Introduction
Once generative artificial intelligence (GenAI) grabbed the world’s attention, in 2023, a technology arms race ensued, with the leading purveyors of large language models (LLMs) outdueling each other to deliver ever larger and more capable models. Innovative businesses jumped on the GenAI bandwagon, experimenting with copilots and plugging into foundation models on the leading public clouds to build custom GenAI capabilities. What many businesses soon learned is that GenAI development costs can be steep and the road to successful production is fraught with pitfalls, roadblocks, and security risks.
As 2024 unfolded, model options grew more numerous, and innovators soon learned that open source models; midsize models; and even small, purpose-built models could often outperform LLMs at a much lower cost. What’s more, businesses learned that high-cost graphical processing unit (GPU) capacity isn’t always necessary but that in the most demanding circumstances, it can be more economical than conventional central processing unit (CPU) capacity, because it whips through GenAI tasks much more quickly.
Building on these learnings, Teradata, in early October 2024, announced two key offerings: ClearScape Analytics Bring Your Own LLM (BYO-LLM) in Teradata VantageCloud Lake and Teradata VantageCloud Lake GPU-Accelerated Compute. This report details these two offerings and related GenAI-based Teradata solutions supporting three use cases: understanding customer complaints, helping banks ensure regulatory compliance, and helping healthcare providers analyze doctors’ notes. Early adopters should use this report to gain a better understanding of Teradata’s emerging offerings that support GenAI innovation.
