Normal Computing builds probabilistic computer chips designed for AI workloads that require handling uncertainty, such as generative models, diffusion processes, and Bayesian inference. Their hardware natively computes with probability distributions rather than deterministic bits, enabling faster and more energy-efficient execution of probabilistic algorithms compared to software simulations on GPUs. Customers include AI researchers, developers building uncertainty-aware models, and enterprises in fields like drug discovery and climate modeling. The company has raised $3.5 million in seed funding from investors including Buckley Ventures, NFDG, and Y Combinator. Founders Antonio Martinez and his co-founders draw from prior work at MIT CSAIL on probabilistic computing architectures. The San Francisco-based startup unveiled its first chip prototype in 2024. Key team members previously worked at Google DeepMind, NVIDIA, and Stanford.