Core Insights - NVIDIA's GTC 2026 showcased a significant shift in the AI industry, with CEO Jensen Huang emphasizing the company's projected revenue of at least $1 trillion by 2027, up from $500 billion last year [4][12][36] Group 1: Event Overview - The GTC 2026 featured 450 sponsoring companies, 1,000 technical sessions, 2,000 speakers, and 110 robots, resembling an annual pilgrimage for the AI industry rather than a typical tech conference [1] - Huang's keynote included a historical overview of NVIDIA's technology evolution, from the GeForce graphics card to the current AI advancements [4][36] Group 2: Token Economy - Huang introduced a comprehensive business model based on token economics, highlighting different pricing tiers for token usage, ranging from free access to $150 per million tokens for advanced tasks [9][12] - The model emphasizes the importance of token throughput and generation rates, which are critical for AI performance [9][12] Group 3: Technological Advancements - The new Vera Rubin AI computing system is described as NVIDIA's most complex, featuring seven types of chips and achieving 3.6 exaflops of computing power [15][20] - Innovations include a 100% liquid cooling solution and the introduction of Co-Packaged Optics (CPO) technology, which enhances data transmission efficiency [21][24][25] Group 4: Integration of Groq - NVIDIA's acquisition of Groq aims to enhance AI inference capabilities, with Groq's architecture designed for high-speed processing, complementing NVIDIA's GPU offerings [29][31] - The integration strategy involves separating inference tasks between Groq and Vera Rubin to optimize performance [33][35] Group 5: Future Developments - Huang announced the upcoming Feynman architecture, which will introduce new GPUs, LPUs, and CPUs, emphasizing the need for higher capacity and bandwidth in future computing solutions [44][47] - The company is also exploring the concept of "space computing" with plans for a data center in space, addressing unique challenges such as heat dissipation in a zero-gravity environment [49][50]

Core Insights - The report discusses the rapid development of the supernode and Scale-up network industry, highlighting the competitive landscape involving Nvidia, Google, AMD, and Huawei, which is challenging Nvidia's dominance in the market [2][22]. Group 1: Nvidia - Nvidia maintains a leading advantage in supernode technology through NVLink and NVLink Switch, with plans to release advanced solutions like GH200 NVL72 and GB200/GB300 NVL72 between 2024 and 2025, expecting a shipment of approximately 2,800 units by 2025 [4][40]. - The NVLink architecture is designed for high bandwidth and low latency data transmission, with NVLink 5 Switch achieving a single GPU-to-GPU bandwidth of 1,800 GB/s and a total bandwidth of 130 TB/s for 72 GPUs by 2025 [5][40]. - Future developments include the introduction of NVSwitch Gen6 and Gen7, which will further enhance GPU-to-GPU communication bandwidth to 3.6 TB/s [5]. Group 2: Huawei - Huawei is working on the Lingqu protocol, which is transitioning to an open standard, although it has not yet gained widespread acceptance in the domestic industry [6]. - The Atlas 950 supernode, expected to launch in Q4 2026, will feature a total computing power of 8 EFLOPS (FP8) and a memory capacity of 1,152 TB, significantly surpassing Nvidia's offerings [7]. - Huawei's approach involves a hybrid design of copper and optical interconnects to balance complexity, reliability, and power consumption while maintaining system scalability [7]. Group 3: Google - Google has established a mature optical interconnect supernode with its TPU series, including TPU v4, TPU v5p, and TPU v7, which are set to be released between 2023 and 2025 [8]. - The TPU v7 will be utilized by Anthropic, which plans to procure nearly 1 million TPU v7 Ironwood AI chips for deployment in its data centers [8]. - Google's competitive edge lies in its unique application of optical circuit switches (OCS) in Scale-up networks, creating a significant technological barrier against competitors [9]. Group 4: AMD - AMD's UALink has emerged as an important open standard, with the first version released in January 2025 and a second version expected in 2026, gaining widespread industry support [10]. - The Helios supernode is positioned as a strong competitor to Nvidia's NVL72 series, featuring a dual-width rack design that allows for future scalability without redesigning infrastructure [10]. - The Helios rack is anticipated to become a mainstream choice in the industry, with significant advantages in power consumption compared to Nvidia's offerings [10].

Investment Rating - The report maintains a "Positive" outlook on the supernode and Scale-up network industry, highlighting its rapid development and potential as a key infrastructure for AI applications [2]. Core Insights - The supernode and Scale-up network are critical infrastructures that break through computing and communication bottlenecks, supporting trillion-level large models and high real-time applications. The report analyzes the progress and advantages of leading AI computing chip manufacturers, including NVIDIA, Google, AMD, and Huawei, in this field [4][24]. Summary by Sections 1. NVIDIA - NVIDIA's leading advantage in supernode technology is based on NVLink and NVLink Switch. The company plans to launch several mature supernode solutions, including GH200 NVL72 and GB200/GB300 NVL72, with an expected shipment of approximately 2,800 units by 2025 [5][6]. - The NVLink technology enables high bandwidth and low latency data transmission, with NVLink 5 Switch supporting a single GPU-to-GPU bandwidth of 1,800 GB/s and a total bandwidth of 130 TB/s for 72 GPUs [6][40]. - Future developments include the introduction of the Vera Rubin NVL144 and Rubin Ultra NVL576, which will increase the number of interconnected GPUs from 72 to 576 [5][6]. 2. Huawei - Huawei has introduced the Lingqu protocol, transitioning to an open standard from version 2.0, although it has not yet gained widespread acceptance in the domestic industry. The company aims to catch up with NVIDIA in supernode performance through a clustered approach [7][8]. - The Atlas 950 supernode, expected to be released in Q4 2026, will have a total computing power of 8 EFLOPS (FP8) and a memory capacity of 1,152 TB, significantly surpassing NVIDIA's offerings [7][8]. 3. Google - Google has established a mature optical interconnect supernode with its TPU series, including TPU v4, TPU v5p, and TPU v7, which have been recognized by external enterprises [9][10]. - The competitive advantage of Google's TPU supernode lies in its unique application of optical circuit switches (OCS), which creates a high barrier to entry in the optical interconnect field [9][10]. 4. AMD - AMD's UALink has become an important open standard, with the 1.0 version released in January 2025 and the 2.0 version expected in 2026. The UALink ecosystem is anticipated to see significant development by 2027, with over 100 member units supporting it [11]. - The Helios rack from AMD is positioned as a strong competitor to NVIDIA's NVL72 series, featuring a dual-width design that balances complexity, reliability, and performance [11]. 5. Investment Strategy - The report suggests a positive outlook for Google, AMD, and domestic supernode manufacturers, as well as for NVIDIA's supply chain, including PCB backplanes, high-speed copper cables, optical modules, and cooling systems [13][14]. - The market is expected to continue reassessing the value of Google, AMD, and domestic supernode sectors as competition intensifies [13].

Core Viewpoint - Starting from 2025, supernodes will become a significant technological innovation direction in the AI computing network, with increasing competition among AI chip manufacturers in both chip performance and Scale up network [1][5]. Group 1: Supernode Development - Nvidia has launched mature supernode solutions, with plans to release GH200 NVL72, GB200/GB300 NVL72, and VR200 NVL72 from 2024 to 2026 [1][3]. - The Blackwell architecture standardizes Scale up with GB200 NVL72 stabilizing the scale at 72 GPUs per cabinet, consisting of 18 Compute Trays and 9 Switch Trays [2]. - The Rubin architecture will enhance bandwidth, with the NVLink 6 Switch increasing single GPU interconnect bandwidth to 3.6 TB/s, up from 1.8 TB/s [2]. Group 2: Nvidia's Competitive Advantage - Nvidia maintains a leading position in the supernode market, with a projected shipment of approximately 2,800 units of GB200/300 NVL72 by 2025 [3]. - Future plans include the introduction of Vera Rubin NVL144 and Rubin Ultra NVL576, expanding interconnected GPUs from 72 to 576 [3]. - Innovations such as NVLink and NVLink Switch are crucial for achieving high bandwidth and low latency in AI training clusters, with NVLink 5 Switch supporting a total bandwidth of 130 TB/s for 72 GPUs [4]. Group 3: Industry Landscape and Investment Strategy - The global supernode competition landscape is still being established, with Nvidia currently in a leading position [6]. - The report suggests monitoring Nvidia's supernode supply chain, including components like PCB backplanes, high-speed copper cables, optical modules, and cooling systems [6]. - Chinese manufacturers are actively participating in the supernode and Scale up network sectors, with potential for domestic firms to gain a competitive edge [6].

Investment Rating - The report maintains a "Positive" outlook on the communication industry [2] Core Insights - The evolution of large language model (LLM) parameters from hundreds of billions to trillions and even hundreds of trillions necessitates tensor parallelism (TP) across servers, making the development of high-bandwidth, low-latency Scale up networks a mainstream technical path in the industry [4][18] - NVIDIA is positioned as a leader in the supernode space, with plans to launch multiple generations of supernodes from 2024 to 2026, including GH200 NVL72, GB200/GB300 NVL72, and VR200 NVL72 [5][43] - The advantages of NVIDIA's supernodes are built on NVLink and NVLink Switch technologies, which support high bandwidth and low latency data transmission essential for AI training clusters [6][86] Summary by Sections 1. High Bandwidth and Low Latency Requirements - The training of LLMs requires extremely high bandwidth and low latency, driving the innovation of supernodes as a key direction in AI computing networks [18] - The need for cross-server tensor parallelism (TP) and expert parallelism (EP) has led to the establishment of Scale up networks [8] 2. NVIDIA's Leading Advantage - NVIDIA's supernode solutions are based on NVLink and NVLink Switch, which have evolved from point-to-point connections to full interconnect communication [33] - The sixth generation of NVLink and NVLink Switch supports GPU-to-GPU communication bandwidth of 3.6TB/s, with total aggregated bandwidth of 260TB/s in the VR NVL72 system [33][75] 3. Supernode Specifications - The GB200 NVL72 supernode features 180 PFLOPS of TF32 Tensor Core computing power, 13.8TB of memory, and a memory bandwidth of 576TB/s, with a total exchange capacity of 129.6TB/s [47][48] - The VR200 NVL72 supernode, set to be released in 2026, will double the total exchange capacity to 259.2TB/s compared to the GB200 NVL72 [70][75] 4. Investment Strategy - Starting from 2025, supernodes will become a significant innovation direction in AI computing networks, with various global manufacturers entering the competition [9] - NVIDIA currently holds a leading position, and attention should be paid to its supernode supply chain, including PCB backplanes, high-speed copper cables, optical modules, and cooling systems [9]

Investment Rating - The investment rating for NVIDIA is "Buy" with a 6-month outlook [6]. Core Insights - NVIDIA reported record revenue of $39.3 billion for Q4 FY2024, a 78% year-over-year increase and a 12% quarter-over-quarter increase, surpassing Bloomberg consensus estimates of $38.2 billion [1]. - The company's data center business achieved a record revenue of $35.6 billion in Q4, exceeding Bloomberg consensus estimates of $34.1 billion, driven by strong sales of the Blackwell architecture [2]. - NVIDIA anticipates continued growth in computing power demand, supported by advancements in multi-modal models and inference optimization, with Blackwell architecture showing a 25-fold improvement in inference throughput compared to Hopper [4]. Revenue and Profit Dynamics - GAAP gross margin was reported at 73%, in line with Bloomberg consensus, while net profit reached $22.1 billion, exceeding expectations of $19.8 billion, resulting in a GAAP net profit margin of 56.2% [1]. - The company expects gross margins to recover to 75% by the end of FY2026 as production capacity ramps up [3]. Business Segment Performance - The networking segment is expected to see growth in the next quarter, with significant revenue increases from Spectrum X and NVLink Switch, as major cloud service providers build large AI infrastructures [5]. - Enterprise AI revenue nearly doubled year-over-year, driven by demand for model fine-tuning and GPU-accelerated data processing [5]. Future Projections - Revenue forecasts for FY2026-2028 have been raised to $228 billion, $335 billion, and $374.2 billion respectively, with GAAP net profit projections also increased to $124.9 billion, $175 billion, and $200 billion [6].

Investment Rating - The investment rating for NVIDIA is "Buy" with a target of over 20% relative return within the next six months [6]. Core Insights - NVIDIA reported record revenue of $39.3 billion for Q4 FY2024, a year-over-year increase of 78% and a quarter-over-quarter increase of 12%, surpassing Bloomberg consensus estimates of $38.2 billion [1]. - The data center business achieved a record revenue of $35.6 billion, exceeding Bloomberg consensus estimates of $34.1 billion, driven by strong sales of the Blackwell architecture [2]. - The company anticipates continued growth in computing power demand, supported by advancements in multi-modal models and inference optimization with the Blackwell architecture [4]. Revenue and Profit Performance - For Q4 FY2024, NVIDIA's GAAP gross margin was 73%, aligning with Bloomberg consensus expectations, while net profit reached $22.1 billion, exceeding expectations of $19.8 billion [1]. - GAAP earnings per share (EPS) for the quarter was $0.89, reflecting an 82% year-over-year increase and a 14% quarter-over-quarter increase [1]. Product and Capacity Dynamics - The initial production of Blackwell has led to a Non-GAAP gross margin of 73.5%, with expectations to recover to 75% by the end of FY2026 as production capacity ramps up [3]. - Blackwell's quarterly production capacity reached 15,000 GPUs, deployed across 350 manufacturing nodes, with major cloud service providers already utilizing Blackwell clusters [3]. Future Projections - Revenue forecasts for FY2026-2028 have been raised to $228 billion, $335 billion, and $374.2 billion respectively, with GAAP net profit projections adjusted to $124.9 billion, $175 billion, and $200 billion [6]. - The company expects the demand for computing power to continue to rise, driven by advancements in AI and cloud computing [4]. Business Segment Growth - The networking segment is expected to recover growth in the next quarter, with significant revenue increases from Spectrum X and NVLink Switch, which are being adopted by major cloud service providers [5]. - Enterprise AI revenue has nearly doubled year-over-year, driven by increasing demand for model fine-tuning and GPU-accelerated data processing [5].

Investment Rating - The investment rating for NVIDIA is "Buy" with a target of over 20% relative return within the next six months [6]. Core Insights - NVIDIA reported record revenue of $39.3 billion for Q4 FY2024, a 78% year-over-year increase and a 12% quarter-over-quarter increase, surpassing Bloomberg consensus estimates of $38.2 billion [1]. - The data center business achieved a record revenue of $35.6 billion, exceeding Bloomberg consensus estimates of $34.1 billion, driven by strong sales of the Blackwell architecture [2]. - The company anticipates continued growth in computing power demand, supported by advancements in multi-modal models and inference optimization, with Blackwell architecture showing a 25-fold improvement in inference throughput compared to the previous generation [4]. Revenue and Profit Performance - NVIDIA's GAAP gross margin was 73%, in line with Bloomberg consensus, while net profit reached $22.1 billion, exceeding expectations of $19.8 billion, resulting in a GAAP net profit margin of 56.2% [1]. - The GAAP earnings per share (EPS) was $0.89, reflecting an 82% year-over-year increase and a 14% quarter-over-quarter increase [1]. Business Segment Highlights - The Blackwell architecture's initial production capacity reached 15,000 GPUs, with deployments across major cloud service providers like Azure, AWS, and GCP [3]. - The networking segment is expected to recover growth in the next quarter, with significant revenue increases from Spectrum X and NVLink Switch, as cloud service providers build large AI infrastructures [5]. - Enterprise AI revenue nearly doubled year-over-year, driven by demand for model fine-tuning and GPU-accelerated data processing [5]. Future Projections - Revenue forecasts for FY2026-2028 have been raised to $228 billion, $335 billion, and $374.2 billion respectively, with GAAP net profit projections of $124.9 billion, $175 billion, and $200 billion [6].