AMZN rumored to be slamming millions of dollars into training new AI model

$Amazon (AMZN.US)$ is investing millions of dollars to train an ambitious large-scale language model (LLM) that it hopes will rival top artificial intelligence (AI) models from OpenAI and $Alphabet-A (GOOGL.US)$ , people familiar with the matter said. The model, called Olympus, has 2 trillion parameters, which could make it one of the largest models being trained, people familiar with the matter said. By comparison, OpenAI's GPT-4, one of the best models available, reportedly has 1 trillion parameters.
The team is led by Rohit Prasad, former head of Amazon's intelligent voice assistant Alexa, who now reports directly to Amazon CEO Andy Jassy. As chief scientist for Amazon General Artificial Intelligence (AGI), Prasad brought in researchers who have been working on Alexa AI and Amazon's science team to work on training models that combine AI efforts across the company with dedicated resources.
LLM is the technology that underpins AI tools that learn from huge datasets to produce human-like responses. Training larger AI models costs more, given the computing power required. During an April earnings call, Amazon executives said the company would increase its investment in LLM and generative AI while cutting back on delivery and shipping operations in its retail division.