TickerTain
TickerTain
NewsroomShortsPortfolioConvergence
NewsroomShortsPortfolioConvergence
←
▶ 1:03 · $SPCX · Top 10 largest AI Datacenters in 2026
episode briefing
SemiAnalysis

Top 10 largest AI Datacenters in 2026

2026-01-06 · 6 company · 4 thematic
sentiment
6 bull0 bear0 neu
quote
“Repurposing existing infrastructure is a reoccurring theme when it comes to XAI. Where in the past or were built, now AI servers are hooking up the latest croc models. Right now, Colossus 1 runs approximately 200,000 Nvidia Hopper GPUs and…”
—unknown
now playing · $SPCX
$SPCX···bullish· highunknown
XAI rapidly scales gigawatt AI infrastructure across two Memphis sites
XAI is executing the fastest large-scale AI infrastructure buildout by repurposing existing industrial facilities and deploying 200k+ Hopper and 110k+ Blackwell GPUs, with both Co…
$OPENAIbullish· mediumunknown
OpenAI Stargate project in Abilene expanding to 6 buildings by mid-2026
OpenAI's Stargate site already operates 100k Blackwell GB200 at 200 MW with six additional data center buildings under construction, positioning it to rise rapidly in the global A…
$AMZN···bullish· highunknown
Amazon deploys 500k custom Trainium 2 chips across two GW-scale AI campuses
Amazon's in-house Trainium 2 silicon is proving competitive at massive scale — half a million chips at the New Carlisle campus alone — with both Mississippi and Indiana sites on m…
$MSFT···bullish· highunknown
Microsoft builds twin 2 GW+ Fairwater campuses with 300k+ Blackwell GPUs
Microsoft is executing parallel gigawatt-scale Fairwater campus deployments in Wisconsin and Georgia, each starting at 150k+ GB200 GPUs with funded expansion paths to 2 GW+ and 70…
$META···bullish· mediumunknown
Meta operates 500 MW+ Columbus cluster using rapid-deploy tent architecture
Meta's Columbus site exceeds 500 MW of installed AI compute by mixing legacy, high-density, and novel tent-based data center designs, prioritizing speed-to-market in the AI infras…
$GOOGL···bullish· highunknown
Google leads with two 1 GW+ multi-campus TPU clusters using distributed training
Google's Omaha and Columbus sites each exceed 1 GW total capacity with 500 MW+ dedicated to AI, leveraging hundreds of thousands of latest-generation custom TPUs across distribute…