The meeting welcomed the financial report of NVIDIA for the second quarter of the fiscal year 2026, discussing future financial forecasts and non-GAAP financial indicators, while emphasizing the risks and uncertainties of forward-looking statements, as well as compliance with financial reporting requirements and disclosure obligations.

NVIDIA's revenue in the second quarter reached $46.7 billion, a year-on-year increase of 56%. The Blackwell platform has performed outstandingly, leading the AI infrastructure market. The company expects AI infrastructure investments to reach $3-4 trillion in the next decade. New products such as GB 300 are accelerating the construction of global data centers, while also facing geopolitical challenges, but continue to drive the establishment of global AI technology standards.

NVIDIA's Spectrum X enhanced Ethernet solution provides the highest throughput and lowest latency in AI workloads, achieving double-digit quarterly and annual growth rates. InfiniBand revenue almost doubled, benefiting from the adoption of XDR technology, significantly increasing bandwidth. NVIDIA's NVLink technology has doubled the speed of communication between GPUs, and its robotics computing platform NVIDIA AGX Orin provides higher artificial intelligence performance and efficiency. The NVIDIA Omniverse platform is driving AI development in automated factories, collaborating with multiple European robotics companies to accelerate the development of robotic systems.

NVIDIA's gaming business revenue reached 4.3 billion US dollars, a year-on-year increase of 49%, benefiting from strong sales of the Blackwell GPU. The company launched the RTX 5060, improving performance and AI capabilities, and plans to upgrade GeForce Now to the level of RTX 5080 in September. In terms of AI applications, NVIDIA collaborated with OpenAI to optimize the GPT model, professional visualization revenue increased by 32%, and automotive computing revenue surged by 69%, thanks to the introduction of the NVIDIA Thor SOC. The company continues to invest in growth, repurchasing stocks and distributing cash dividends totaling 10 billion US dollars in the second quarter, with the board of directors approving an additional 60 billion US dollars for repurchasing authorization.

Looking ahead to the third quarter of the 2026 fiscal year, it is expected that total revenue will reach 54 billion, with significant growth, not including shipments from H20 Chinese customers. Emphasis is placed on AI, especially breakthroughs in reasoning and agent AI, driving growth in data centers and networks. The Blackwell MV Link 72 system increases computing efficiency, with plans to expand into the 3-4 trillion US dollar AI infrastructure market over the next five years, in response to industry changes.

Discussed sustainable growth strategies for the Chinese business in the fourth quarter, as well as the competitive environment facing ASIC projects and AI market trends. Emphasized strengths in the complex accelerated computing field, including platform versatility, architectural adaptability, and system-level solutions, to maintain market leadership.

The dialogue emphasized the technical complexity required to build high-performance AI platforms, including the integration of efficient GPUs, CPUs, memory systems, and network architectures, aimed at achieving large-scale AI computing. NVIDIA, with its advantages in energy efficiency, performance, and cost-effectiveness, has become the preferred choice for cloud service providers and startups. Its solutions cover the full stack from chips to AI super factories, demonstrating strong market competitiveness and growth potential.

The conversation revolved around the huge investments in data center infrastructure in the next ten years, mentioning that investment scale will soar from billions to three to four trillion US dollars. It focused on analyzing the driving role of the AI revolution on capital expenditure, as well as NVIDIA's industry contribution as an AI infrastructure provider. The discussion emphasized potential bottlenecks such as power supply, as well as the strategic importance of improving efficiency and reducing costs through performance enhancements.

The dialogue discussed NVIDIA's long-term development prospects in the Chinese market, especially the importance of black box architecture authorization. It is expected that the AI software market in China will reach $50 billion this year with an annual growth rate of about 50%. China has approximately 50% of the world's AI researchers and numerous leading open source models, which have a significant impact on global AI applications. Open source models promote the custom development of AI systems for global enterprises, which is particularly critical in the SaaS and robotics fields. American companies need to maintain their competitiveness in the Chinese market in order to drive American technology stack to become the global standard. H20 has been approved by some enterprises, and the entry of black box architecture into the Chinese market is possible, but it is necessary to continue to promote the rationality and importance of relevant policies.

The conversation focuses on the market potential of Spectrum XGS, discusses its applications in data transmission and networking layers, as well as the possibility of achieving a $10 billion annual revenue scale in the Ethernet product line, evaluating its market opportunities and growth prospects.

Three types of network technologies were discussed, including NVLink and InfiniBand for scaling out, emphasizing the revolutionary role of NVLink 72 in building large virtual GPUs and the low latency advantage of InfiniBand in supercomputing and model development.

The dialogue discussed the importance of network technology in AI factories, especially the low latency and low jitter characteristics of Infiniband and the new Spectrum Ethernet (including Spectrum X and XGS), and their significant contribution to efficiency improvement. It was pointed out that the correct network selection can significantly enhance factory performance, bringing billions of dollars in benefits, emphasizing NVIDIA's investment and achievements in this field.

Discussed the contributions of Blackwell and Hopper to the growth in data center revenue, as well as the synergistic effects of networking equipment, emphasizing that Blackwell will continue to lead the growth, while Hopper and networking equipment will maintain stable sales.

Discussed the impact of performance improvements in AI on the efficiency of data centers, emphasizing the annual cycle of product upgrades, and how increasing performance per watt and reducing costs can increase customer revenue and profit margins. Looking ahead, AI capabilities and the economic benefits of data centers will be further enhanced through continuous technological innovation and architectural optimization.

The discussion explored a 50% compound annual growth rate in the AI market, as well as the impact on the growth of data center revenue. It mentioned a significant increase in funding for AI-native startups, as well as the high interest and demand for AI from large enterprises, SaaS, and industrial AI. It emphasized the crucial role of Blackstone and Rubin platforms in AI factory construction, as well as the extensive prospects for the future application of AI technology in enterprise services, industrial automation, and robotics.