AI Basics Part 4: The Artificial Intelligence Infrastructure and the hardware needed
What infrastructure and hardware are needed for AI? In this fourth of four presentations about Artificial Intelligence and how it can be used in practice, Pere Sivecas, Spend Matters Development PM and Software Architect, examines the practical needs involved in rolling out a large language model (LLM). Beginning with a short history about how 3D gaming and Ethereum brought new interest in using Graphics Processing Units (GPUs) for AI, he shows the current GPU market landscape. From there, he examines differences in the needed GPU infrastructure, options, costs, advantages and disadvantages between having an in-house LLM and turning to an AI-as-a-service provider.