SpaceX Hands Anthropic 300 Megawatts at Colossus 1 - And They Want Data Centres in Orbit Next
0:00 / 1:00
Sources
News
SpaceX Hands Anthropic 300 Megawatts at Colossus 1 - And They Want Data Centres in Orbit Next
calendar_today Date:
schedule Duration: 1:00
visibility Views: 216
database
Summary Report
Anthropic has signed a deal for all of SpaceX's Colossus 1 - 220,000 GPUs and 300 megawatts within the month - and the two are exploring orbital data centres.
- 01. Anthropic will use the entire compute capacity of SpaceX's Colossus 1 data centre.
- 02. That is over 220,000 Nvidia GPUs and more than 300 megawatts of new capacity, deploying within the month.
- 03. It adds to existing 5-gigawatt deals with Amazon and Google, $30 billion of Azure capacity, and a $50 billion Fluidstack arrangement.
- 04. SpaceX has not previously been a serious name in AI infrastructure, making the partnership an unusual one.
- 05. Anthropic and SpaceX are also exploring multiple gigawatts of orbital AI compute - data centres in space.
Anthropic has signed a major deal with SpaceX to take the complete compute capacity of the Colossus 1 data centre, gaining access to over 220,000 Nvidia GPUs and 300 megawatts of power. The deployment is set to begin within the month, representing a significant expansion of Anthropic's computing infrastructure.
This arrangement adds to Anthropic's already substantial compute portfolio, which includes 5-gigawatt deals with Amazon and Google, $30 billion worth of Azure capacity, and a $50 billion Fluidstack arrangement. The scale suggests Anthropic believes its current infrastructure remains insufficient for its AI development goals.
The choice of SpaceX as a partner marks an unusual development in AI infrastructure. SpaceX has not previously been recognised as a major player in AI computing services, making this partnership with one of OpenAI's primary competitors particularly noteworthy.
Beyond the immediate data centre deal, both companies are exploring the possibility of multiple gigawatts of orbital AI compute capacity - essentially data centres operating in space. This represents a potential paradigm shift in how AI infrastructure could be deployed, taking the compute race beyond terrestrial limitations.
Meta Data
LLM:
