Core Insights - Meta Platforms plans to spend between $115 billion and $135 billion on capital expenditures in 2023, marking an 87% year-over-year increase at the high end of its guidance [1] Group 1: Meta's AI Investments - The majority of Meta's capital expenditure will focus on its artificial intelligence buildout, which includes optimizing data centers for efficiency [1][3] - Meta's AI initiatives include the Llama large language model and algorithms that determine content delivery to over 3.5 billion monthly active users [3] Group 2: Partnerships with Chipmakers - Meta has signed multi-billion dollar contracts with Nvidia and AMD to utilize their GPUs and CPUs in its data centers, benefiting both companies significantly [2][4][5] - The deal with Nvidia involves co-designing next-generation AI models optimized for Nvidia's hardware, focusing on data centers for LLM training [4] - The partnership with AMD will deploy MI450 GPUs and 6th-generation EPYC CPUs, aimed at enhancing inference performance [5][6] Group 3: AMD's Competitive Position - AMD's MI450 GPUs are expected to outperform Nvidia's Rubin platform in terms of price performance due to advanced architecture and high-bandwidth memory [6] - Analysts predict that AI inference will dominate AI compute starting this year, positioning AMD favorably with its long-term deal with Meta [7]