Most people are arguing about whether AI data centers are a bubble.
Wrong layer.
AI needs power.
Power needs data centers.
Data centers need capital.
The FUD this month is real: $150B+ of U.S. projects stalled or cancelled, Texas auditing the queue, New York freezing large new sites, interconnection filings that look like a terawatt of demand.
A lot of that pipeline is phantom.
Queue position is not power.
A press release is not a transformer.
What does not disappear is the constraint.
The sites that get permitted, interconnected, and financed become scarce. Everything else is noise.
That’s the filter I care about not the headline megawatts. The megawatts that actually turn on.
I’ll write from that stack:
where AI demand hits the grid
which data-center stories survive permitting and power
how capital is being structured around those assets.
If you follow one account for the AI → power → compute → crypto, make it this one.
@Freedom_By_40 Let's ride. That's could be a great September. Remember Tom Lee said he thinks September could be great with most people thinking the opposite.
$CIFR just keeps adding fuel to the story.
Cipher is developing natural-gas pipelines that could enable up to 2.5 GW of new power generation across multiple sites.
More power = more potential HPC capacity.
The pipeline just got a LOT more interesting. ⚡️
$IREN is more than a Bitcoin miner.
It’s building a massive power + data center platform positioned to capitalize on the explosive demand for AI compute.
Secured power, expanding capacity, and hyperscaler demand could make IREN one of the most interesting AI infrastructure plays through 2027. ⚡️