Elon Musk has been generating buzz around the upcoming versions of his AI chatbot, Grok, developed by his company xAI. Musk recently hinted that Grok 3 will be a significant leap after it undergoes training using 100,000 Nvidia H100 GPU chips, a high-end component crucial for AI development. The estimated cost of these chips ranges from $3 billion to $4 billion, though it’s uncertain whether Musk’s company purchased them outright or is renting from cloud service providers. This ambitious step is a substantial increase from Grok 2, which required around 20,000 H100s for training. Musk’s plans come amid a broader trend of tech giants like Meta stockpiling these GPUs to enhance their AI capabilities, with Meta expected to have about 350,000 H100 GPUs by the end of 2024. The competition for AI talent has intensified, as exemplified by Meta’s substantial GPU collection influencing hiring decisions. While Grok 1 and Grok 1.5 are available to early testers, Grok 2 is set to launch in August, with Grok 3 planned for release by year’s end.

Elon Musk Teases Major Advancements for Grok 3 AI Chatbot
Elon Musk’s xAI aims to revolutionize AI with Grok 3, training on 100,000 GPUs.
1–2 minutes










