NVIDIA Corporation (NVDA) — Company Research

Last Updated: 4 September 2026

NVIDIA is the world's largest company by market capitalisation, at roughly $5.5 trillion, and it reached that position by supplying the accelerated-computing platform on which almost all frontier AI is trained and served. In the quarter ended 26 July 2026 the company reported revenue of $96.22bn, up 106% year on year, with Data Center alone contributing $89.02bn. What has changed most in 2026 is not the growth rate but the structure of the business around it: NVIDIA has issued $25bn of bonds, committed $366bn of future purchase and lease obligations, agreed to acquire Hugging Face for $12.93bn, and taken on a $105bn residual value guarantee against OpenAI's data-centre leases in Ohio. This report sets out what the filings say, without price targets or analyst ratings. The most recent reported period throughout is fiscal Q2 2027, reported on 26 August 2026.

1. Company Snapshot

Field	Value
Ticker / exchange	NVDA — Nasdaq Global Select Market
Headquarters	2788 San Tomas Expressway, Santa Clara, California, USA (Delaware incorporated)
Founded	1993
CEO / Leadership	Jen-Hsun (Jensen) Huang, founder, President and Chief Executive Officer. Colette Kress, EVP and Chief Financial Officer. Nicholas Parker became EVP, Worldwide Field Operations on 24 Aug 2026.
Employees	Approximately 42,000 in 38 countries at the end of FY2026, of which about 31,000 are in research and development (FY2026 Form 10-K)
Revenue (FY2026, year ended 25 Jan 2026)	$215,938m
Net income (FY2026, GAAP)	$120,067m
Most recent reported quarter	Fiscal Q2 2027, ended 26 Jul 2026, reported 26 Aug 2026: revenue $96,221m, GAAP diluted EPS $2.46
Market capitalisation	Approximately $5.52 trillion (3 Sep 2026 close of $228.45 on 24.147bn shares)
Fiscal year end	Last Sunday in January. FY2027 ends in late January 2027.
Dividend	$0.25 per share per quarter, raised from $0.01 with effect from the 26 Jun 2026 payment

2. Bull and Bear Case

Bull Case

• Growth has not decelerated at scale: revenue of $96.22bn in fiscal Q2 2027 was up 18% sequentially and 106% year on year, and guidance for fiscal Q3 2027 is $108.0bn plus or minus 2% — another 12% sequential step, explicitly assuming zero Data Center compute revenue from China.
• Margin structure is holding through the ramp: GAAP gross margin was 75.0% in the quarter, up 2.6 points year on year, with GAAP operating income of $63.73bn on operating expenses of just $8.41bn. Guidance for Q3 is 74.0% gross margin plus or minus 50 basis points.
• Vera Rubin is in full production, on schedule: the platform is running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius. The NVL72 configuration is claimed at up to 30 times more work per watt for agentic AI than the prior generation, which matters when power, not silicon, is increasingly the constraint.
• Cash generation funds everything without strain: trailing twelve-month operating cash flow was $134.36bn against capital expenditure of $7.35bn. The company returned roughly $26.0bn to shareholders in the quarter and still held $99.0bn of remaining buyback authorisation at quarter end.
• The software and networking moat is widening, not just the GPU: CUDA remains the default development target, and NVIDIA now sells the full rack — Vera CPU, NVLink 6, ConnectX-9, BlueField-4 and Spectrum-6 — rather than a chip. The pending Hugging Face acquisition extends that into the model distribution layer.

Bear Case

• Commitments have outrun visibility: total future commitments reached $366bn at 26 Jul 2026, of which supply and capacity commitments alone jumped from $119bn to $279bn in a single quarter, with roughly $120bn falling due in the remainder of FY2027. Inventories rose to $31.58bn from $21.40bn at year end. If the AI capital-expenditure cycle pauses, these are contractual, not discretionary.
• Circular financing is now on the balance sheet: NVIDIA is simultaneously supplier, investor and credit backstop to its own customers — a $105bn cap on residual value guaranties for OpenAI's Ohio leases, $3,529m notional of land, power and shell guarantees for AI cloud partners against just $712m of partner escrow, $25bn of committed equity investments, and non-marketable securities that more than doubled to $51.16bn.
• Earnings quality has deteriorated: GAAP other income of $7,773m in the quarter was almost entirely $7,771m of gains on equity securities. That is why non-GAAP EPS of $2.22 sits below GAAP EPS of $2.46. Marketable equity securities rose from $12.89bn to $42.78bn in six months, so reported earnings now move with the share prices of NVIDIA's own investees.
• China is effectively closed, in the company's own words: NVIDIA states it was "effectively foreclosed" from China's data-centre compute market, took a $4.5bn H20 charge in FY2026 and a further $0.4bn H200 charge in the first half of FY2027, and shipped less than 1% of Data Center revenue there. H200 units must be inspected in the United States and carry a 25% import tariff NVIDIA cannot pass on.
• Custom silicon is growing faster than merchant GPUs: TrendForce puts custom-ASIC shipment growth at 44.6% in 2026 against 16.1% for merchant GPUs. The programmes behind that — Google TPU, Amazon Trainium, Microsoft and Meta, designed with Broadcom and Marvell — sit inside the same hyperscalers that represent NVIDIA's 13–16% individual revenue concentrations.

3. Business Segments

NVIDIA reports two segments, Compute and Networking and Graphics, and separately discloses revenue by market platform. That market-platform disclosure was restructured from fiscal Q1 2027: the former Data Center, Gaming, Professional Visualization, Automotive and OEM breakdown was replaced by Data Center, split into Hyperscale and AI Clouds, Industrial and Enterprise, plus Edge Computing. Prior periods were recast. Figures below are for fiscal Q2 2027.

Segment	% of revenue	What it is
Data Center — Hyperscale	50.6% ($48,710m)	GPUs, CPUs, networking and full racks sold to the largest cloud service providers. Up 102% year on year.
Data Center — AI Clouds, Industrial and Enterprise (ACIE)	41.9% ($40,313m)	Specialist AI cloud operators, sovereign AI programmes, enterprise and industrial deployments. Up 138% year on year.
Edge Computing	7.5% ($7,198m)	PCs, game consoles, workstations, AI-RAN base stations, robotics and automotive. Up 27% year on year.
Reportable segment: Compute and Networking	91.8% ($88,299m)	The accounting segment covering data-centre compute, networking and automotive platforms. Segment operating income $62,696m.
Reportable segment: Graphics	8.2% ($7,922m)	GeForce, professional visualisation, virtual GPU software and cloud gaming. Segment operating income $3,899m.

Geographically in the same quarter, revenue by customer headquarters was United States $60,074m, Taiwan $26,985m, China including Hong Kong $7,880m and other regions $1,282m. One direct customer accounted for 16% of total revenue in the quarter; across the first half of FY2027 three direct customers accounted for 16%, 15% and 13%.

4. Business Model and Moat

How it makes money. NVIDIA designs accelerated-computing systems and sells them, increasingly as complete racks rather than discrete chips, to cloud providers, AI laboratories, enterprises and governments. It manufactures nothing itself — TSMC fabricates the silicon and partners assemble the systems — so the model converts design and software leadership directly into gross margin. At 75.0% GAAP gross margin and only $8.41bn of quarterly operating expenses against $96.22bn of revenue, incremental revenue drops through at extraordinary rates.

Where the defensibility sits. The durable asset is CUDA and the software stack built on it. Two decades of libraries, frameworks and developer habit mean that a competing accelerator must be not merely cheaper but sufficiently better to justify rewriting a workload. NVIDIA has extended that from the chip to the rack: NVLink, Spectrum-X Ethernet, BlueField DPUs and the reference system designs make the whole cluster the product, which is far harder to substitute piecemeal than a GPU.

What has changed in the model this year. NVIDIA has moved upstream and downstream simultaneously. Upstream it has locked in supply, taking capacity commitments from $119bn to $279bn in one quarter and signing a multi-year memory partnership with SK hynix. Downstream it is underwriting demand — guaranteeing residual values on OpenAI's leases, investing equity in AI model makers and infrastructure financiers, and convening more than $500bn of third-party financing platforms with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR. The pending $12.93bn Hugging Face acquisition adds the model distribution layer.

Where the model is weakest. NVIDIA's customers are also its most credible competitors. The hyperscalers that generate its concentrated revenue are the same organisations funding custom accelerator programmes, and they have both the volume and the engineering depth to displace merchant silicon on their own internal workloads. Meanwhile the supply chain has no redundancy: essentially all leading-edge output comes from a single foundry, and advanced packaging and high-bandwidth memory are both hard constraints.

5. Financial Health

All figures below are taken from NVIDIA's quarterly earnings press releases, Forms 10-Q and 10-K, and the SEC XBRL company facts dataset. Fiscal years end on the last Sunday in January and are labelled by the calendar year in which they end.

Split basis. NVIDIA executed a four-for-one stock split effective 20 Jul 2021 and a ten-for-one split effective 10 Jun 2024. The per-share figures below are stated on the fully adjusted post-2024 basis. The FY2022 row required manual restatement: the latest SEC-tagged value for that year, $3.85, was filed in February 2024 before the ten-for-one split and has never been restated in a subsequent Form 10-K.

Fiscal Year	Revenue ($m)	YoY %	GAAP EPS	Adjusted EPS	Dividend/share	Long-term debt (YE)
FY2022 (ended 30 Jan 2022)	$26,914m	+61.4%	$0.385	$0.444	$0.016	$10,946m
FY2023 (ended 29 Jan 2023)	$26,974m	+0.2%	$0.17	$0.334	$0.016	$9,703m
FY2024 (ended 28 Jan 2024)	$60,922m	+125.9%	$1.19	$1.296	$0.016	$8,459m
FY2025 (ended 26 Jan 2025)	$130,497m	+114.2%	$2.94	$2.99	$0.034	$8,463m
FY2026 (ended 25 Jan 2026)	$215,938m	+65.5%	$4.90	$4.77	$0.040	$7,469m

Long-term debt above is the non-current balance as tagged in each Form 10-K. It rose sharply after the year end: NVIDIA issued $25.0bn of senior notes across seven tranches maturing 2028 to 2056 in June 2026, taking non-current long-term debt to $32,366m at 26 Jul 2026, with $1,000m of current debt representing the 3.20% notes due 2026. FY2026 gross margin of 71.1% was depressed by the $4.5bn H20 China inventory write-down taken in the first quarter of that year.

Quarter / Half	Revenue	Adjusted EPS	GAAP EPS
Q2 FY2027 (ended 26 Jul 2026)	$96,221m	$2.22†	$2.46
Q1 FY2027 (ended 26 Apr 2026)	$81,615m	$1.87†	$2.39
Q4 FY2026 (ended 25 Jan 2026)	$68,127m	$1.62	$1.76
Q3 FY2026 (ended 26 Oct 2025)	$57,006m	$1.30	$1.30
FY2026 full year (ended 25 Jan 2026)	$215,938m	$4.77	$4.90

† From fiscal Q1 2027 NVIDIA's non-GAAP measures no longer exclude stock-based compensation. The FY2027 adjusted figures are therefore calculated on a stricter basis than the FY2026 rows and the two are not directly comparable. On the restated basis including stock-based compensation, Q4 FY2026 adjusted EPS would have been $1.59.

Cash flow and balance sheet, taken from the fiscal Q2 2027 earnings release and Form 10-Q. For the six months to 26 Jul 2026: operating cash flow $74,421m, purchases related to property, equipment and intangible assets $4,434m, depreciation and amortisation $2,124m, share repurchases $39,044m and dividends paid $6,290m. For FY2026: operating cash flow $102,718m, capital expenditure $6,042m, depreciation and amortisation $2,843m. At 26 Jul 2026 NVIDIA held cash and equivalents of $22,443m plus marketable debt securities of $34,143m and a further $42,783m of marketable equity securities; total shareholders' equity was $228,984m, accounts receivable $63,059m and inventories $31,575m.

Guidance for fiscal Q3 2027, given on 26 Aug 2026: revenue of $108.0bn plus or minus 2%, GAAP and non-GAAP gross margin of 74.0% plus or minus 50 basis points, GAAP operating expenses of approximately $9.2bn and non-GAAP operating expenses of approximately $9.0bn, with a full-year tax rate of 16.0% to 18.0%. The outlook assumes no Data Center compute revenue from China.

6. Valuation Metrics

Raw metrics, September 2026. Not opinions on whether the stock is cheap or expensive.

Metric	Value
Share price (3 Sep 2026 close)	$228.45
Market cap	~$5,516bn (24.147bn shares outstanding per the 10-Q cover page × $228.45)
Trailing P/E (GAAP)	~28.9x (trailing twelve-month GAAP diluted EPS of $7.91 = first-half FY2027 $4.85 plus Q4 FY2026 $1.76 plus Q3 FY2026 $1.30). Note that GAAP earnings are flattered by $23.7bn of equity-securities gains in the first half; on trailing non-GAAP EPS of $7.01 the same calculation gives approximately 32.6x, though the non-GAAP definition changed mid-series.
P/E (forward)	~23.4x, derived from company guidance rather than consensus: Q3 FY2027 revenue of $108.0bn at 74.0% gross margin less $9.0bn non-GAAP operating expenses implies roughly $70.9bn of operating income, or about $2.44 per share after tax at the 17% midpoint on 24.1bn shares, annualised at four times. This assumes no further sequential growth and excludes non-operating equity gains.
P/S (TTM)	~18.2x (market cap ~$5,516bn / trailing twelve-month revenue $302,970m)
Enterprise value	~$5,493bn (market cap ~$5,516bn + total debt $33,366m − cash, equivalents and marketable debt securities $56,586m, per the 26 Jul 2026 balance sheet. Excludes $42,783m of marketable equity securities.)
EV/EBITDA (TTM)	~27.3x (EV ~$5,493bn / EBITDA $201,266m; EBITDA = trailing operating income $197,579m + depreciation and amortisation $3,687m, both built from the income statement and cash flow statement across the second half of FY2026 and the first half of FY2027)
P/FCF	~43.4x (market cap ~$5,516bn / free cash flow $127,006m; FCF = trailing operating cash flow $134,360m − trailing capital expenditure $7,354m)
Price/book	~24.1x (market cap ~$5,516bn / total shareholders' equity $228,984m at 26 Jul 2026)
52-week high	$236.54
52-week low	$164.07
Short interest (% of float)	~1.23% (285,956,804 shares short at the 14 Aug 2026 settlement date against a float of 23.127bn shares)
Days to cover	2.35

Two cautions. The trailing GAAP multiple is calculated on earnings that include large non-operating gains on equity investments, which can reverse; net unrealised gains on publicly held equity securities were $12.5bn in the first half alone. The forward multiple rests on a single quarter of company guidance annualised, because NVIDIA does not guide earnings per share or full-year revenue. Compare the live picture using ChartsView Live Charts.

7. What Are They Building

NVIDIA spent $7,054m on research and development in fiscal Q2 2027 alone, up 64% year on year, and $18,497m across FY2026 — three and a half times the FY2022 figure.

Vera Rubin. The current flagship platform entered full production and is deployed at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius. It comprises six co-designed chips: the Rubin GPU, Vera CPU, NVLink 6 switch, ConnectX-9 SuperNIC, BlueField-4 DPU and Spectrum-6 Ethernet switch. The NVL72 rack pairs 72 Rubin GPUs with 36 Vera CPUs at roughly 260 TB/s of aggregate NVLink bandwidth, rated at 3.6 exaflops of NVFP4 inference and 2.5 exaflops of training, fully liquid cooled, and claimed at up to 30 times more work per watt for agentic AI than the prior generation.

Vera and Rubin CPX. NVIDIA describes Vera as the first CPU built for AI agents, with SpaceXAI announced as a deployer on 24 Aug 2026. Rubin CPX is a specialised GPU class with 128GB of GDDR7 optimised for the compute-bound prefill phase of million-token-context inference, paired with standard Rubin GPUs in the NVL144 CPX rack.

Networking and storage. Spectrum-6 switch systems supporting both pluggable and co-packaged optics are being deployed across large AI facilities. NVLink Fusion was expanded with NVHBM custom high-bandwidth memory on 26 Aug 2026, and BlueField-4 underpins an AI-native inference context memory storage tier.

Roadmap beyond Rubin. Rubin Ultra is targeted for the second half of 2027 in the Kyber NVL576 rack, moving to a quad-die GPU module. Reports in July 2026 suggested Kyber could slip toward 2028; NVIDIA responded that its roadmap is intact but published no revised schedule. The post-Rubin generation, Feynman, is expected in 2028 paired with the Rosa CPU, using stacked GPUs and optical NVLink. The Groq 3 LPX interactive inference accelerator, arising from the Groq licensing transaction, entered full production on 24 Aug 2026.

8. Competitors and Market Position

Market capitalisations below were re-checked live on 3–4 September 2026.

Peer	Market cap (September 2026)	Key 2025 metric
Broadcom (AVGO)	~$1,699bn	Trailing twelve-month revenue $75.5bn at 76.3% gross margin; the principal custom-XPU design partner to the hyperscalers
Alphabet (GOOGL)	~$4,189bn	Trailing twelve-month revenue $445.9bn; operates the Tensor Processing Unit programme, the largest internal custom accelerator effort
Amazon (AMZN)	~$2,793bn	Trailing twelve-month revenue $775.7bn; develops Trainium and Inferentia, yet agreed on 26 Aug 2026 to take two million additional NVIDIA GPUs
AMD (AMD)	~$745bn	Trailing twelve-month revenue $41.3bn; signed six-gigawatt agreements with both Meta and OpenAI; Instinct estimated at 5–7% of the AI accelerator market
Intel (INTC)	~$485bn	Trailing twelve-month revenue $57.0bn at 38.9% gross margin and loss-making on a trailing basis; AI GPU share estimated near 1%
Marvell (MRVL)	~$188bn	Trailing twelve-month revenue $9.45bn; second-source custom AI silicon designer to the hyperscalers

Estimates of NVIDIA's share of the AI accelerator market for 2026 range from 70% to 86%, narrowing to roughly 75–81% for the first half of 2026 once hyperscaler custom silicon is counted in the denominator. The structural pressure is visible in shipment growth rather than share: TrendForce puts custom-ASIC shipment growth at 44.6% in 2026 against 16.1% for merchant GPUs. Two other companies matter more as dependencies than as rivals. TSMC, at roughly $2,163bn, fabricates essentially all leading-edge NVIDIA silicon and is the single most critical supplier. Micron, at roughly $1,082bn, is one of three qualified high-bandwidth memory suppliers, all of which have sold out 2026 HBM capacity. NVIDIA has also noted in its own filings that exclusion from China helped competitors there build larger developer and customer ecosystems.

9. Insider Activity

The most striking feature of NVIDIA's 2026 Form 4 record is what founder, President and Chief Executive Officer Jensen Huang did not do: he made no open-market sales of NVIDIA stock through 4 September 2026, a break from prior years' pre-arranged selling programmes. His only dispositions were gifts — 117,925,204 shares on 18 Mar 2026 and a further 400,000 on 16 Jun 2026, transactions typical of transfers to family trusts or charitable vehicles — plus mandatory tax withholding on vesting. Director Mark Stevens, by contrast, has been by far the largest seller. All transactions below are from Forms 4 filed under CIK 0001045810.

Name	Date	Type	Shares	Price	Value	Plan Type
Mark A. Stevens (Director)	31 Aug – 2 Sep 2026	Sale	1,848,501	$220.06–$226.27	~$410.8m	No plan cited
Timothy S. Teter (EVP, General Counsel)	31 Aug 2026	Sale	30,000	$216.84–$218.53	~$6.5m	Rule 10b5-1
Mark A. Stevens (Director)	18 Jun 2026	Sale	885,000	$209.70–$210.44	~$186.0m	No plan cited
Mark A. Stevens (Director)	2–4 Jun 2026	Sale	1,000,000	$217.66–$222.38	~$221.1m	No plan cited
Stephen C. Neal (Director)	3 Jun 2026	Sale	15,500	$215.73	$3.34m	No plan cited
Colette Kress (EVP and CFO)	20 Mar 2026	Sale	62,645	$171.97–$177.51	~$11.0m	Rule 10b5-1
Ajay K. Puri (then EVP, Worldwide Field Operations)	18 Mar 2026	Sale	300,000	$181.31–$182.94	~$54.8m	Rule 10b5-1
Aarti S. Shah (Director)	19 Mar 2026	Sale	19,000	$176.71	$3.36m	Rule 10b5-1

Mark Stevens sold roughly 3,955,183 shares for approximately $856m across 2026, and his filings do not carry the Rule 10b5-1 footnote that the officer sales do. There were no open-market purchases by any insider during the period. Remaining Form 4 activity was routine tax withholding on vesting, annual director stock grants and the March 2026 executive equity awards.

10. Key Risks

• Extreme customer concentration: one direct customer was 16% of total revenue in fiscal Q2 2027, and three direct customers were 16%, 15% and 13% of first-half revenue. NVIDIA's own risk factors state that revenue could be adversely affected if it loses or is prevented from selling to any of these customers. Indirect concentration in a handful of hyperscalers is greater still.
• China and export controls: NVIDIA states it has been effectively foreclosed from China's data-centre compute market. It took a $4.5bn H20 charge in FY2026 and a $0.4bn H200 charge in the first half of FY2027, shipped less than 1% of Data Center revenue to China in the quarter, and faces a 25% US import tariff on H200 units that it cannot pass to customers. The replacement for the rescinded AI Diffusion rule remains undefined.
• Purchase and lease commitments of $366bn: supply and capacity commitments of $279bn, cloud service agreements of $29bn, uncommenced data-centre leases of $25bn, equity investment commitments of $25bn and capital expenditure of $8bn, with roughly $120bn falling due in the remainder of FY2027. NVIDIA's risk factors already acknowledge that long lead times combined with inaccurate demand estimates have led to supply and demand mismatches.
• Counterparty and vendor-financing exposure: the $105bn cap on residual value guaranties for OpenAI's Ohio leases, $3,529m notional of land, power and shell guarantees against $712m of partner escrow, $4.7bn of maximum variable-interest-entity loss exposure, $4,800m notional of public company warrants and $51.16bn of non-marketable securities together convert customer credit risk into NVIDIA balance-sheet risk.
• Competition from custom silicon: custom-ASIC shipment growth of 44.6% in 2026 against 16.1% for merchant GPUs, driven by Google, Amazon, Microsoft and Meta programmes designed with Broadcom and Marvell, targets precisely NVIDIA's largest customers. AMD has signed six-gigawatt agreements with both Meta and OpenAI.
• Supply-chain concentration: NVIDIA manufactures nothing, depending on third parties to fabricate, assemble, package and test. Essentially all leading-edge output comes from TSMC, advanced packaging capacity is a hard constraint, and high-bandwidth memory is sold out across all three qualified suppliers for calendar 2026.
• Earnings quality and reversal risk: GAAP other income of $7,773m in the quarter was almost entirely equity-securities gains, and marketable equity securities rose from $12.89bn to $42.78bn in six months. Reported GAAP earnings now move materially with the share prices of NVIDIA's own ecosystem investees and can reverse.
• Antitrust and regulatory scrutiny: the Form 10-Q discloses increased regulator interest in the European Union, United States, United Kingdom, South Korea, Japan and China, including broad information requests from EU competition regulators. China's antitrust authorities published a preliminary finding in September 2025 that NVIDIA violated conditions of the Mellanox approval, the US Department of Justice has issued subpoenas over exclusivity practices, and the French Competition Authority is examining whether gaming and data-centre GPUs are separate product categories. The pending Hugging Face acquisition requires approvals in that environment.
• Power and physical build-out constraints: NVIDIA is now securing land, power and shell capacity directly, an acknowledgement that grid capacity rather than silicon is becoming the gating factor. It has committed $25bn to data-centre leases that have not yet commenced, dependent on construction completion, with terms of up to twenty years.

11. Recent Developments

• 03 Sep 2026 — NVIDIA agrees to acquire Hugging Face for $12.93bn. The definitive agreement was signed on 2 Sep 2026 and comprises approximately $11.9bn of cash to stockholders plus up to roughly $1.0bn of retention equity, expected to close in the first half of calendar 2027 subject to regulatory approvals. NVIDIA committed to keep the platform open to any model or dataset and to continue supporting other silicon vendors, and added a new standalone risk factor covering government restrictions on open-source models.
• 31 Aug 2026 — Director Mark Stevens begins the largest insider disposal of the year, selling 1,848,501 shares for approximately $410.8m across 31 Aug to 2 Sep 2026.
• 26 Aug 2026 — fiscal Q2 2027 results. Revenue of $96.22bn, up 106% year on year, GAAP diluted EPS of $2.46, GAAP gross margin of 75.0%, and Q3 guidance of $108.0bn assuming no Data Center compute revenue from China. Approximately $26.0bn was returned to shareholders in the quarter with $99.0bn of buyback authorisation remaining.
• 26 Aug 2026 — AWS agrees to deploy two million additional NVIDIA GPUs alongside next-generation infrastructure for agentic and physical AI, a significant expansion of the existing collaboration. NVLink Fusion was expanded with NVHBM custom high-bandwidth memory the same day.
• 17 Aug 2026 — $105bn residual value guarantee for OpenAI data-centre leases. NVIDIA entered residual value guaranties with SB Energy covering leases for roughly 4.25 gigawatts of IT load at the PORTS-Pike campus in Pike County, Ohio, with discretion to credit-support a further 3.8 gigawatts. OpenAI is the tenant and has agreed to reimburse and indemnify NVIDIA. Obligations trigger on OpenAI insolvency or payment failure and are expected to begin in 2028.
• 11 Aug 2026 — more than $500bn of third-party AI infrastructure financing platforms announced with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, subject to definitive agreements.
• 02 Jul 2026 — leadership transition in worldwide sales. Ajay Puri, EVP Worldwide Field Operations and a 21-year veteran, notified the company on 28 Jun 2026 of his intention to retire. Nicholas Parker, previously EVP and Chief Business Officer of Microsoft's Worldwide Sales and Solutions, took the role on 24 Aug 2026.
• 17 Jun 2026 — $25.0bn senior notes issued across seven tranches maturing between 2028 and 2056, taking non-current long-term debt from $7.5bn to $32.4bn, alongside a $25.0bn commercial paper programme with nothing drawn.
• 07 Jun 2026 — multi-year memory partnership with SK hynix covering next-generation memory across NVIDIA's forward product roadmap, part of a broader effort to secure high-bandwidth memory allocation across all three qualified suppliers.
• 20 May 2026 — $80bn additional buyback authorisation and a 25-fold dividend increase. The board approved the repurchase authorisation on 18 May 2026 and raised the quarterly dividend from $0.01 to $0.25 per share, first paid on 26 Jun 2026.

12. Key Dates

• 10 Sep 2026 — record date for the $0.25 per share quarterly dividend declared with fiscal Q2 2027 results.
• 01 Oct 2026 — payment date for that $0.25 per share quarterly dividend.
• Expected Nov 2026 — fiscal Q3 2027 results for the quarter ending in late October 2026. NVIDIA had not issued a conference-call scheduling release as of 4 Sep 2026; the prior-year equivalent was reported on 19 Nov 2025. Aggregator estimates cluster around late November. Track confirmations on the ChartsView Economic Calendar.
• Expected Feb 2027 — fiscal Q4 and full-year 2027 results, based on the pattern of the FY2026 report on 25 Feb 2026.
• 14 Mar 2027 — GTC AI Conference 2027 opens at the San Jose McEnery Convention Center, running to 18 Mar 2027. Registration is stated to open later in autumn 2026 and the keynote line-up has not been announced.
• Expected Jun 2027 — 2027 annual meeting of stockholders. The 2026 meeting was held virtually on 24 Jun 2026 and results were filed on 30 Jun 2026.
• TBD — completion of the Hugging Face acquisition, which NVIDIA expects in the first half of calendar 2027 subject to regulatory approvals.

Lease commencements at the OpenAI and SB Energy Ohio campus are expected to be ready for service beginning in 2028, which is when NVIDIA's residual value guarantee obligations could first be triggered. No regional GTC date for Washington DC in 2026 has been confirmed.

Discuss this company with other ChartsView members in the Forum.

---

Disclaimer: This research is produced by ChartsView for educational and informational purposes only. It does not constitute financial advice or a recommendation to buy or sell any security. All information is sourced from publicly available company filings, press releases, and official data. ChartsView does not use analyst opinions or third-party ratings. Always conduct your own due diligence and consider your personal financial situation before making investment decisions. Past performance is not indicative of future results.