Musk Plans to Double Nvidia Chip Count at Colossus 2 by Year-End
Summary
In a post on X, Musk said xAI's Memphis-based AI compute cluster Colossus 2 currently runs 110,000 Nvidia GB200 chips and 440,000 GB300 chips, and gave his most detailed expansion timeline to date: another 220,000 GB300s next week, an expected 220,000 more in November, and possibly another 220,000 in late December "if we're lucky." That would more than double the cluster's Nvidia chip count by year-end.
Bubble analysis
This is direct evidence of the AI compute arms race: a still-unprofitable AI company plans to more than double its chip count within months, implying tens of billions of dollars of continued capex acceleration. Chip purchases running far ahead of current AI revenue is exactly the mechanism bubble skeptics cite — spending outrunning returns.
Related signals
linked by shared tags
- Nvidia-Backed Firmus Plans Australian IPO at $30 Billion Target Valuation#datacenter#nvidia
- Tencent Reportedly Signs Data Center Lease with Oracle, Gaining Access to 100,000 Cutting-Edge AI Chips' Compute#datacenter#chips
- Report: Tencent Leases Oracle Southeast Asia Data Centers — About 100,000 Advanced AI Chips, Deal Valued at Roughly $7 Billion#datacenter#nvidia
- CICC: Research Methods and Insights on AI Compute Hardware — Compute Chips, Storage, Equipment and Consumer Electronics, Sept 2026#chips
- Toshiba Plans to Double HDD Capacity for AI Data Centers, Investing About 60 Billion Yen#datacenter