We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
OpenAI Blog·Jan 9·3 min read
OpenAI and SoftBank Group partner with SB Energy
TL;DR
Training advanced AI models requires massive computational power, but building sufficient data center capacity has been a bottleneck. OpenAI and SoftBank Group partnered with SB Energy to construct multi-gigawatt AI data center campuses, including a 1.2 GW facility in Texas for the Stargate initiative.
✦ Why It Matters
Engineers can plan AI infrastructure investments knowing major capacity expansion is underway; researchers gain access to computational resources for large-scale model development.
Key Takeaways
- Plan AI workloads around announced 1.2 GW Texas facility availability timeline.
- Evaluate whether your models require multi-gigawatt scale infrastructure for training.
- Monitor SoftBank-OpenAI partnership announcements for additional data center locations.
- Consider power and cooling constraints when designing large-scale AI systems.
Full Summary
Large language models and advanced AI systems demand enormous computational resources during training and inference (the process of running trained models on new data). Existing data center capacity has struggled to keep pace with AI development, creating a supply constraint.
OpenAI and SoftBank Group announced a partnership with SB Energy to develop multi-gigawatt AI data center campuses—facilities capable of delivering gigawatts (billions of watts) of power. The first concrete project is a 1.2 GW facility in Texas supporting Stargate, OpenAI's infrastructure initiative for scaling AI capabilities.
This partnership combines OpenAI's AI expertise, SoftBank's investment and operational experience, and SB Energy's power infrastructure capabilities. The collaboration addresses both power generation and data center construction, tackling the dual constraints of electricity supply and physical infrastructure.
This enables sustained growth in AI model training and deployment at scale.