AI data centers are set for explosive growth.
Bank of America estimates ~212 GW of new capacity will be needed between 2026 and 2030, with annual deployments rising from 13 GW this year to 55.7 GW by decade’s end.
Nvidia still leads, but the race is widening fast. AMD, Google’s TPU, AWS Trainium, Microsoft Maia, Meta’s MTIA and OpenAI are all scaling custom silicon aggressively. Power, cooling and grid interconnects are becoming the real bottlenecks, not just chips.
The infrastructure buildout required will reshape energy demand and semiconductor supply chains for years.
$NVDA $AMD $GOOGL $AMZN $MSFT
AI data centers are about to explode 📈
Bottom-up analysis shows ~212GW of incremental capacity needed from 2026–2030.
Annual deployment:
• 2025: 13.0 GW
• 2030E: 55.7 GW
Nvidia still dominates, but AMD, Google TPU, AWS Trainium, Microsoft Maia, Meta MTIA & OpenAI are all scaling fast.