$IREN is now hiring operators from the hyperscalers it wants to compete with.
Nicholas Sandberg joins as Head of US Operations after 20 years across $ORCL and $AMZN AWS to help turn IREN’s expanding US power and data center footprint into a scaled AI infrastructure platform.
Terrible news for $ASTS, obviously. 😂
Japan is now:
• Building Tactical AI satellites with NTT Data
• Integrating AI into military command & decision-making
• Creating a multi-layered space architecture
• Building sovereign, resilient LEO communications
• Funding the Rakuten/AST SpaceMobile J-LEO network
• Targeting AST-powered direct-to-device service in Q4 2026
So $ASTS is increasingly sitting at the intersection of:
AI + DEFENSE + SOVEREIGN COMMS + EMERGENCY CONNECTIVITY + DIRECT-TO-DEVICE BROADBAND.
But please…
Tell me again how AST SpaceMobile is just a speculative satellite-phone company with no real use case. 🤡
Important nuance:
Japan has NOT publicly named AST SpaceMobile as the connectivity layer for the Defense Ministry’s Tactical AI satellite program.
Yet.
The pieces are getting VERY interesting. 🛰️🇯🇵🔥
P.S. Investing gets a lot easier once you learn to read past the headline. 😉
I read $AXTI's latest 10-Q this morning, and two things in it don't fit together in a way that's worth a minute of your time.
First: AXT cannot ship indium phosphide substrates to customers in the United States. China put InP on its export control list in February 2025. Its subsidiary Tongmei got the first permits four months later, covering certain customers in Europe and Japan. The UK and Canada followed. The US did not.
The filing's own words: "We are unable to estimate when we will receive the necessary export permits to resume shipping our indium phosphide substrates to the U.S." It calls InP export permits "the most significant challenge we currently face."
Nineteen months and counting.
Second: in that same quarter, two American optical companies signed up for that capacity anyway.
On June 26, $COHR — a Pennsylvania corporation — signed a three-year development and supply agreement for six-inch InP substrates, with AXT committing to expand its Beijing plant across 2026 through 2028. On July 26, $LITE signed a six-year capacity reservation running to the end of 2031, carrying two deposits of $43.5 million each.
The first was due within 30 business days. The second is scheduled for sometime in 2028. Both apply as shipment credits against future purchases.
So: two US buyers committed six years and cash up front to a Chinese fab, for wafers that per the most recent filing cannot legally leave China bound for a US customer.
That's the whole post, really. It's an observation about what conviction costs, more than a call on the stock.
There's a tariff layer underneath as well. US duties on many Chinese imports including these substrates went from 25% to 50% in January 2025, to 60% in February, to 70% in March. The Supreme Court struck down many IEEPA tariffs in February 2026, though other authorities survive and the administration has signalled it may rebuild under Section 122. China's revised Foreign Trade Law took effect March 1, 2026, with expanded power to prohibit or restrict exports.
What I don't know, and won't pretend to: the filings never say where those substrates get delivered. Lumentum Operations LLC is a Delaware entity, but Lumentum runs fabs outside the US. Some or all of this capacity may be destined for non-US sites, in which case the permit problem doesn't bind the way it appears to. I'd want that answered before sizing anything.
The direction holds either way though. When a buyer wires eight figures to reserve six years of output from a supplier that currently cannot ship to it, the buyer is telling you the bottleneck is real and that it expects the paperwork to resolve. Chokepoint analysis usually means guessing who's constrained. Here somebody paid to tell you.
What would change my mind: the US permit landing. The moment it does, the scarcity premium in this arrangement starts to compress, and the interesting question moves from access to plain capacity — where the answer is just whoever builds fastest.
$AEHR Late-fail economics
A failed AI die found after HBM + advanced package is a 10x–100x cost event.
$AEHR durable edge is not “another burn-in box.”
It is screening at wafer before the expensive stack is assembled.
FOX-XP is the only production-proven full-wafer platform at AI-processor power.
Sonoma already won two consecutive generations at a lead hyperscaler on the package side.
3 attractive stocks you've probably never heard of.
1. $FTAI - Owns, leases and repairs jet engines.
Converting old jet engines into 25MW gas turbines for data centers.
- 55% YoY revenue growth
- 60% 2YR FWD EPS Growth
Currently trading at 0.6x PEG
2. $CLS - Hardware systems for Cloud providers and Hyperscalers. Connectivity, Racks and Storage.
- 66% YoY revenue growth
- 42% 2YR FWD EPS Growth
Currently trading at 22x NTM P/E
3. $VIST - Shale oil from Argentina’s Vaca Muerta. Exploration, production and crude exports.
- 67% YoY revenue growth
- 36.5% 2YR FWD EBITDA
Currently trading at 7.5x NTM P/E
Buying stocks before they get popular on X is how you win big.
$SOFI CEO Anthony Noto says “85% of our products are non lending” as SoFi continues moving beyond its roots as a lender.
That mix is what could ultimately push the market to value SoFi more like a financial platform than a lending business.
$CIFR will soon be a setup to buy, not yet.
Maybe soon, if we manage to grind above the 1D HMA, which currently sits at $15.5, and escape this week long falling wedge, $CIFR will be a buy.
On your watchlists so we manage to catch the setup on time.
ELON MUSK BOUGHT $6.69 BILLION WORTH OF TESLA $TSLA ...
According to the latest insider transaction data, Elon Musk acquired 286.4 MILLION shares of Tesla on June 17th, valued at approximately $6.69 BILLION.
CREDO INSIDERS HAVE BEEN SELLING MILLIONS IN $CRDO ...
Recent insider transactions include:
1) COO Yat Tung Lam: $47.96M+ sold
2) CTO Chi Fung Cheng: $39.8M+ sold
3) CLO James Laufman: $1.82M sold
4) CEO William Joseph Brennan: $1.49M sold
5) CFO Daniel Fleming: $595K sold
That’s roughly $92 MILLION in insider selling across the transactions shown...
After listening to $CRDO call, I can conclude that for me, it seems like a no-brainer.
Always liked the company profile.
Founder led, huge growth, very profitable with high margins(68-70%).
I'm might be interested again.
Who is buying?
$GOOGL Waymo has cut estimated robotaxi hardware costs from ~$115K per vehicle to just ~$20K with its 6th-gen system.
A big driver is vertical integration including a custom 5nm chip capable of 1,000 TOPS that processes camera, lidar and radar data in real time.
$GOOGL owned Waymo is opening paid robotaxi rides to the public in Denver, San Diego and Tampa.
That brings fully autonomous service to 14 U.S. cities as Waymo keeps scaling into a broader commercial network.