$IREN entire 2GW Sweetwater Hub has now been included in ERCOT Batch Zero as Base Load.
Sweetwater 1: 1,400MW
Sweetwater 2: 600MW
Total: 2,000MW
Great news. 👏
If you’re impressed with Astra, just remember it was pretrained using more than 100,000 GPUs.
Now imagine what’s coming over the next 12–18 months.
As an investor, I think the best way to play this is to own the bottlenecks.
Just remember, it was retail investors who found stocks like $PLTR, $IREN, $HIMS, $NBIS, $AMD, $HOOD, $RKLB and $ASTS in the single digits, long before they became Wall Street favourites.
CEO Jensen Huang said “AGI has arrived” and that Astra was trained on 100K+ Blackwell GPUs.
Plus, another 400K GPUs are coming online next. That’s 4x as many GPUs.
As an investor, the opportunity looks something like this:
AI factories / Neoclouds:
$IREN, $NBIS, $CRWV
$IREN doesn’t need every customer to be hyperscaler-sized today. It benefits if some of today’s smaller AI companies become substantially larger tomorrow.
Just look at how quickly some of its customers are already growing:
You don’t build wealth by waiting for everyone else to agree with you.
Find undervalued companies before the big institutions do. I found $HIMS at $6, $IREN at $6 and $ONDS at $1.80. I built my conviction and started my positions at those levels.
Berkshire Hathaway CEO recently said he has “always had the strong view that energy would be the constraint” when it comes to the data-center buildout, and that they still see it as a significant opportunity.
This is nothing new.
Stop trying to control the market.
Control your own decisions.
If the thesis is intact, hold.
If the price gives you an opportunity, buy.
If the thesis breaks, act.
Everything else is noise.
$HOOD and $SOFI listed within two months of each other in 2021.
One is 3.5x its first-day close. The other is still 20% underwater five years later.
2022 → HOOD fell to $6.81, SOFI to $4.24.
2023–24 → both crawled sideways off the floor businesses to enter.
Most investors feel optimistic after prices have already risen and pessimistic after prices have already fallen. That is why they repeatedly buy high and sell low. Sometimes, successful investing requires doing the opposite.
A huge amount of the capital Anthropic raises will ultimately flow straight back into AI infrastructure because compute is the bottleneck. More GPUs, more memory, more data centers, more networking and more power.
That is why I want to own the picks and shovels.
$IREN & $SKHY
Anthropic IPO will likely pull liquidity and attention away from other AI stocks. They will be competing for capital currently sitting in NVDA, neoclouds and semiconductors. This would be a temporary liquidity event, so I would treat any sell-off as an opportunity.
$IREN remains my highest-conviction pick for the rest of the year, and with four months still to go, I think there is a very good chance we see this stock double from here. There will obviously be volatility along the way.