Meta is maintaining a massive supercluster of tens of thousands of Nvidia GPUs to support large-scale AI model training while ensuring operational reliability for its complex infrastructure.
Key Points
- Meta’s Prometheus supercluster utilizes tens of thousands of Nvidia GPUs to sustain continuous AI model training.
- HPE plans to deploy specialized AI hardware to the Moon before the end of the year.
- Brave is implementing trusted execution environments to provide end-to-end encryption for AI chat interactions.
- EQT acquired a majority stake in the Swiss cybersecurity firm Acronis at a valuation exceeding $3.5 billion.
- Microsoft is addressing a zero-day vulnerability affecting on-premises SharePoint servers that remained unpatched after initial security updates.
- Ubuntu released the "Stonking Stingray" beta, featuring increased Rust integration and new speech-to-text capabilities.