China Securities Co., Ltd.: Focus on Computex2026 and continue to recommend the computing power industry chain.

China Securities Co., Ltd. released a research report stating that the upcoming Computex 2026, with the theme of "AI Together," will see AI infrastructure evolving from chips to the construction of cabinets and systems. The value and importance of CPU, interconnect, liquid cooling, power supply, and other components continue to rise. The exhibition will also feature the first Siasun Robot&Automation zone, showcasing edge AI PCs, embodied intelligence, AI services from Siasun Robot&Automation, and more, demonstrating the comprehensive implementation of AI from chips to end applications. The prosperity of the computing power industry chain is being passed down from upstream to downstream whole machine manufacturers, with both Lenovo and Dell's latest financial reports exceeding market expectations. Both companies have AI orders exceeding 20 billion RMB, and the increase in the proportion of cabinet-level product shipments is driving the increase in the value of the whole machine manufacturers' industry chains, and also significantly improving their profitability. Computex 2026 is approaching, with AI hardware entering the era of racks and systems. From June 1st to 5th, COMPUTEX 2026 and GTC Taipei will be held in conjunction with the theme of "AI Together." NVIDIA CEO Huang Renxun is expected to deliver a keynote speech at 11:00 on June 1st. Additionally, leading CEOs from Qualcomm, ARM, Marvell, Intel, and others will also speak. The core theme of AI hardware at this exhibition has shifted from the previous focus on "buying GPUs" to "buying complete racks and accompanying systems," covering topics such as Vera Rubin GPU/Vera CPU/LPU, CPO/silicon photonics, 800V DC power, liquid cooling, edge AI, Siasun Robot&Automation, and embodied intelligence. Vera CPU is optimized for AI inference and intelligent scene scenarios, achieving 1.5 times the efficiency (single-core performance) compared to x86, 2 times the energy efficiency, and 4 times the rack density. Intel and NVIDIA will also release new self-developed SoC chips. Intel's Arc G3E is based on the Panther Lake architecture and 18A process, designed for Windows handheld devices to meet the demands of portability and high battery life. The chip was officially released on May 28th, with a live product demonstration expected at the exhibition. NVIDIA's first consumer-grade PC SoC chip, the N1X, will be officially released on the first day of the exhibition, marking NVIDIA's entry into the PC platform chip market. The chip, developed jointly with MediaTek, uses a 3nm process and Arm architecture, and is expected to be applied in the field of Arm gaming laptops, avoiding direct competition with x86 in mainstream business markets and instead focusing on AI performance and graphics performance in workstations and high-end gaming laptops. If NVIDIA can replicate its software and hardware dominance in data centers on the PC side, leveraging its CUDA ecosystem and AI computing power, it may shake the foundation of the Wintel alliance in the long run. This also suggests that the landscape of the PC chip market may evolve from a duopoly of Intel/AMD to a multi-polar landscape of x86+ARM. Looking at the supply chain and OEM levels, Dell and Lenovo have confirmed the start of whole machine research and development based on N1X, with the first products expected to be launched sequentially in the first half of 2026. If NVIDIA's PC SoC shipments can capture a market share in the millions in the AI PC market, it will bring clear incremental benefits to MediaTek (for chip design collaboration) and high-end product lines of brand OEMs. Marvell will focus on showcasing 1.6T ZR/ZR+ pluggable optical modules, 800G switch cluster solutions, CXL3.0 controllers, etc., at the exhibition, signifying a shift in computing capabilities from single-chip capabilities to interconnectivity, power supply, cooling, and delivery systems. Additionally, the Siasun Robot&Automation zone will be established for the first time at the Taipei exhibition, showcasing edge AI PCs, embodied intelligence, AI services from Siasun Robot&Automation, further demonstrating the comprehensive implementation of AI from chips to end applications. Lenovo and Dell's financial results exceeding expectations confirm the high prosperity of the computing power industry chain. Lenovo and Dell recently released their latest quarterly financial reports. Lenovo's FY2026 revenue reached $83.075 billion, a year-on-year increase of 20%, with a net profit attributable to shareholders of $1.912 billion, a year-on-year increase of 42%. In Q4 FY26, revenue reached $21.588 billion, a year-on-year increase of 27%, marking the highest growth rate in the past 5 years. AI-related revenue grew by 84% year-on-year, accounting for 38%, indicating that AI has evolved from an incremental business to a core variable in terms of profit and valuation. Lenovo's significant profit growth is largely driven by its ISG infrastructure solutions business, which returned to profitability in the latest quarter, with gross profit margin increasing from the mid-single-digit level to over 10%, and an operating profit margin of 3.6% for the quarter, driving profitability for the entire year. The improvement in Lenovo's ISG business profit margin is due to both an increase in revenue scale and the gradually starting shipments of high-margin cabinet products, as well as the acquisition of high-end storage manufacturer Infinidat, enhancing comprehensive solution capabilities, moving away from low-margin assembly business, with AI server order backlog exceeding $21 billion. On the Dell side, in the first quarter of fiscal year 2027, revenue was $43.842 billion, an 88% year-on-year increase, with a net profit of $3.438 billion, a 256% year-on-year increase, and Non-GAAP EPS of $4.86, a 214% year-on-year increase, significantly higher than market expectations of $2.94. AI server revenue reached $16.1 billion, a staggering 757% year-on-year increase, with AI orders of $24.4 billion and a backlog of $51.3 billion, reaching a historical high. Overall, Dell has raised its full-year AI server revenue forecast to about $60 billion, a 144% year-on-year increase, and its full-year revenue forecast has been revised upwards to $169 billion, the highest it has been. The excellent performance of Dell and Lenovo reflects the high prosperity of the computing power industry chain. The continued arms race in global cloud providers' AI infrastructure is being transmitted downward from upstream chip and storage sectors to OEMs/ODMs. At the same time, domestic computing networks are progressing steadily, with leading internet companies also advancing AI infrastructure construction, raising their full-year capital expenditure expectations. The high prosperity of AI infrastructure in both overseas and domestic markets is resonating. As capital expenditure for AI by North American cloud providers and global Neocloud customers enters an intensive delivery phase, AI servers have become the core growth engine for both companies, driving high-speed growth for Dell and Lenovo's ISG businesses. With the continuous penetration and high-end trend of AI PCs, Lenovo's high-end PC market share has reached 50%, with a rapidly increasing proportion of AI-related revenue. While Dell has not separately disclosed the proportion of AI PCs, the significant expansion of its operating profit margin is also benefiting from the upgrade of its product portfolio in high-end commercial and AI terminal products, driving a systematic increase in average selling prices and gross margins.