Core Viewpoint - The emergence of DeepSeek's low-cost AI models has led to significant declines in the stock prices of major U.S. semiconductor companies and tech giants, raising questions about their spending on AI development and infrastructure [1][5][6]. Group 1: Impact on U.S. Tech Giants - NVIDIA's stock fell approximately 17%, equating to a loss of around 5.6 million to train using fewer than 2,050 GPUs, in stark contrast to Meta's use of 16,000 NVIDIA H100 chips for its Llama 3 model [4]. Group 3: Spending Trends and Future Outlook - U.S. tech giants are projected to spend around 3 billion on model training in 2024 [6]. - Meta is planning capital expenditures between 65 billion for AI in 2025, indicating a trend of increasing investment despite the competitive pressure from low-cost alternatives [6]. - The Stargate joint venture, involving OpenAI, SoftBank, and Oracle, aims to invest $500 billion over the next four years in new AI infrastructure [7]. Group 4: Semiconductor Companies' Prospects - Amphenol, Broadcom, and NVIDIA are identified as key semiconductor stocks poised to benefit from the ongoing deployment of AI and Generative AI technologies [8]. - Amphenol's diversified business model and strong demand across various sectors, including defense and commercial air, are expected to drive growth [10]. - Broadcom's strong demand for networking products and custom AI accelerators, along with its acquisition of VMware, positions it well for future growth [12]. - NVIDIA is anticipated to benefit from increased usage of its GPUs, with a focus on the Hopper and Ampere architectures, despite the competitive landscape introduced by DeepSeek [15].
3 Top-Ranked Semiconductor Stocks to Buy Despite DeepSeek Disruption