Anthropic, Musk and SpaceX: what the WSJ report means for you

The Wall Street Journal has reported that Anthropic co-founder Tom Brown used Republican connections to settle a disagreement over model safety and to win over Elon Musk. The reported result is a deal for Anthropic to use SpaceX's computing power. Many details are still unconfirmed.

According to the Wall Street Journal's report, as summarised by Rio Rundown, Brown used his Republican ties to end a standoff over model safety in June. The same effort won over Musk, the report says, and produced a compute deal with SpaceX worth $1.25 billion a month.

"Compute" simply means the computing power, usually in large data centres, that AI companies need to train and run their models.

If that monthly figure held for a full year, it would come to about $15 billion. That is our own arithmetic, not a figure the WSJ reported.

Techmeme also listed the WSJ report on 10 October.

One caution: this article is based on a summary of the WSJ story, not the full article. The details of the June standoff and the terms of the deal have not been verified. Anthropic has not been quoted in the material we have.

The notes do not say how this affects everyday users of Anthropic's tools, so nothing about prices, access or features can be drawn from it yet.

What is known is limited. A reported deal of this size points to a large amount of computing capacity. Whether that changes anything you see when you use an AI tool has not been stated.

Anthropic plans a pre-IPO investor day on 14 October. An IPO, or initial public offering, is when a private company first sells shares to the public. The investor day is the next scheduled event connected to the company's finances, though the notes do not say whether the SpaceX deal will be discussed.

Anthropic and SpaceX have not confirmed the reported terms in the material available to us.