What it is
A foundation model is a large-scale artificial intelligence model, often a large language model (LLM) or a multimodal model, that has been extensively pre-trained on a massive dataset covering broad domains. This extensive pre-training allows the model to learn general patterns, representations, and capabilities that can then be fine-tuned or adapted for numerous specific tasks, rather than being built from scratch for each application. These models serve as a "foundation" upon which many other AI applications can be built.
Foundation models have become central to the AI industry, driving innovation across various sectors, from generative AI applications to enterprise solutions. Companies leverage foundation models to build custom AI tools, reducing development time and costs. The performance and capabilities of these models, often measured by their parameters and context window, influence the competitive landscape among AI developers and cloud providers. Their widespread adoption impacts demand for compute, model training, and fine-tuning services.
Why it matters
Foundation models are the building blocks for many modern AI applications, making them crucial for understanding the core technology and market trends in AI.
Reviewed under editorial standardsUpdated September 26, 2026Not investment advice