- Elon Musk’s xAI plans to launch Grok-2 in August and Grok-3 by year-end.
- Grok-3 training will use 100,000 Nvidia H100 chips, costing $3-$4 billion.
- xAI may spend $10 billion renting cloud servers from Oracle.
Elon Musk’s xAI Chatbots: Launch Timeline and Nvidia Chip Requirements
AI company, xAI by Elon Musk is developing new iterations of its AI chatbot, Grok, similar to OpenAI’s ChatGPT and Google Gemini. Recently, Musk revealed the launch timeline for Grok-2 and Grok-3, as well as the number of Nvidia chips needed to train them.

Grok-2 Launching in August, Grok-3 by Year-End
Musk announced on social media platform X that Grok-2 will launch in August and will not be trained on data from other AI models, addressing concerns about AI models being trained on each other’s data. “Sadly quite true. It takes a lot of work to purge LLMs from the Internet training data. Grok 2, which comes out in August, will be a giant improvement in this regard,” Musk stated.
"Models training on each other's data is like a human centipede effect" 💀💀 pic.twitter.com/SFtDzMpeER
— Beff – e/acc (@BasedBeffJezos) June 30, 2024
Following up, Musk confirmed that Grok-3 would launch by the end of the year and will be trained using 100,000 Nvidia H100 chips. “Grok 3 end of year after training on 100k H100s should be really something special,” he added. This is a significant increase from the 20,000 H100 chips required to train Grok-2, as mentioned by Musk in an April interview with Nicolai Tangen, head of Norway’s sovereign fund.
Grok 2 in August will be Based AGI 🔥 https://t.co/7sEWIDjbsF
— Beff – e/acc (@BasedBeffJezos) July 1, 2024
Cost of Training Grok-3 with Nvidia H100 Chips
Training Grok-3 using 100,000 Nvidia H100 chips represents a substantial investment. Each H100 chip costs between $30,000 and $40,000, which means xAI could spend between $3 billion and $4 billion on AI chips for Grok-3. Alternatively, the company may rent GPU compute from cloud service providers.

xAI’s Cloud Server Plans
A report by The Information in May indicated that xAI has been in discussions with Oracle to spend $10 billion over multiple years to rent cloud servers. Musk’s companies have already acquired H100 chips in recent years, with some reports suggesting these chips have been used for other purposes, such as diverting chips bought for Tesla to X.
Check out more on Artificial Intelligence(AI) around the globe!
Donald Trump’s AI-Generated Dance Video with Elon Musk Gets Viral!
Krutrim to Launch First AI Chip, Bodhi 1, by 2026
OpenAI Unveils GPT-4o mini: Powerful AI at an Affordable Price
Here’s the Full List of 28 US AI Startups That Have Raised $100M or More in 2024
Claude Adds Prompt Engineering Support to Enhance AI Apps
Bumble Users Can Now Report AI-Generated Profiles
OpenAI and Arianna Huffington Back AI Healthcare Venture
Elon Musk Reveals xAI’s Grok Chatbot Launch Timeline
How Tokenization is Limiting Generative AI Models
OpenAI Breach: The Security Threat to Companies
Source: Elon Musk on when xAI will launch next AI chatbots and number of Nvidia’s AI chips needed to train them (msn.com)