Meta to double AI investment in 2026

- Meta's capital expenditure on AI is projected to reach between $115 billion and $135 billion in 2026, a significant increase from the roughly $72.2 billion spent in 2025. A large portion of this investment will go towards data centers, servers, and networking equipment. - To secure its AI hardware supply chain, Meta has entered into multi-year, multi-billion dollar agreements with both AMD and Nvidia. The deal with AMD is valued at over $100 billion for up to 6 gigawatts of GPU capacity and includes a performance-based warrant for Meta to acquire up to 10% of AMD's shares. - For ML system design interviews, candidates are expected to architect end-to-end pipelines for problems like news feed ranking or content safety systems. Interviewers assess the ability to handle large-scale data ingestion, discuss trade-offs between latency and accuracy, and design for reliability and privacy. - Common data structures and algorithms (DSA) questions for ML engineering roles involve topics like graph traversal to detect arbitrage, array manipulations such as moving zeros, and understanding the trade-offs of different sorting algorithms. While fundamental, DSA is often a filtering stage in the interview process. - Meta's open-source strategy is central to its AI development, with models like LLaMA 3 and frameworks such as PyTorch being made publicly available to foster a broader ecosystem and accelerate innovation. This approach allows developers to build upon Meta's technology for a wide range of applications. - To build a strong portfolio, aspiring ML engineers should create projects that utilize modern MLOps tools for automation, monitoring, and deployment. Familiarity with tools like Kubeflow for pipeline orchestration, MLflow for experiment tracking, and containerization with Docker are key skills. - Top tech companies seek T-shaped engineers who possess both broad software engineering skills and deep expertise in modern AI. This includes proficiency in Python, a strong foundation in mathematics and statistics, and experience with deep learning architectures. - The company's Fundamental AI Research (FAIR) team is focused on advancing machine intelligence with a focus on AI perception. Recent projects include a Perception Encoder for image and video tasks and a Dynamic Byte Latent Transformer, which operates at the byte level to improve efficiency and robustness over traditional token-based models.