NVIDIA Q2 FY27: Revenue +106% to $96.2B, net income +126% to $59.7B, on Blackwell Ultra ramp

The Supreme Court dismissed NVIDIA's certiorari petition as improvidently granted in December 2024, and the Ninth Circuit's partial reversal took effect in February 2025, remanding the securities class action to district court. The current filing references Note 10 for developments since January 2026, indicating the case has progressed to active district court proceedings after the Supreme Court's dismissal ended NVIDIA's appellate efforts.

The baseline disclosed that one plaintiff in the Delaware Chancery derivative case (Horanic v. Huang) was voluntarily dismissed with prejudice in August 2025. The current filing's reference to Note 10 for developments since January 2026 suggests further procedural changes in the derivative litigation landscape, though the specific nature of those changes is not detailed in the Item 1 Legal Proceedings section itself.

NVIDIA introduced a new data center architecture, Vera Rubin, which began production shipments in Q3 FY27. The company will ship both Blackwell and Rubin systems concurrently and is experiencing supply constraints. This represents a new product generation beyond Blackwell.

Supply and capacity commitments surged from $119B in Q1 FY27 to $279B as of Q2 FY27 — a $160B sequential increase. This reflects NVIDIA's aggressive pre-positioning to secure manufacturing capacity for anticipated multi-year demand, but also concentrates execution risk and capital exposure.

NVIDIA entered into guarantees capping at $105B to support land, power, and shell buildout for a 4.25 GW data center campus in Ohio, with 20-year leases to OpenAI. The guarantees become effective as each of nine data centers is placed in service (expected to begin FY29), and NVIDIA's exposure declines as OpenAI fulfills lease payments. This is a major off-balance-sheet commitment tied to a single customer's infrastructure deployment.

NVIDIA introduced a new business model in Q2 FY27 with select AI cloud partners: NVIDIA commits to $36B (six-year duration) of cloud service agreements, the AI clouds procure NVIDIA infrastructure, and NVIDIA participates in revenue share if the AI clouds sell capacity to third parties. The commitments decrease as capacity is used by third parties or by NVIDIA for R&D. This model addresses AI cloud partners' infrastructure constraints but introduces revenue-timing complexity and off-balance-sheet exposure.

NVIDIA entered into MOUs with large capital providers to establish financing platforms targeting $500B+ of third-party capital for AI infrastructure deployment. These are preliminary (may not lead to definitive agreements), and NVIDIA may optionally provide limited residual-value support on specific projects. This is a new ecosystem-financing initiative to address customer capital constraints.

NVIDIA disclosed $99B in equity investments and $25B in equity investment commitments as of Q2 FY27. The baseline filing did not disclose comparable figures. This represents a major balance-sheet and off-balance-sheet exposure to ecosystem companies, likely including AI startups, cloud partners, and infrastructure providers.

The China licensing narrative shifted from H20 (FY26) to H200 (FY27). In FY27 H1, NVIDIA incurred a $0.4B charge for H200 excess inventory and purchase obligations; H200 shipments under the new licensing program are subject to a 25% U.S. import tariff (due to a required U.S. inspection process), which NVIDIA cannot pass to customers. H200 shipments account for <1% of Data Center revenue in Q2 FY27. The baseline H20 charge was $4.5B in Q1 FY26; Q2 FY26 saw $650M H20 revenue to a non-China customer. The product, charge magnitude, and tariff mechanics all changed.

Q2 FY27 revenue was $96.2B (up 106% YoY, 18% QoQ) vs Q2 FY26 revenue of $46.7B (up 56% YoY, 6% QoQ). Operating income was $63.7B in Q2 FY27 (up 124% YoY) vs $28.4B in Q2 FY26 (up 53% YoY). The YoY growth rate more than doubled, driven by the Blackwell Ultra ramp.

Data Center revenue was $89.0B in Q2 FY27 (up 117% YoY, 18% QoQ) vs $41.1B in Q2 FY26 (up 56% YoY, 5% QoQ). FY27 introduced a new customer-mix breakdown: Hyperscale $48.7B (up 102% YoY, 13% QoQ) and AI Clouds, Industrial, & Enterprise (ACIE) $40.3B (up 138% YoY, 25% QoQ). The baseline did not report this split; it stated CSPs represented ~50% of Data Center revenue in Q2 FY26. The ACIE category is growing faster than Hyperscale.

Q2 FY27 gross margin was 75.0% (up 2.6 pts YoY, flat QoQ) vs Q2 FY26 gross margin of 72.4% (down 2.7 pts YoY, up 11.9 pts QoQ). The YoY comparison flipped from a decline to an increase, driven by improved mix from Blackwell Ultra. The baseline's sequential jump reflected the non-recurrence of the Q1 FY26 H20 charge.

Q2 FY27 operating expenses were $8.4B (up 55% YoY, 10% QoQ) vs Q2 FY26 opex of $5.4B (up 38% YoY, 8% QoQ). The YoY growth rate accelerated from 38% to 55%, driven by higher compute infrastructure and compensation costs.

R&D compute infrastructure spend grew 127% YoY in Q2 FY27 (120% in H1 FY27) vs 68% YoY in Q2 FY26 (69% in H1 FY26). The growth rate nearly doubled, reflecting NVIDIA's internal AI infrastructure buildout. Compensation growth was stable at ~30% YoY in both periods.

Q2 FY27 other income was $7.8B (vs $2.8B in Q2 FY26), driven by $7.8B in gains from equity securities (vs $2.2B in Q2 FY26). The baseline labeled this line "Other income (expense), net" and did not break out equity-security gains separately. The FY27 presentation is more granular and the dollar magnitude tripled, reflecting unrealized gains in NVIDIA's $42.8B marketable equity portfolio.

Q2 FY27 repurchases were $19.7B (94M shares) vs Q2 FY26 repurchases of $9.7B (67M shares). H1 FY27 repurchases were $39.8B (203M shares) vs H1 FY26 repurchases of $24.2B (193M shares). Dollar spend increased significantly (roughly doubled in Q2, up 64% in H1) despite similar share counts, reflecting the higher stock price.

Q2 FY27 dividends were $6.0B vs Q2 FY26 dividends of $244M. H1 FY27 dividends were $6.3B vs H1 FY26 dividends of $488M. The quarterly dividend increased from $0.01/share to $0.25/share in May 2026, a 25x increase, driving the dollar surge.

NVIDIA issued $25.0B of senior unsecured notes in June 2026. Total debt increased from $8.5B (net) as of Q2 FY26 to $33.4B (net) as of Q2 FY27. The maturity profile shifted: $1.0B due in one year, $15.0B in 1-5 years, $7.5B in 5-10 years, $10.0B beyond 10 years. This is a major debt raise to support general corporate purposes, likely including supply commitments and ecosystem investments.

The commercial paper program size increased from $575M to $25.0B. No commercial paper was outstanding in either period, but the authorization expansion reflects NVIDIA's need for short-term liquidity flexibility to support its larger scale of operations and commitments.

Unrecognized tax benefits increased from $2.9B (Q2 FY26) to $5.0B (Q2 FY27), a $2.1B increase. This reflects growing tax-position uncertainty as NVIDIA's income and cross-border operations scale. The company is under IRS examination for FY23 and FY24.

NVIDIA entered into guarantees capped at $105 billion to support an OpenAI-affiliated customer's data center leases at SB Energy's Ohio campus, covering approximately 4.25 gigawatts of IT load across nine phases beginning in fiscal 2029. The guarantees are triggered by tenant defaults, decrease over each 20-year lease term, and terminate if OpenAI achieves a satisfactory credit rating or the lease completes. In exchange, the site will exclusively host NVIDIA AI infrastructure, and NVIDIA holds an option to provide additional credit support for approximately 3.8 additional gigawatts.

NVIDIA issued $25 billion of senior unsecured notes in June 2026 across seven tranches (maturities ranging from 2028 to 2056, rates 4.25% to 5.625%) for general corporate purposes. This is a major debt raise, increasing long-term debt from $7.5 billion to $32.4 billion period-over-period.

NVIDIA increased supply and capacity commitments from $119 billion last quarter to $279 billion as of July 26, 2026, securing memory and manufacturing capacity for data center infrastructure systems to meet long-term demand across current and future product architectures. The baseline filing did not disclose this level of detail or magnitude.

NVIDIA entered into $36 billion of AI cloud agreements (fiscal 2028–2032+) under which AI clouds procure NVIDIA data center infrastructure and NVIDIA commits to cloud service agreements. The AI clouds can unilaterally stop providing capacity to NVIDIA and sell to third parties at more advantageous rates; NVIDIA's commitments decrease as capacity is used by third parties or by NVIDIA for R&D, and NVIDIA participates in revenue share from third-party customers if certain criteria are met. This is a new partnership structure not disclosed in the baseline.

NVIDIA held $42.8 billion of publicly-held equity securities as of July 26, 2026 (up from $12.9 billion at January 25, 2026), of which $36.9 billion are subject to short-term lock-up restrictions and $5.0 billion are subject to lock-up restrictions through December 2027. The baseline filing disclosed $3.2 billion of publicly-held equity securities with $2.8 billion subject to short-term lock-up. This is a major increase in equity holdings and lock-up exposure.

NVIDIA held $47.9 billion of non-marketable equity securities as of July 26, 2026 (up from $3.8 billion at July 27, 2025), including $3.3 billion of equity-method investments in infrastructure financiers. Those equity-method investments deemed VIEs had a maximum loss exposure of $4.7 billion; NVIDIA is not the primary beneficiary and does not consolidate the VIEs. This is a new disclosure of VIE exposure and a major increase in non-marketable equity holdings.

NVIDIA received warrants to purchase shares of publicly-traded common stock with terms of three to five years in the second quarter of fiscal 2027. The warrants are classified as equity derivatives with a fair value of $824 million as of July 26, 2026 (Level 3 measurement), and the corresponding benefit was substantially deferred. This is a new derivative instrument not disclosed in the baseline.

NVIDIA entered into land, power, and shell guarantees for select AI cloud partners' data center lease obligations in the event of their default, with a maximum gross exposure of $3.5 billion. The baseline filing did not disclose these guarantees.

NVIDIA entered into $20 billion of data center leases (fiscal 2028–2032+) with terms of approximately fifteen years, expected to commence between fiscal 2028 and 2029, that NVIDIA expects to reassign to third parties. The baseline filing did not disclose these leases.

NVIDIA accrued $986 million as of July 26, 2026 (down from $3.9 billion at January 25, 2026) related to a Groq, Inc. non-exclusive license agreement. The cash flow statement shows $2.9 billion paid to Groq in the first half of fiscal 2027. This is a new disclosure of a major license agreement.

Customer advances increased from $80 million at July 27, 2025 to $2.8 billion at July 26, 2026, reflecting a major increase in prepayments from customers for data center infrastructure. The baseline filing disclosed $80 million of customer advances at July 27, 2025.

NVIDIA changed its revenue presentation from "Revenue by End Market" (Data Center split into Compute and Networking, plus Gaming, Professional Visualization, Automotive, OEM and Other) to "Revenue by Market Platform" (Data Center split into Hyperscale and AI Clouds, Industrial, & Enterprise, plus Edge Computing). The new presentation reflects NVIDIA's focus on AI cloud partnerships and hyperscale customers. The baseline filing used the prior presentation.

Inventory provisions for the first half of fiscal 2027 were $1.6 billion, down from $3.2 billion in the first half of fiscal 2026. The baseline filing disclosed $3.2 billion of inventory provisions for the first half of fiscal 2026. This reflects lower obsolescence charges as NVIDIA's product mix stabilizes.

Excess inventory purchase obligations decreased from $3.2 billion at July 27, 2025 to $2.1 billion at July 26, 2026, and the charge for the first half of fiscal 2027 was $501 million, down from $3.1 billion in the first half of fiscal 2026. This reflects lower excess-capacity charges as NVIDIA's supply commitments align with demand.

Product warranty liabilities increased from $2.1 billion at July 27, 2025 to $2.9 billion at July 26, 2026, reflecting higher warranty accruals as data center infrastructure deployments scale. The baseline filing disclosed $2.1 billion of product warranty and return provisions at July 27, 2025.

Operating lease assets increased from $2.1 billion at July 27, 2025 to $5.4 billion at July 26, 2026, and long-term operating lease liabilities increased from $1.8 billion to $5.0 billion, reflecting major new data center leases. The baseline filing disclosed $2.1 billion of operating lease assets and $1.8 billion of long-term operating lease liabilities at July 27, 2025.

Goodwill increased from $5.8 billion at July 27, 2025 to $21.1 billion at July 26, 2026, reflecting major acquisitions in the first half of fiscal 2027. The baseline filing disclosed $5.8 billion of goodwill at July 27, 2025 and a $567 million increase from acquisitions in the first half of fiscal 2026.

Intangible assets increased from $755 million at July 27, 2025 to $3.0 billion at July 26, 2026, reflecting major acquisitions in the first half of fiscal 2027. The baseline filing disclosed $755 million of intangible assets at July 27, 2025.

Unearned stock-based compensation increased from $14.0 billion at July 27, 2025 to $19.4 billion at July 26, 2026, reflecting major new equity grants in the first half of fiscal 2027. The weighted average recognition period increased from 2.2 years to 2.6 years.

NVIDIA increased its quarterly cash dividend from $0.01 per share to $0.25 per share on May 18, 2026, a 25x increase. The baseline filing did not disclose this dividend increase.

NVIDIA's Board approved an additional $80 billion in share repurchase authorization on May 18, 2026, increasing the total authorization to $99.3 billion as of July 26, 2026. The baseline filing disclosed $14.7 billion of remaining authorization as of July 27, 2025, and an additional $60 billion approved on August 26, 2025 (after the baseline period end).

NVIDIA added a new risk factor disclosing multi-year capital commitments for supply/capacity, cloud services, and data center leases; financial guarantees and credit support for customers/partners building AI infrastructure; agreements with AI clouds under which NVIDIA may be obligated to purchase unsold capacity; and specific guarantees with SB Energy relating to PORTS Technology Campus leases tied to OpenAI performance. The disclosure also mentions August 2026 memoranda of understanding with large capital providers for independent financing platforms. These arrangements expose NVIDIA to counterparty default, execution delays, power/permitting constraints, and the risk of bearing lease costs or purchasing capacity it cannot resell.

The baseline discussed H20 licensing (August 2025 licenses granted but no shipments/revenue) and a potential 15% USG revenue share. The current filing replaces this with H200 licensing detail: February 2026 licenses granted, PRC restrictions prevented full sales, $0.4B charge for excess H200 inventory/obligations, shipments under 1% of Data Center revenue, a 25% tariff on H200s due to required U.S. inspection, and NVIDIA's inability to pass the tariff to customers. This reflects a shift from H20 to H200 product, concrete inventory charges, and new tariff burden.

The baseline framed China data center foreclosure as a conditional future risk ("we may be unable"). The current filing states it as present fact: "we are unable" and "we were effectively foreclosed" as of Q2 FY2027. It adds that this foreclosure has already helped competitors build ecosystems to challenge NVIDIA worldwide, and frames the impact as ongoing unless NVIDIA can return with a product approved by both governments. This is a hardening from hypothetical to realized competitive harm.

The baseline mentioned an ongoing investigation into whether U.S. export-control compliance discriminates against China customers. The current filing adds that on September 15, 2025, China's antitrust regulators published a preliminary finding that NVIDIA's compliance with U.S. export controls (offering degraded products) violated the terms of China's Mellanox acquisition approval. This escalates from inquiry to formal preliminary adverse finding, increasing the risk of penalties or restrictions on NVIDIA's China operations.

NVIDIA added disclosure of two legislative developments: the GAIN AI Act (passed Senate October 2025), which would restrict the Trump Administration's ability to modify Biden-era export controls and allow private parties to review/overturn licensing decisions; and the Remote Access Security Act (RASA), under consideration, which could prohibit cloud services to companies with China-based ultimate parents. Both could materially impact NVIDIA's business by constraining export-control flexibility or cloud-service customer base.

NVIDIA added that the French Competition Authority is questioning whether gaming and data center GPUs are separate product categories, an inquiry that may affect export controls on gaming products in France and Europe. This is a new regulatory development that could expand the scope of export restrictions to gaming products if the authority concludes the categories overlap.

The baseline mentioned state AI regulations under consideration or enacted. The current filing adds that new state laws took effect on January 1, 2026, and specifies they may impact AI models/systems "including those capable of autonomous action." This reflects the transition from pending to effective regulations and adds detail on the scope (autonomous-action systems).

NVIDIA added a new risk factor on indebtedness, disclosing $33.5B in senior notes outstanding as of July 26, 2026, and a $25.0B commercial paper program (no amounts outstanding). The disclosure covers refinancing risk, debt-service burden, covenant-breach acceleration risk, and credit-rating impact. This is a standard debt-risk disclosure for a company with material outstanding debt.