Wall Street's imagination cannot keep pace with NVIDIA's "speed".
NVIDIA has once again delivered an impressive financial report, with revenue, operating profit and earnings per share all hitting record highs.
On August 26 local time in the United States, NVIDIA released its financial results for the second quarter of fiscal 2027 ending July 26, 2026. According to the earnings report, NVIDIA's revenue in the quarter reached 96.221 billion US dollars, up 106% year-on-year and 18% quarter-on-quarter; net profit reached 59.688 billion US dollars, up 126% year-on-year; diluted earnings per share was 2.46 US dollars, up 128% year-on-year.
Compared with the revenue of 81.615 billion US dollars in the previous quarter, NVIDIA continues to refresh its growth record. In the same period last year, the company's single-quarter revenue was only 46.743 billion US dollars, and it has almost doubled its growth in one year.
After the earnings report was released, NVIDIA's after-hours share price once fell by about 1.3%, then jumped by more than 4%, which also reflects the divergence and changes in market focus: in the past, investors were concerned about whether NVIDIA could continuously exceed expectations, but now they are more concerned about how long the AI infrastructure construction can continue, and whether the next-generation chips and new business models can support future growth.
From the perspective of profitability, NVIDIA still maintains an extremely high level.
In the second fiscal quarter, the company's operating profit reached 63.734 billion US dollars, up 124% year-on-year; calculated on a Non-GAAP basis, the net profit was 53.954 billion US dollars, up 118% year-on-year. In the same period, the company's gross profit margin reached 75%, higher than 72.4% in the same period last year, and slightly higher than 74.9% in the first fiscal quarter.
Jensen Huang, founder and CEO of NVIDIA, said: "Artificial intelligence has reached an inflection point. It is doing useful work, and its tokens are creating productivity and profitability. Now, computing is revenue."
Now, more AI labs and startups are expanding rapidly, and new directions such as the open-source model ecosystem and physical AI are also developing, and the entire AI industry is entering a broader construction cycle.
At the same time, NVIDIA continues to increase the intensity of shareholder returns. In the second fiscal quarter, the company returned about 26 billion US dollars to shareholders through stock repurchases and cash dividends. As of the end of the quarter, the company's remaining stock repurchase authorization limit was still about 99 billion US dollars.
01. Data center revenue surges, AI cloud customers grow faster
The core driving NVIDIA's current round of growth is still the data center business.
In the second fiscal quarter, NVIDIA's data center revenue reached 89.023 billion US dollars, up 117% year-on-year and 18% quarter-on-quarter, contributing almost all of the company's revenue increment. With global enterprises and cloud service providers continuing to invest in AI infrastructure construction, the demand for NVIDIA GPUs remains at a high level.
NVIDIA adjusted the disclosure method of its data center business in the first fiscal quarter, dividing customers into two categories: Hyperscale and ACIE.
Among them, the revenue from Hyperscale customers reached 48.71 billion US dollars, up 102% year-on-year and 13% quarter-on-quarter, mainly from large public clouds and Internet companies.
At the same time, the revenue from AI cloud, industrial and enterprise customers (ACIE) reached 40.313 billion US dollars, up 138% year-on-year and 25% quarter-on-quarter. This business covers AI-native companies, enterprise customers, sovereign AI customers, and hyperscale computing demands using AI cloud services.
The new classification method shows that AI computing power demand is spreading from a few cloud giants to enterprises, governments and more industry scenarios. However, investors still focus on the profitability and long-term demand behind different customer types, not just the growth of order scale.
The Chinese mainland market remains an important variable in the financial report.
NVIDIA stated that the revenue from data center Hopper products shipped to the Chinese mainland in the second fiscal quarter accounted for less than 1% of the total data center revenue. At the same time, the company did not include any data center computing revenue from the Chinese mainland in its performance outlook for the third fiscal quarter.
In addition to the data center business, the edge computing business achieved revenue of 7.198 billion US dollars in the second fiscal quarter, up 27% year-on-year and 13% quarter-on-quarter. Among them, sales of Blackwell workstations drove growth, but consumer PCs were affected by rising memory and system prices, which partially offset the growth.
02. Blackwell Ultra ramps up, Vera Rubin goes into full production
NVIDIA is promoting the upgrade of its product roadmap from a single GPU to a complete computing platform.
In the second fiscal quarter, Blackwell Ultra became an important factor driving the growth of the data center business. NVIDIA stated that the growth of data center revenue in the quarter mainly came from the large-scale deployment of Blackwell Ultra infrastructure. As cloud service providers and AI enterprises continue to build large-scale AI computing clusters, Blackwell is entering a broader commercial deployment stage.
At the same time, NVIDIA has begun to prepare for the next-generation product cycle. In the second fiscal quarter, the company announced that the Vera Rubin platform has been fully put into production, and related rack systems are running on cloud platforms of partners such as CoreWeave and Google Cloud.
The Rubin platform not only includes GPUs, but also covers CPUs, networks, software and system-level solutions. Among them, Vera CPU is the first CPU launched by NVIDIA designed for AI agents, and this product is planned to be adopted by the world's leading technology providers.
In addition, NVIDIA also announced that the NVIDIA Groq 3 LPX for interactive AI inference has been fully put into production. NVIDIA hopes to strengthen its competitiveness in real-time AI inference scenarios through products such as Groq 3 LPX.
In addition to hardware products, NVIDIA is also strengthening its software ecosystem.
In the second fiscal quarter, the company launched the DSX platform, providing infrastructure builders with a complete solution for designing, building and operating large-scale AI factories. The platform integrates computing, network, software and systems to help customers build larger-scale AI infrastructure.
In terms of AI software, NVIDIA continues to expand the NVIDIA Agent Toolkit and enhance development capabilities through PhysicsNeMo and CUDA-X libraries. The company stated that it is cooperating with global software platform providers to launch new software, open-source models and partner programs.
For NVIDIA, product competition is no longer limited to the performance of a single chip, but revolves around the complete ecosystem from chips, networks, software to systems. However, the market is still paying attention to the speed of transition from Blackwell to Rubin, and whether the new generation of platforms can continue to drive customers to expand investment in AI infrastructure.
03. AI infrastructure enters the financing stage, NVIDIA takes on more construction roles
As the scale of AI data centers continues to expand, NVIDIA is participating in more infrastructure construction.
The financial report for the second fiscal quarter shows that as of July 26, 2026, NVIDIA's total future commitment amount reached 3600 billion US dollars. This includes 2790 billion US dollars in supply and production capacity commitments, 290 billion US dollars in cloud service agreements, 230 billion US dollars in capital expenditures, and 250 billion US dollars in equity investments.
The above commitments are mainly related to the future expansion of AI infrastructure. Among them, the most concerned is the SB Energy PORTS-Pike project in Ohio that NVIDIA participates in.
NVIDIA stated that the company provides credit support for SB Energy's technology park in Ohio. The project initially involves about 4.25GW of land, power and plant construction, used to host OpenAI's NVIDIA infrastructure. The guarantee obligation provided by NVIDIA is up to 1050 billion US dollars, and will take effect in stages after conditions such as the data center reaching a serviceable state are met.
At the same time, NVIDIA is promoting larger-scale capital investment in AI infrastructure. The company has announced the establishment of strategic partnerships with institutions such as Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, with the goal of mobilizing more than 5000 billion US dollars in third-party capital in the future for AI infrastructure construction.
Jensen Huang believes that AI infrastructure construction is entering a new stage, and a large amount of capital investment will be needed in the future to support the development of AI model training, inference and more applications.
However, as NVIDIA participates in more infrastructure projects, the market has begun to pay attention to the capital pressure brought by this model.
As of the end of the second fiscal quarter, NVIDIA's cash, cash equivalents and marketable debt securities totaled 56.6 billion US dollars. At the same time, the company issued 25 billion US dollars of senior unsecured notes in the second fiscal quarter for general corporate purposes. NVIDIA stated that the company's financial position remains robust, and future investments will mainly focus on supporting supply chains, infrastructure and long-term growth needs.
04. Q3 revenue will exceed 100 billion US dollars, competitive pressure begins to appear
For the next quarter, NVIDIA has given expectations for continued growth.
The company expects revenue in the third quarter of fiscal 2027 to reach 108 billion US dollars, with a fluctuation of plus or minus 2%; it expects gross profit margins under both GAAP and Non-GAAP standards to be 74%. NVIDIA also emphasized that this guidance does not include any data center computing revenue from China.
Compared with the 96.2 billion US dollars revenue in the second quarter, NVIDIA expects to maintain growth in the third quarter. However, the market focus has shifted from single-quarter growth to the long-term competitive landscape.
At present, competition in the AI chip market is expanding. NVIDIA emphasized in its earnings report that the company is maintaining its advantages through a complete computing platform, including processors, interconnection technology, software, algorithms, systems and services. The company hopes to meet the needs of AI training and inference through this entire ecosystem.
But at the same time, competitors are accelerating their layout. AMD continues to launch data center products, and Google is also developing self-developed TPU chips. Large technology companies are not only important customers of NVIDIA, but also investing resources to develop their own computing platforms.
Investors are particularly concerned about the development of the AI inference market. As AI applications increase, computing demand is expanding from model training to inference services. NVIDIA strengthens its inference capabilities through the Vera Rubin platform, Groq 3 LPX and software ecosystem, hoping to continue to maintain its leading position in the market.
However, future growth still needs to answer several questions: whether Blackwell Ultra and Rubin platforms can continue to drive customers to increase procurement; whether AI infrastructure investment can maintain the current speed; and whether NVIDIA's share in the AI computing market will be affected as customers develop more self-developed chips.
Revenue of 96.2 billion US dollars is another milestone for NVIDIA in the AI wave.
For NVIDIA, the record-breaking financial data proves the success of the past cycle, and the challenge in the next stage is how to continue to maintain its core position after the AI industry enters a larger-scale construction stage.
This article is from the WeChat Official Account "Tencent Tech", written by Su Yang, and published by 36Kr with authorization.