	n/a
Peer median (ex NVDA)		36,009	+27.8%	53.2%	22.2%	20.6%	18.8%	18.6%	18.9%	2,136	120

Marvell's net margin exceeds its operating margin because of ~$1.7bn of non operating gains. Qualcomm's FY2025 net income absorbed a $7.1bn one time tax charge. Intel's equity excludes $12.1bn of non controlling interests.

What the sector comparison establishes

NVIDIA is 1.8× TSMC and 3.4× Broadcom by revenue, and its FY2026 revenue alone equals 27% of the entire 2025 global semiconductor market ($791.7bn, SIA). No company has held that share of a hardware industry in modern history.
Margin is the moat, not scale. A 60.4% operating margin against a peer median of 22.2% means NVIDIA earns roughly $3 of operating profit for every $1 a median peer earns on the same revenue. Only TSMC (50.8%) is in the same conversation, and TSMC spends 33% of revenue on capex to get there against NVIDIA's 2.8%.
R&D intensity is misleading in both directions. At 8.6% of revenue NVIDIA looks underinvested against AMD (23.4%) and Arm (56.4%). In dollars, its $18.5bn budget is the largest in the group, at 2.3× AMD's and 1.3× Intel's. Percentage intensity here measures revenue, not commitment.
The sector is bifurcating, not rising together. Micron (+48.9%) and Marvell (+42.1%) are riding the same AI wave; Intel is shrinking with negative free cash flow; Texas Instruments and Qualcomm are in analog and handset cycles largely disconnected from it. "Semiconductor sector performance" is no longer a single variable.
Memory is where the margin pressure comes from. Micron's 48.9% growth and Gartner's forecast of a further ~47% DRAM price increase in 2026 are NVIDIA's cost line. The Q2 FY2027 CFO said the $279bn commitment jump was "primarily related to the procurement of memory."

Valuation comparison at 27 Aug 2026
Company	Mkt cap $bn	Fwd P/E	EV/Sales	EV/EBITDA
NVIDIA	5,053	18.4×	16.6×	24.9×
TSMC	1,970	19.4×	13.5×	19.0×
Broadcom	1,690	22.5×	23.0×	41.3×
Micron	1,060	6.5×	11.5×	15.2×
AMD	785	43.3×	18.8×	81.2×
Intel	464	53.1×	8.5×	28.8×
Arm	268	105×	51.4×	249×
Texas Instr.	239	27.1×	12.7×	25.9×
Marvell	215	53.4×	24.8×	79.8×
Qualcomm	175	17.7×	4.1×	15.2×

The counterintuitive result. NVIDIA is the second cheapest large cap in its own sector on forward earnings, behind only Micron. Consensus is therefore not pricing hypergrowth. It is pricing a rapid convergence to normal. That is what makes the bear case financially survivable and the bull case, if it happens, violent.

Hyperscaler capex — 2023 to 2025 are SEC filed cash purchases of property & equipment. Microsoft is on a June fiscal year (FY24/FY25/FY26). 2026E is company guidance and press reporting, not filed data.

Revenue by geography — Revenue by customer headquarters location. NVIDIA changed from billing location to headquarters location in Q3 FY2026 and recast prior periods; the two bases are not comparable and should not be spliced.

6Five years in the market Monthly closes, Aug 2021 to Aug 2026

Five year total return — Monthly closes, split adjusted, indexed to 100 at the Aug 2021 close. SMH is the VanEck Semiconductor ETF and SPY the SPDR S&P 500 ETF, used as sector and market proxies. Aug 2026 uses the 26 Aug close of $209.66; the live 27 Aug print of $225.70 is not in the series.

The finding that reframes the valuation debate. NVIDIA returned 837% over five years, twelve times the S&P 500, and the price to earnings multiple fell 58% along the way. Earnings per share rose 1,957% while the price rose 756%. Not one percentage point of the five year return came from the market paying more for a dollar of NVIDIA earnings; the multiple has been a headwind for five straight years. Anyone arguing the stock is a multiple expansion story is arguing against the tape.

Five year market statistics, Aug 2021 to Aug 2026
Measure	NVDA	SMH	SPY
Total return	+837%	+310%	+70%
Annualised return	56.4%	32.6%	11.2%
Annualised volatility	50.4%	34.9%	16.0%
Return per unit of risk	1.04	0.82	0.45
Worst drawdown, monthly close	−62.9%	−40.0%	−24.8%
Best month	+36.3%	+32.2%	+10.5%
Worst month	−32.0%	−17.6%	−9.6%
Positive months	38 of 60	38 of 60	37 of 60
Beta to SPY	2.20	1.53	1.00
Correlation to SPY	0.70	0.83	1.00

Computed from monthly closes. Return per unit of risk uses a 4% cash rate. Worst month for NVDA was April 2022 and the best was May 2023.

Return decomposition by fiscal year
At FY close	Price $	Diluted EPS $	Trailing P/E
FY2022, Jan 2022	24.49	0.385	63.6×
FY2023, Jan 2023	19.54	0.174	112.3×
FY2024, Jan 2024	61.53	1.19	51.7×
FY2025, Jan 2025	120.07	2.94	40.8×
FY2026, Jan 2026	191.13	4.90	39.0×
Trailing, Aug 2026	209.66	7.92	26.5×
Change over the period	+756%	+1,957%	−58%

Where the price is now
Market data	Value	Read
Live price	$225.70	+7.65% on the Q2 FY2027 print
Prior close	$209.66	26 Aug, the anchor used throughout
52 week range	$164.04 to $236.54	95% of the 52 week high
Implied volatility	40.6%	58th percentile of the last 52 weeks
Realised volatility	35.9%	30 day; options carry a 4.7 point premium
Call/put volume	1.63	below the 1.79 average
Daily value	$31.4bn	liquidity is not a constraint at any institutional size
Dividend	$0.25	up 25 fold in June 2026

Mapping the five drawdowns to their drivers

Episode	Peak	Trough	Fall	Recovered	What caused it, and what it teaches
Rate shock and crypto unwind	Nov 21, $34.65	Oct 22, $10.81	−68.8%	May 23	Fed tightening, a gaming channel glut after the crypto bust, and the first China A100 and H100 export ban in Sep 2022. Revenue was flat, not falling, and the stock still lost two thirds.
Post ChatGPT consolidation	Aug 23, $50.27	Oct 23, $39.23	−22.0%	Nov 23	Expanded controls removed the A800 and H800 workarounds, plus profit taking after the May 2023 guidance shock. Revenue grew 206% that quarter.
Summer air pocket	Jun 24, $140.76	Aug 24, $90.79	−35.5%	Oct 24	Reports of a Blackwell mask respin met the unwind of the yen carry trade. A pure positioning event with no change to the earnings path.
DeepSeek and tariffs	Jan 25, $153.13	Apr 25, $86.62	−43.4%	Jul 25	DeepSeek R1 questioned the compute intensity of frontier training, then the H20 licence ban forced a $4.5bn charge and reciprocal tariffs hit. The only episode with a genuine fundamental cause, and it cost 43%.
2026 digestion wobble	May 26, $236.54	Jun 26, $189.86	−19.7%	pending	Custom ASIC share fears, the $25bn debt issue and memory cost inflation. Not yet resolved; the Q2 print has taken the price back to 95% of the high.

Peaks and troughs are intramonth highs and lows, so these are deeper than the monthly close drawdowns plotted above. Recovery is the first month whose close exceeded the prior peak close.

The market bull case

The de rating is the bull case. At 26.5× trailing earnings the stock is cheaper than at any point in the five years charted, and cheaper than it was at $24 in January 2022 when it traded on 63.6×. Every beat lowers the multiple mechanically. If the multiple merely holds at 26.5× while base case FY2029 EPS of $14.02 arrives, the price is $372, above our $336 base target, which already assumes further compression to 24×.
Positioning is not stretched. Call to put volume of 1.63 today is below the 1.79 average, and implied volatility of 40.6% sits at only the 58th percentile of the last year despite an earnings print that moved the stock 7.7%. There is no euphoria signature in the option tape.
Options price the base case as reachable, not heroic. The one year implied one standard deviation band is $150 to $339. The $336 base target sits inside it. On a three year horizon the base case is a +0.6 sigma outcome, which is to say ordinary.

The market bear case

Five drawdowns of 20% or more in five years, one of 69%. At 50% annualised volatility a 30% fall is roughly a one sigma event. Owning this stock means accepting a 30% drawdown as normal weather, not as a thesis break.
Every drawdown happened while revenue was rising. In 2022 revenue was flat and the stock lost two thirds. In 2024 the earnings path never changed and the stock lost 35%. Fundamental strength has never once prevented a drawdown, which is why the fundamental case and the market case have to be held separately.
The bull case is a market capitalisation problem. At $225.70 the company is worth $5.4tn. The base target implies $8.1tn and the bull target $16.6tn. The bull case asks NVIDIA alone to create roughly $11tn of value in three years, more than the entire 2025 market capitalisation of the next four semiconductor companies combined.
Beta of 2.20 means this is not a diversifier. Correlation to the S&P 500 is 0.70 and to the semiconductor sector 0.77. In a market drawdown NVIDIA does not protect a portfolio, it amplifies the loss by roughly two times.

Reconciling the scenarios with the option market. Measured against implied volatility of 40.6%, our three year targets are a −1.6 sigma bear, a +0.6 sigma base and a +1.6 sigma bull. That is a near symmetric spread and it means the scenario set is neither timid nor sensational by the market's own yardstick. Note what it also implies: on a one year view the bear case at $76 is a −2.7 sigma event, so it is not a 12 month forecast. The bear case is a multi year unwind, not a crash.

What the price history says about the scenario probabilities. Assigning 25% to the bear case may look harsh next to a 56% annualised five year return. It is not. Over the five years charted, NVDA spent 15 of 61 months more than 20% below its prior peak. That is 25% of the time, so a 25% weight on a bad outcome is precisely the historical base rate of this stock sitting in a deep drawdown.

7Critical factors driving the bull case Ranked by contribution to variance in the FY2029 outcome

#	Factor	Reading	Insight: why it drives the best case, and what would break it	Leverage
1	AI infrastructure capex	~$725bn 2026E big four	The big four hyperscalers spent ~$409bn in 2025 (SEC actuals) and plan ~$725bn in 2026, a 77% increase. Neoclouds, sovereign programmes and enterprise sit on top. The insight practitioners miss: NVIDIA's revenue is not a share of this number, it is a share of the compute portion of it. Roughly 55% to 60% of an AI data centre's cost is the accelerator system; the rest is land, power, shell and cooling. That is precisely why NVIDIA is now guaranteeing power projects. Every dollar diverted to megawatts is a dollar not spent on GPUs, and grid interconnect is now the binding constraint, not silicon. Breaks if: any hyperscaler cuts capex guidance, or depreciation schedules on 2024 and 2025 GPU fleets shorten.	Very high
2	Supply commitments as forward demand	$279bn vs $119bn one qtr ago	Commitments rose $160bn in a single quarter, described as "primarily related to the procurement of memory." Management says it has secured components "for the next several years." The insight: this is the closest thing NVIDIA discloses to a backlog, since the company reports none. But it is a cost commitment, not a revenue contract. It is bullish only if it was placed against customer orders. At a 71% gross margin, $279bn of committed COGS implies roughly $960bn of associated revenue if fully consumed. That is the bull case stated as arithmetic. Breaks if: demand softens. NVIDIA states it "may not be able to reduce our supply commitments in time, at the same rate, or at all." The FY2026 H20 episode ($4.5bn charge) is the working example. The supply line is only part of it: total future commitments reach $366bn, being $279bn of supply and capacity, $29bn of cloud services, $25bn of data centre leases not yet commenced, $25bn of equity investments and $8bn of capex.	Very high
3	Rubin cycle & system content	Ramp Q4 CY2026	Vera Rubin enters production shipment in Q3 CY2026 with volume in Q4 into H1 2027. Each generation has raised revenue per rack: the shift from HGX boards to full NVL72 racks multiplied dollar content per deployment. The insight: content expansion is dilutive to margin even while it lifts revenue. NVIDIA's own FY2026 10-K attributes the gross margin decline partly to "our business model transitioned from offering Hopper HGX systems to Blackwell full scale datacenter solutions." A bull case built on system content therefore cannot also assume 75% gross margins indefinitely. Breaks if: Rubin slips a quarter, or customers hold for Rubin and pause Blackwell orders (the Osborne effect).	High
4	Networking attach rate	$31.4bn FY2026 +142%	Networking grew from $8.6bn (FY2024) to $31.4bn (FY2026) and is the least discussed part of the story. The insight: networking travels with rack scale deployment and is defensible in a way that raw compute is not. A customer can substitute a TPU for a GPU far more easily than it can rip out NVLink and InfiniBand from a 9,216 accelerator pod. Networking is the switching cost that converts a chip sale into a platform. Breaks if: Ethernet based open standards (UEC, Broadcom Tomahawk) take share in scale out fabric.	Med. high
5	China re entry optionality	Zero	China fell from 20% of revenue (FY2024) to 9% (FY2026). The Q3 FY2027 outlook assumes no Data Center compute revenue from China. The February 2026 H200 licence permits sales under a 25% import duty NVIDIA cannot pass through and has generated no revenue. The August 2025 H20 licences produced just $60m. The insight: this is a genuine free option. It costs nothing in the base case and adds directly to it if exercised. But it is a two sided option: Beijing is steering state buyers toward Huawei Ascend, so the market may be lost regardless of US policy. Value it as an option, not a forecast.	Medium
6	Memory cost inflation	DRAM +47% 2026E	SK hynix is sold out of HBM through 2026; DDR5 contract prices have more than doubled; Gartner forecasts a further ~47% DRAM increase in 2026. NVIDIA's filings state "our product and solution pricing generally does not fluctuate with short term changes in our costs." The insight: this is the most least modelled risk in the sector, and it is the reverse of the usual semiconductor story, because the constraint is on the input side rather than the demand side. Each 100bp of gross margin is worth roughly $0.22 of FY2029 EPS in our base case. A 400bp memory driven compression removes ~$0.90 of EPS and, at 24× exit, roughly $22 of value per share, before any second order effect on the multiple itself. Bull case requires: that the multiyear SK hynix partnership announced in August 2026 locks in pricing, not just volume.	Negative
7	Custom ASIC displacement	~27.8% of units 2026E	Google TPU v7, Amazon Trainium3, Meta MTIA 400 and Microsoft Maia 200 are all in volume. Custom silicon is projected at 27.8% of AI chip units in 2026, growing 44.6% versus 16.1% for merchant GPUs. Broadcom reports a $73bn AI backlog. The insight: unit share is the wrong metric. ASICs win inference workloads the buyer already understands and can amortise, while frontier training remains CUDA bound. The risk to NVIDIA is not displacement, it is mix: as inference grows faster than training, the addressable share of the pool shrinks even as the pool expands. NVIDIA's estimated accelerator share has already fallen from ~87% (2024) to ~75% (2026E).	Negative
8	Customer concentration & vendor financing	36% from 2 customers	One direct customer was 22% of FY2026 revenue and a second 14%; three receivable counterparties represent 56% of the balance. NVIDIA added a new risk factor in FY2026 on counterparty exposure, disclosing $99bn of equity investments, $25bn of investment commitments, $36bn of cloud services commitments to partners who simultaneously buy its products, and a guarantee capped at $105bn on SB Energy data centres leased to OpenAI. The insight: this is the structural weakness of the bull case. Revenue growth increasingly depends on customers whose ability to pay depends on capital NVIDIA is helping to arrange. The 10-Q says plainly that AI clouds "currently lack the ability to secure long term infrastructure contracts and investment grade financing capacity." The SB Energy guarantee shows the scale: 4.25 gigawatts in Ohio on 20 year leases to OpenAI, which NVIDIA states represents about 1.5 million of its GPUs, or $150bn to $200bn of its own revenue, per infrastructure generation. Circularity of this kind is invisible in the income statement until it is not.	Negative

8Three year scenario analysis FY2027E to FY2029E, base year FY2026 actual

FY2027 is close to locked. Half of it is reported ($177.8bn) and the third quarter is guided ($108.0bn), a figure the company has beaten by an average of 5.2% in each of the last four quarters. Only the fourth quarter is a genuine variable, and lead times beyond 12 months mean the product mix shipping in Q4 FY2027 has largely already been committed. The scenarios diverge almost entirely in FY2028 and FY2029, where the analytical work belongs.

Scenario financials ($m except per share)
	BULL (20% probability)			BASE (55% probability)			BEAR (25% probability)
$m	FY27E	FY28E	FY29E	FY27E	FY28E	FY29E	FY27E	FY28E	FY29E
Revenue	411,837	658,939	889,568	405,837	543,821	636,271	395,837	411,670	321,103
Growth %	+90.7	+60.0	+35.0	+87.9	+34.0	+17.0	+83.3	+4.0	−22.0
Gross margin %	74.6	74.0	73.5	74.4	72.5	71.0	74.0	66.0	59.0
Operating expenses	35,800	47,972	61,404	35,500	46,150	57,226	35,200	42,240	44,352
Operating income	271,431	439,643	592,435	266,442	348,120	394,527	257,720	229,462	145,099
Operating margin %	65.9	66.7	66.6	65.7	64.0	62.0	65.1	55.7	45.2
Tax rate %	16.5	17.0	17.0	16.5	17.5	18.0	16.5	18.0	19.0
Net income	228,502	367,062	494,207	224,338	289,344	325,974	217,047	190,291	119,990
Diluted shares (m)	24,150	23,600	23,000	24,150	23,700	23,250	24,150	23,900	23,700
Diluted EPS ($)	9.46	15.55	21.49	9.29	12.21	14.02	8.99	7.96	5.06
Free cash flow	159,268	275,874	405,330	157,462	240,988	294,449	154,058	189,823	163,743
FCF margin %	38.7	41.9	45.6	38.8	44.3	46.3	38.9	46.1	51.0
Valuation implication
FY2026→FY2029 revenue CAGR	60.3%			43.4%			14.1%
Exit P/E applied to FY2029E EPS	32.0×			24.0×			15.0×
Implied FY2029 value per share	$688			$336			$76
vs $209.66 close, 26 Aug 2026	+228% (+48.6% p.a.)			+60% (+17.1% p.a.)			−64% (−28.7% p.a.)

Probability weighted outcome

At 20/55/25 the expected FY2029 value is $342 per share, a 63% gain over three years or 17.7% annualised. That is a respectable but unremarkable expected return for an asset whose 25th percentile outcome is a 64% loss. The distribution is skewed left in probability and skewed right in magnitude, the reverse of what most investors assume they are buying.

Read the base case as a bear case in disguise. Base assumes revenue almost triples from FY2026 to FY2029 and NVIDIA still only returns 17.1% annually, because the exit multiple compresses from 45× trailing to 24×. Nearly all of the base case return comes from earnings growth being partially given back through de rating. Investors need the bull case to be paid for the bear case risk.

Core assumptions behind each scenario
Assumption	Bull	Base	Bear
FY2027 Q4 revenue	$126bn	$120bn	$110bn
Q3 FY2027 versus the $108bn guide	+5% beat	in line	in line
Big four capex 2027	>$1.0tn	~$850bn	~$700bn
NVDA share	holds ~75%	falls to ~65%	falls to ~50%
Rubin ramp	on time, +ASP	on time, flat ASP	one quarter slip
Memory cost trend	locked via SK hynix	+15% pass through gap	+40%, no pass through
China Data Center	partial re entry	zero	zero + Huawei share gain
Inventory writedowns	under 1% of revenue	1% to 2% of revenue	5%+ on excess commitments
Opex growth	+34% / +28%	+30% / +24%	+20% / +5%
WC drag	36% of incremental revenue in all cases
Capex	3.0% of revenue in all cases
Investment gains	excluded, treated as non recurring

9Sensitivity analysis What moves the answer

EPS sensitivity heat map — FY2029E diluted EPS. Revenue CAGR measured from the FY2027 base of $405.8bn; operating expenses held at the base case FY2029 level of $57.2bn; tax 18%; 23,250m diluted shares. Base case ($14.02) sits at approximately +17% CAGR and 71% gross margin.

Value per share ($); anchor price $209.66
P/E	$3.50	$5.00	$6.50	$8.00	$9.50	$11.00	$12.50	$14.00	$17.50	$21.50
14×	49	70	91	112	133	154	175	196	245	301
18×	63	90	117	144	171	198	225	252	315	387
22×	77	110	143	176	209	242	275	308	385	473
26×	91	130	169	208	247	286	325	364	455	559
30×	105	150	195	240	285	330	375	420	525	645
35×	122	175	228	280	332	385	438	490	613	753

The three scenario outcomes are the cells at 15×/$5.06 (bear, $76), 24×/$14.02 (base, $336) and 32×/$21.49 (bull, $688). Note that the anchor price of $209.66 is recoverable from a wide range of the grid, roughly $6.50 EPS at 32× or $14.00 at 15×. That is exactly why the market and its critics can both defend their positions from the same numbers.

Ranking of the sensitivities

Driver	Range tested	FY28E EPS swing	%
Revenue	±10%	$2.74	±11.2%
Gross margin	±300bp	$1.14	±4.7%
Tax rate	15% to 21%	$0.89	±3.6%
Share count	±3%	$0.74	±3.0%
Operating expenses	±15%	$0.48	±2.0%

Why the ranking matters more than the ranges. Revenue is 2.4× more powerful than gross margin and 5.7× more powerful than opex. At a 64% operating margin, cost control is nearly irrelevant to the outcome. NVIDIA cannot manage its way out of a demand shortfall, and does not need to manage costs to deliver an upside. Every hour of analytical effort belongs on the demand side: hyperscaler capex guidance, custom silicon share, and whether the $279bn commitment was placed against orders. Modelling opex to the nearest hundred million is wasted work.

Non linearity the grid understates. The two variables are not independent in the bear case. A demand shortfall triggers inventory and purchase obligation writedowns, so revenue falling 20% does not hold gross margin at 71%. It drives it toward 59%, as modelled. The FY2026 H20 charge showed the mechanism: a 2.6 point gross margin hit from a single excess commitment event.

10What would change the view Monitorable indicators, in order of signal value

Leading indicators: bearish
Indicator	Current	Trigger level
Supply commitments q/q	+$160bn	Any decline
Days sales outstanding	60, was 45	>75 days
Inventory + commitments / fwd revenue	0.77×	>1.0×
Gross margin guidance	74.0%	<70%
Inventory provisions as % of revenue	1.0%	>3%
Big four capex guidance revision	Rising	Any cut
Top customer concentration	22% / 14%	>25% single
Vendor financing commitments	$105bn cap	Drawn, or expanded

Leading indicators: bullish
Indicator	Current	Trigger level
China Data Center revenue	<1%, $0 guided	Any material amount
Networking revenue growth	+142% FY26	Sustained >80%
Rubin volume ramp timing	Q4 CY2026	On or ahead
CoWoS packaging capacity	120k to 140k wpm	Gap closes below 5%
Gross margin recovery	75.0% Q2	Sustained >75%

The three risk factors that matter, from Item 1A

"One direct customer accounted for 22% and another for 14% of total revenue." Three receivable counterparties represent 25%, 18% and 13% of the balance. FY2026 10-K, Item 1A

"We may not be able to reduce our supply commitments in time, at the same rate, or at all." That is against $279bn of commitments and lead times "of more than 12 months." FY2026 10-K Item 1A; Q2 FY2027 10-Q

New in FY2026: risks from "long term capacity purchase obligations and financial guarantees" and requests to "offer financing arrangements to support our customers' and partners' buildout of datacenter infrastructure." FY2026 10-K, Item 1A. The only new risk factor heading versus FY2025

The steelman for the bears

The strongest bear argument is not that AI demand is fake. It is that NVIDIA has begun underwriting its own demand. The company holds $99bn of equity investments, has committed $25bn more, guarantees up to $105bn of data centre obligations for a customer's landlord, and has signed $36bn of cloud services purchases with partners who simultaneously buy its products. None of this is improper and all of it is disclosed. But it means a portion of reported revenue is backed by capital NVIDIA itself supplied. In a demand downturn those exposures correlate perfectly with the revenue decline, which is precisely when diversification is needed. That is the mechanism by which a 22% revenue decline becomes a 64% equity drawdown in the bear case.

The steelman for the bulls

The strongest bull argument is not the growth rate. It is that at 18.4× forward earnings, consensus already assumes the deceleration. NVIDIA trades below Broadcom, Texas Instruments, AMD and Arm on forward earnings while growing faster than all of them and earning three times the peer median operating margin. If FY2028 revenue merely grows 34%, half the FY2026 rate, the stock is on 18× earnings two years out. The bull case does not require the boom to continue; it requires it not to break.

Where reasonable people disagree, and why. Bulls and bears here are not disputing the historical numbers, which are filed and audited. They are disputing one unobservable: whether the $279bn supply commitment was placed against customer purchase orders or against management's forecast. NVIDIA discloses no backlog, so the question cannot be settled from the filings. Anyone claiming certainty in either direction is asserting something the disclosure does not support. The honest position is that this is a genuinely two sided distribution, and the correct response to a two sided distribution is position sizing, not conviction.

Sources. All historical financial data extracted from SEC EDGAR filings via SEC-API.io: NVIDIA 10-K accession 0001045810-26-000021 (FY2026), 0001045810-25-000023 (FY2025), 0001045810-24-000029 (FY2024), 0001045810-22-000036 (FY2022); 10-Q accession 0001045810-26-000075 (Q2 FY2027), 0001045810-26-000052 (Q1 FY2027), 0001045810-25-000230 (Q3 FY2026); 8-K accession 0001045810-26-000073 (Q2 FY2027 results and Q3 guidance, filed 26 Aug 2026). Peer data from each company's most recent 10-K or 20-F. Hyperscaler capital expenditure from Microsoft 10-K 0001193125-26-323660, Alphabet 10-K 0001652044-26-000018, Amazon 10-K 0001018724-26-000004 and 10-Q 0001018724-26-000026, Meta 10-K 0001628280-26-003942. Industry revenue from the Semiconductor Industry Association (6 Feb 2026) and WSTS (2 Dec 2025). Market capitalisations, forward multiples and 2026 capex plans are third party estimates, not filed data, and are identified as such throughout.

Method and limitations. Ratios are computed from filed XBRL; balance sheet ratios use two point averages where a prior year is available. Per share figures are restated for the 4:1 (July 2021) and 10:1 (June 2024) splits. Q4 FY2026 figures are derived as the full year less the nine months reported in the Q3 FY2026 10-Q, because NVIDIA files no fourth quarter 10-Q; these are arithmetic differences of filed figures, not directly reported. Revenue by geography changed basis in Q3 FY2026 from billing location to customer headquarters location; the two series are not comparable. Non GAAP measures changed definition in Q1 FY2027 to include stock based compensation. Market statistics are computed from monthly closing bars and are therefore lower than the intramonth figures quoted in the drawdown table. Valuation throughout is anchored on the 26 August 2026 close of $209.66 rather than the live intraday price, so that it stays consistent with the peer multiples gathered the same day. Forward scenarios are the author's estimates and are not forecasts; they exclude gains on unlisted equity securities, which have been material to recent reported net income. Third party market share, capex guidance and forward multiple figures could not be verified against primary sources and should be treated as indicative.

This document is financial analysis prepared from public filings for informational purposes. It is not investment advice, and the author is not a registered investment adviser. Scenario outputs are illustrative model results, not price targets or recommendations. Prepared 27 August 2026. Data current to the Q2 FY2027 10-Q filed 26 August 2026. Powered by SEC-API.io

NVIDIA Corporation, Financial and Scenario Analysis, 27 August 2026Data: SEC EDGAR via SEC-API.io MCP Server