The $6B round was valued at a $24B post-money valuation. The fundraising total puts xAI in roughly the same class as OpenAI’s $10B investment from Microsoft in 2023 and Anthropic’s ~$6B this year. This compares to the valuation of OpenAI a year ago at around $29B post-money (today it’s about $93B) and Anthropic’s nearly $18B post valuation earlier this year (around $20B today). In other words, while xAI’s valuation is a huge number, and it’s in the same zip code as these two. It’s worth noting that OpenAI’s valuation last year deserved to be higher than xAI because OpenAI’s business was further along.

As for where the money will be invested, the business is capital-intensive and requires aggressive infrastructure investment in GPUs (mostly from Nvidia), hiring talent to train the model, and for distribution. Given the cost of the investment phases, I expect the company will be raising an additional +$6B in a year.

We believe there are only a handful of companies that have all three elements – infrastructure, data, and distribution – with a chance to build a leading AI model: OpenAI (GPT), Google (Gemini), Meta (Llama), and xAI (Grok). We’re confident in the investment for four reasons:

1) We believe there will be about four primary proprietary foundation models that will effectively power the Western World. Generative AI is just scratching the surface and its utility will take a significant jump once we hit general intelligence.

2) xAI’s model is built to be “truthful, competent, and maximally beneficial for all of humanity.” This is a distinction from some of the mishaps from some of the existing foundation models and gives xAI an opportunity to be the model of choice for the sciences. Additionally, businesses that tend to be more right-leaning will likely prefer building on top of xAI given the model will likely better align with their culture.

3) xAI has the benefit of training on data from X. These 250m daily active users will give xAI the world’s best data for real-time intention and news. This is a core advantage that xAI has versus other model companies is access to Twitter/X data for training and distribution. X Corp, the parent of X, owns 25% of xAI. X data is unique versus Google and Meta in that it is more real time and may create an opportunity for Grok to be the best AI model for interacting about immediate events and information. This real time data can be integrated into inference through the use of Retrieval-Augmented Generation (RAG). This means you don’t have to train the model to have the most updated results.

4) The Elon factor. His involvement is central to the success of the company given his vision has a unique power to attract world-class talent and deep pockets to fund what will be an incredibly capital intensive 5-year infrastructure build.

In classic Elon science fiction subculture, he named xAI’s chatbot “Grok” in a nod to Robert Heinlein’s 1961 novel Stranger in a Strange Land, in which he made up the word. It means “to understand intuitively.”

As mentioned, the fundamental advantage of Grok is that it has real-time knowledge of the world via the 𝕏 platform. Grok is powered by Grok-1, xAI’s LLM, which competes with Llama (Meta), GPT-4 (OpenAI), Claude (Anthropic), and Gemini (Google).

Like the other leading models, Grok is multimodal, meaning it can process information via several forms of visuals like images, documents, and screenshots. Musk has described the model as “advancing both our multimodal understanding and generation capabilities are important steps in building beneficial AGI that can understand the universe.”

Today, Grok has a fraction of the power of GPT-4, with OpenAI maintaining about a 2-4 year head start over xAI. In AI years, that is a massive lead in the race reach general intelligence. That said, Grok does have two advantages over OpenAI; access to the 𝕏 data and a model that will likely have a greater appeal with the hard science community.

You could say that the concept of xAI was founded in 2018.  Taking a step back, in 2015 Musk co-founded OpenAI, the company behind ChatGPT, but stepped down from the board in 2018. For about 4 years there was little said about OpenAI beyond Elon’s occasional comments that AI could someday overpower humans. As OpenAI shifted to a for-profit organization, Musk began taking issue with the how the business was being built, and publicly accused OpenAI of training on copyrighted material and “training the AI to lie.”

Elon then committed that he was “going to start something which I call ‘TruthGPT’, or a maximum truth-seeking AI that tries to understand the nature of the universe.” and said TruthGPT “might be the best path to safety” that would be “unlikely to annihilate humans.”

Then Musk acquired Twitter in 2022, rebranding it to 𝕏 in 2023, and founding xAI the same year. The company had raised just under $1B, most of which came from Musk, prior to the recent $6B round.

Disclaimer