Nvidia delivered a quarter that beat expectations across the board and, for the first time, provided a formal revenue framework for the next fiscal year, projecting roughly 70% year-on-year growth. However, JPMorgan believes this guidance may underestimate the true explosive potential of the agentic AI wave and sovereign demand.
Nvidia's fiscal second-quarter revenue (through July) reached $96.1 billion, up 106% year-on-year, marking the fourth consecutive quarter of accelerating growth. Meanwhile, the company set its fiscal third-quarter revenue guidance midpoint at $108 billion, exceeding market expectations by approximately $3.5 billion. More notably for the market, management provided its first formal revenue framework for fiscal 2028 (ending January 2026), projecting approximately 70% year-on-year growth—but explicitly characterized this as a conservative estimate under "supply-constrained" conditions, noting that in an unconstrained supply scenario, the actual pace of demand would be significantly higher.
According to trading desk sources, JPMorgan analyst Harlan Sur noted in a report that, given the accelerating adoption of agentic AI applications, the ongoing mobilization of global sovereign compute demand, and the additional supply flexibility being released by Nvidia's supply chain partners, the fiscal 2028 framework "is likely still conservative." JPMorgan raised its price target on Nvidia from $280 to $320 (corresponding to December 2027) while maintaining an "Overweight" rating.
Quarterly Results: Data Center Accelerates Across the Board, Sovereign AI a Bright Spot
Nvidia's fiscal Q2 2027 (July quarter) revenue was $96.1 billion, exceeding market expectations by about $4.1 billion. Data center revenue reached $89 billion, up 117% year-on-year, also surpassing the market estimate of $87.3 billion.
Within the data center segment, revenue from hyperscale cloud providers was $48.7 billion, up 13% quarter-on-quarter, driven by sustained high demand for Blackwell Ultra. Revenue from ACIE (AI cloud, enterprise, and international) was $40.3 billion, up 25% sequentially and 138% year-on-year, driven by demand from AI-native clouds, enterprises, and sovereign customers. Notably, sovereign AI revenue (primarily through regional emerging cloud providers) grew 35% quarter-on-quarter and more than tripled year-on-year, pushing international revenue mix to 38%.
Additionally, networking revenue grew 18% sequentially, with Spectrum-X Ethernet products up 2.6 times year-on-year. Edge computing revenue was $7.2 billion, up 13% quarter-on-quarter. Shareholder returns hit a record $26 billion in the quarter ($20 billion in buybacks plus $6 billion in dividends), and the year-to-date free cash flow return ratio has reached 60%, exceeding the previously planned target of "over 50%."
Fiscal 2028 Framework: Agentic Inference Expands TAM, Supply Remains the Primary Constraint
Management's fiscal 2028 revenue framework—roughly 70% year-on-year growth, corresponding to JPMorgan's estimate of approximately $695 billion—attributes growth primarily to TAM expansion driven by agentic inference. Key data points supporting this framework include: hyperscale backlog has exceeded $2 trillion; the Vera Rubin platform is expected to generate over $40 billion in revenue per gigawatt of compute, well above Grace Blackwell's $25 billion and Hopper's $18 billion; and CPU revenue is expected to more than double in fiscal 2028 as Vera Rubin ramps.
Vera Rubin production shipments began this month (August), with purchase orders in place from all major hyperscalers, AI clouds, and system OEMs. Management characterized this as "the fastest product ramp in Nvidia's history," noting that the order tail for the Blackwell product line is "rapidly narrowing." Vera Rubin is expected to contribute approximately 20% of data center revenue in the fiscal third quarter, with hyperscale expected to re-accelerate in the fiscal fourth quarter and into fiscal 2028 as Vera Rubin supply expands.
Management also explicitly described the fiscal 2028 framework as guidance under a supply-constrained scenario, stating that the actual pace of demand in an unconstrained environment would be substantially higher. JPMorgan believes the pace of agentic AI adoption and the global mobilization of sovereign demand will make this framework conservative.
Gross Margin Trajectory Revised Lower, but HBM Price Lockups Provide Downside Protection
One of the main expectation gaps in this earnings report was the adjustment to gross margin guidance. Fiscal third-quarter gross margin guidance is 74.0%, still slightly above the market consensus of approximately 73.5%, but down about 100 basis points sequentially from the prior quarter, due to memory cost inflation exceeding previous expectations.
Management further provided a longer-term gross margin trajectory: margins are expected to bottom out in the 71%-72% range in the fiscal fourth quarter (FQ4 2027), then gradually recover as price increases take effect from the start of fiscal 2028 Q1, with full-year fiscal 2028 gross margin expected in the 72%-73% range.
For investors, a key risk mitigation signal is that Nvidia has locked in supply commitments and known pricing for the majority of its HBM requirements for fiscal 2028, with these memory contracts covering a three-year forward period. JPMorgan describes this as a "downside floor" for gross margins, providing some hedge against near-to-mid-term margin uncertainty. Of Nvidia's approximately $350 billion in total supply commitments, the majority is related to memory (primarily HBM).
Financing Architecture Expansion: Revenue Sharing and a $500 Billion Private Platform Create New Revenue Streams
Nvidia has significantly expanded the infrastructure financing architecture supporting forward demand, which JPMorgan views as a key highlight of this quarter.
Specifically, there are several layers. First, Nvidia's PORTS-Pike guarantee supports the construction of OpenAI's Ohio data center, involving approximately 4.25 gigawatts of IT load, with an additional pre-emptive right for roughly 3.8 gigawatts. This is expected to generate approximately $15 billion to $20 billion in revenue per product generation, spanning multiple 20-year upgrade cycles.
Second, Nvidia has entered into revenue-sharing agreements with certain emerging cloud providers. Nvidia provides a base pricing guarantee while capturing 50% of the difference between actual market rental prices and the base price. JPMorgan describes this as an "additional recurring revenue stream" layered on top of hardware sales. As the current 5 to 20 managed revenue-sharing projects come online, this is expected to contribute billions of dollars in incremental revenue over time.
Third, Nvidia, together with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR, has built a $500 billion private capital financing platform to support infrastructure construction for frontier AI labs. Nvidia's residual credit exposure cap per individual financing is approximately 25%, with no actual capital committed to date. Additionally, Nvidia has cumulatively invested nearly $50 billion in frontier AI labs to date. OpenAI has committed to large-scale deployment of Nvidia AI infrastructure by 2030 (approximately 12 gigawatts cumulative). AI lab customers supported by Nvidia's balance sheet are expected to contribute approximately one-quarter of fiscal 2028 revenue.
In response to market concerns about "circular financing," management characterized these arrangements as being measured by demand intensity, ecosystem returns, and the investment-grade credit ratings of the underlying compute purchasers. JPMorgan stated it broadly agrees with this framework but expects the discussion to continue.
JPMorgan raised its price target on Nvidia to $320, based on approximately 20 times price-to-earnings (corresponding to CY26 EPS of $15.87). At the current share price of approximately $209, the fiscal 2028 forward P/E is approximately 13 times, and fiscal 2028 forward EV/EBITDA is approximately 10 times.
Comments