晚来一步的高通，能否在AI芯片领域逆袭英伟达？

Core Viewpoint - Qualcomm has launched a new AI chip series aimed at the lucrative AI data center market, intending to challenge Nvidia's dominance in this rapidly growing sector [2][3] Group 1: Product Launch and Market Strategy - Qualcomm's new AI200 chip series is set to ship next year, with its first customer being Saudi AI startup Humain, which plans to deploy systems with a total power of 200 megawatts by 2026 [2] - The AI200 products will be available in various forms, including standalone chips, expansion cards for existing devices, and complete rack servers, allowing Qualcomm to compete directly with Nvidia and other chip manufacturers [3] - The new chips are built on neural processing units (NPU), originally used in smartphones, designed to accelerate AI tasks without significant power consumption [3][4] Group 2: Financial Implications and Market Position - Analysts suggest that even a small market share in the over $500 billion AI accelerator market could generate billions in new revenue for Qualcomm [3] - Qualcomm's stock rose 11% to $187.68, marking its highest price since July 2024 and the largest single-day gain since April of this year [2] - The company aims to reduce its reliance on smartphone sales, which have slowed, by expanding into automotive and personal computer chips, with this being its first product targeting the data center market [4] Group 3: Competitive Landscape - Nvidia's data center revenue is projected to exceed $180 billion this year, significantly outpacing the total revenue of all chip manufacturers, including Qualcomm [4] - Qualcomm's new chips will feature unprecedented memory capacity, with low-power dynamic random-access memory (LPDDR) reaching up to 768GB, which is crucial for AI processing [4][5] - The growth of AI business is expected to help Qualcomm offset revenue losses from Apple, which has contributed about 20% of Qualcomm's revenue in recent years but is now shifting towards in-house chip development [5]