Compute shortages/pricing hikes; AWS $50B/Trainium/HBM4 but MS-AWS friction/10x costs/$121B burn/IPO ties/Stargate slowdown
Key Questions
What compute challenges is OpenAI facing?
CFO highlights shortages causing missed opportunities, $600B uncertainty, and $121B 2028 burn pressuring inference costs at 50%+ of revenue.
How might this affect pricing for users?
Inference demands could lead to hikes for Plus/Pro subscriptions as costs 10x amid capacity constraints.
What infrastructure deals offset shortages?
AWS $50B Trainium, Samsung HBM4, Tata/AMD for Stargate/Oracle/MS Abilene; Nvidia Rubin and caching help.
What geopolitical risks exist?
Iran threats to Stargate data center in Abu Dhabi amid MS-AWS friction and GitHub tensions.
How do leadership and IPO tie into risks?
Brockman $110B concerns and pricing risks linked to geo/leadership issues pre-IPO.
CFO shortages/$600B uncertainty/$121B '28 burn/inference 50%+ rev pressuring hikes (Plus/Pro); AWS Trainium $50B/Samsung HBM4/Tata/AMD offset Stargate slowdown/Oracle/MS Abilene; Iran threats Stargate Abu Dhabi; MS-AWS tension/GitHub/Brockman $110B/Nvidia Rubin/caching; Sora shutdown frees compute. Capacity/pricing risks geo/leadership.