07 December 2023
If you are covering AMD’s latest announcement about its new chip, please see the following comment from Ben Barringer, tech analyst at Quilter Cheviot:
"AMD last night announced the launch of its new MI300 chip, a significant development in the world of AI. This chip, featuring an impressive count of 150 billion transistors, positions AMD as a formidable contender in the AI and machine learning industry.
"A notable aspect of the MI300 is its dual capability in both training and inference. While its performance in training is comparable to NVIDIA’s best chips, it reportedly surpasses them in inference tasks. Inference, the process of using trained AI models, is crucial for the practical application of AI technologies and could easily be a bigger market than training chips.
"The event also highlighted AMD's collaboration with major tech giants such as Microsoft, Meta, and Oracle to name a few on stage during the announcement. This indicates a broad industry support and potential for future partnerships leveraging AMD's new technology.
"However, AMD is also actively enhancing its operating systems and networking capabilities, suggesting a comprehensive approach to AI solutions. This move is seen as a direct challenge to NVIDIA, not only through superior hardware but also through a robust open sourced ecosystem of software and networking solutions.
"The introduction of the MI300 chip is particularly significant for major cloud service providers as it offers them an alternative to NVIDIA, potentially reducing their dependency on a single supplier and diversifying the AI technology landscape.
"AMD did not announce any changes in its financial forecasts during this event. However, it did adjust its market forecasts, projecting a growth to $400 billion by 2027, a substantial increase from the previous estimate of $150 billion. This revised forecast, reflecting a 70% growth rate over four years, underscores AMD’s confidence in the expanding AI market.
"Furthermore, AMD is venturing into the development of AI chips for personal computers, marking a foray into edge AI. Edge AI involves processing part of the AI workload on local devices like PCs or smartphones, enabling faster and more efficient data processing. This move could significantly impact how AI is integrated into everyday technology, making it more accessible and widespread."
