1. Intel Intensifies Competition
Last week, Intel announced its plans to launch the Gaudi3 chip in 2024, directly challenging Nvidia’s H100, which currently dominates the AI app market for server farms. Intel’s Gaudi3 chip offers superior performance, competitive pricing, and reduced total cost of ownership, aiming to increase its share in the AI accelerator market next year. Additionally, Intel is set to release the Core Ultra chip, designed for AI app usage on personal computers. According to CEO Pat Gelsinger, the AI PC is expected to be the highlight of 2024.
2. Google’s Gemini and Infinity TPU
Alphabet’s recent unveiling of the Gemini large language models (LLM) drew much-deserved attention. With Gemini Ultra surpassing current AI models on 30 of 32 benchmarks, including outperforming human experts on the MMLU test, this new LLM model showcases remarkable advancements.
Despite this, the introduction of Google’s tensor processing unit (TPU) chip, specifically designed for training AI models, received less recognition. Google’s Cloud TPU v5p, used to train Gemini, signifies a significant leap in their AI chip technology.
Google’s partnership with Nvidia remains strong, but recent developments indicate that Google is diversifying its offerings, demonstrating a desire for a less reliant future on Nvidia.
While Nvidia’s GPUs are currently the go-to choice for AI apps, challenges loom on the horizon.
Investors have high expectations for Nvidia’s sustained growth. However, any disruptions to this growth may impact Nvidia’s stock price. Intel’s Gaudi3 and Google’s new chip could attenuate the demand for Nvidia’s GPUs to some degree.
Although Nvidia’s momentum is unlikely to be entirely derailed by these developments, a tougher year may lie ahead in 2024. The stock’s remarkable tripling and soaring profits of 2023 may be difficult to replicate.