At the base of this stack is the infrastructure layer, consisting of semiconductors and associated hardware and software necessary for AI computing, such as accelerated processors, memory, storage, networking and packaging, and the sophisticated machinery and software required to design and produce it all. Nvidia's chips and software are key to this layer, but so too are the offerings of numerous other companies such as Broadcom, Cadence, and ASML.
On top of this infrastructure sits the AI model layer. This includes foundational Large Language Models (LLMs) as well as domain-specific models that are tailored to particular industries or use cases such as pharmaceuticals. LLMs are currently led by the large technology companies such as Google, Meta and Microsoft as well as newer players like OpenAI and Anthropic.
Next up are AI engineering tools that enable enterprises across the whole economy to operationalise LLMs to harness AI. Companies such as ServiceNow, which is helping businesses deploy generative AI across their operations, are particularly well positioned, as are companies like Snowflake and MongoDB which help businesses organise and use their data for AI-driven tasks.
Finally, at the top of the stack is the applications layer, where AI meets its customers to drive prices down and up the quality of products and services, both for business-to-business and business-to-consumer. This is currently led by the likes of Spotify and Netflix through recommender systems, though it is also underpinning a range of other leading consumer-facing businesses such as Uber and Airbnb.
