Elon Musk’s xAI Secures $6 B Funding: What It Means for AI Supercomputing

Elon Musk founded xAI in March 2023 in Nevada, immediately purchasing 10,000 GPUs.

The team includes former staff from OpenAI, DeepMind, Microsoft and Tesla, with AI safety director Dan Hendrycks serving as an advisor.

Musk has valued xAI at $18 billion and announced a $6 billion Series B round, raising approximately 435 billion CNY. The funding will be used to bring the first products to market, build advanced infrastructure and accelerate future technology research.

xAI has already launched Grok, an upgraded version of OpenAI’s ChatGPT, currently available to X Premium subscribers.

The company’s goal is to build artificial general intelligence and uncover the true nature of the universe.

Funding‑backed Musk begins hiring AI talent

In March 2024 xAI released the Grok‑1 model, a 3.14‑trillion‑parameter expert‑mixture model, surpassing OpenAI’s GPT‑3.5 (1.75 trillion) and Llama 2 by four times. The following month Grok‑1.5V added multimodal capabilities, handling text, images, documents and tables, and outperformed peers on the RealWorldQA benchmark, especially in physical‑world understanding.

The round’s investors include Valor Equity Partners (led by Antonio Gracias), ARK Invest, Vy Capital, Andreessen Horowitz, Sequoia Capital, Fidelity Management & Research and Kingdom Holding, among others.

Previously, xAI had sought $1 billion in equity financing and later aimed for up to $6 billion, which has now been secured, doubling the original target and raising the valuation from $15 billion to $18 billion.

Investors are betting on the synergy between xAI and Musk’s social platform X as a massive data source for training large language models.

xAI plans to train Grok 2.0 on 20,000 GPUs and eventually on 100,000 Nvidia H100 chips for a next‑generation AI supercomputer. Musk also disclosed that Tesla owns 35,000 Nvidia H100 GPUs for autonomous driving, with plans to double that by year‑end.

By fall 2025, xAI intends to launch a new data center whose performance will be four times that of the current largest AI data center (≈ 400 exaflops), requiring 100,000 Nvidia H200 chips, which are expected to be available in early 2025.

The massive investment reflects the high cost of competing in AI hardware, talent and technology, with major tech firms and venture capital pouring billions into AI startups, and highlights the competitive landscape against OpenAI, Meta’s Llama, Anthropic’s Claude and Google’s Gemini.