Join me on Moby! Use my link: https://t.co/7YV4xL8Lzb. Definitely one of the best mobile apps in the game. Actually the best one just not everyone knows yet @mobyagent link above for yours
Memories yo and more to be had LFG! @wealthonmars thanks for being one who opened the door and continues to do so for others. The value of what I’ve learned there is hard to quantify but truly grateful I found ya!
$SPCX is increasingly turning into a neocloud provider, and that's really bullish for the AI infrastructure trade because it reframes Colossus from a proprietary xAI training asset into a merchant compute platform.
Grok is (even by Elon's own admission) lagging frontier AI peers. But SpaceX has raced to build arguably the largest and most valuable AI compute cluster in the world. Rather than let massive GPU capacity sit under-monetized while xAI fights uphill in the model race, SpaceX is leasing that capacity to compute-starved AI labs that need scale now.
The recent Anthropic and $GOOGL deals first suggested this was more than excess-capacity monetization.
Then, today's new Reflection deal (reportedly worth up to $6.3 billion) makes it look like a repeatable business model.
In other words, SpaceX may have accidentally become the AWS of frontier AI compute, simply because it can bring enormous powered GPU clusters online faster and cheaper than almost anyone else. And by shifting to a neocloud model, SpaceX is now becoming a quasi-guaranteed massive buyer of AI hardware.
And that, obviously, is extremely bullish for the sellers of that AI hardware.
$NVDA is the first that comes to mind. But the sharper plays are in the plumbing: $CRDO for active electrical cables and optics, $ANET $AVGO $MRVL for AI networking and switching silicon, $MU $RMBS for memory bandwidth, $STX $P $NTAP for storage, $VRT $ETN for power and thermal infrastructure, and $CEG $GEV $VST for the power bottleneck.
Model economics may be brutal, but infrastructure economics are improving. Whether Claude, Gemini, Grok, Reflection, or another model wins, they all need more compute.
This is the “buy the companies getting the AI checks, not the companies writing them” thesis being validated in real-time by an unlikely validator of that thesis in SpaceX.