Hey, it's Gabriel.
These are my private notes this week on building wealth, raising my family, and breaking the generational patterns I inherited.
Quick note before we get in: I went quiet for two weeks. I was head-down launching the Two-Hour CEO Skool. To make up for it, you're getting two letters from me this week. This is one. The other lands later in the week.
⚡ The Leverage
Four months sunk into OpenClaw. I switched to Hermes anyway.
For the past 4 months, OpenClaw was my main AI agent. The system running my mornings, my inbox, my calendar, the heartbeat of my company.
It kept breaking down.
So this week, I officially made the switch I'd been putting off. I retired OpenClaw and made Hermes the main agent I run everything through.
Yes, I'd already invested a lot into OpenClaw. Yes, I could have kept patching it. That's exactly the trap I refused to fall into — the Sunk Cost Fallacy.
Quick frame for anyone who hasn't run into the idea:
The Sunk Cost Fallacy is the mistake of keeping something alive because you've already poured time, money, or pride into it — not because it's still the right thing.
The money and the months you've already spent are gone. You can't get them back. The only question that matters is: from this point forward, what wins?
More importantly — in an AI world moving this fast, you have to hold everything loosely. The tool you build your stack on today might not be the right one in three months. Don't marry your infrastructure. Date it.
Hermes is more reliable, simpler, and much easier to use. So here's the proof of what the switch actually unlocked:
This week, using Hermes + my second brain, I launched the Two Hour CEO Skool on my own. Zero ads. Zero employees. First 100 founding members in 5 days. Five-figure launch.
This is genuinely an amazing time to be alive — when one person, with the right second brain and the right agent, can run a one-man company that used to need a team.
Quick reminders since I went dark for two weeks:
The Two Hour CEO Skool filled all 100 founding spots and enrolment is currently closed. We reopen enrolment this Friday — join the waitlist here.
If you haven't grabbed the AI Second Brain Blueprint — the same architecture Hermes now runs on — it's free here.
The Uncrowded Take: Don't keep funding something because you've already spent on it. Especially in AI. Hold everything loosely. When it's time to switch, switch.
❤️ The Partnership
Worship in front of your kids. Especially when it's hard.
This week, our family went through something pretty terrible. I can't go into the specifics. What matters is what we did about it.
Ayla is 5. Avia is 3.
The instinct as parents — especially with kids this young — is to hide the hard things from them. Two reasons we usually do it:
There's nothing they can actually do about it since they are so young
We don't want to burden them
Both reasons are real. Both reasons are also wrong in a quieter way, I didn't see until this week.
Zoey and I have a vision for our kids: that they grow up with their own personal relationship with Jesus. Not borrowed. Not inherited from us. Theirs.
The way I believe they'll get there is by watching Zoey and me have a personal relationship with Jesus ourselves. Worshipping God when life is good — and worshipping God when life isn't.
So this week, we sat them down. We told them what happened, as much as a 5-year-old and a 3-year-old can hold. And then we prayed together as a family.
They didn't fully understand it. They probably won't for a long time.
But here's what I believe they'll carry as they grow up: a memory of mom and dad gathering them in when things were hard, naming it honestly, and worshipping anyway. Declaring out loud, as a family, that God would bring us through it.
Life is not always going to be rosy and smooth. When we hit a rough patch, we stay together. We worship together. We pray together. We declare that God will bring us out.
That's not a lesson I can teach with words. It only transmits through being seen doing it.
If I get to pass one inheritance to my kids — bigger than any business I build or money I leave behind — it's this. That they can run to God whenever they are in trouble.
The "Uncrowded" Move: Don't only worship in front of your kids when things are good. The hard weeks are when the real lesson lands.
💰 The Portfolio
$700 billion is moving. Look underneath where it lands.
Meta, Google, Amazon, and Microsoft have collectively committed over $700B in AI infrastructure spend for 2026 alone. Signed budget. Board-approved. Already flowing.
The interesting question isn't whether AI gets built out. It's which layer underneath captures durable economics from that spend.
Roughly five buckets:
Chips — NVDA and the rest of the silicon ecosystem (Broadcom, AMD)
Power — natural gas, nuclear restarts (Vistra, Constellation), and the companies running the grid
Data centres — the companies that own and lease the physical buildings AI runs out of (DLR, EQIX)
Cooling and electrical kit — Vertiv, Eaton — the picks-and-shovels under the picks-and-shovels
Networking — the cables and switches that tie every data centre to every other one
Personally, I'm leaning hardest into cooling and electrical infrastructure. As data centres keep scaling, demand for the kit underneath them doesn't slow — it compounds.
A reminder on the underneath play: 1–2 months ago, when I saw the OpenClaw rush and Mac Minis selling out everywhere, I flagged memory stocks as worth a look. Micron is up roughly 90% in the last month after announcing there's a RAM constraint across the entire market.
That's the pattern I keep coming back to. Look one layer beneath the headline.
I still hear "AI bubble". I'm still not buying it. A bubble in the model layer is not a bubble in the power layer. We're 2–3 years into a supercycle, not weeks into a hype cycle. Will there be a correction along the way? Probably. That's normal. But if you're in for the long term, the next 3–5 years are going to be wild.
Companies are not slowing their spend. They have to spend to catch up and stay ahead. That keeps data centres getting built. That keeps compute and energy constrained for years. That keeps the picks-and-shovels selling.
Signal vs. Noise: The noise is which model launched this week. The signal is the layer underneath every model needs to run.
🧬 The Protocol
Drink the coffee. Then immediately go to sleep.
Last issue I told you I was cutting coffee. That's still going — and honestly, you don't know how addicted you are to something until you try to quit it.
This week I tried something else. Weirder. The caffeine nap.
Here's the idea. Caffeine works by blocking the adenosine receptors in your brain — adenosine is the chemical that builds up through the day and makes you feel sleepy. A short nap also clears adenosine. Stack the two together — coffee, then nap — and you hit the same target from two directions at once.
The protocol I've been running:
Drink a cup of coffee. Cold preferred, because you want to down it, not sip it — caffeine should be hitting while you're already lying down.
Get into bed. Full blackout mask on.
Set an alarm for exactly 25 minutes. No more. Past 25 minutes you slip into deeper sleep and wake up groggy — the opposite of what you want.
When the alarm goes off, you stand up noticeably sharper than a normal nap leaves you. The caffeine has kicked in while you slept, and the adenosine has been cleared.
It's been pulling me through the harder weeks. There's a lot going on right now — the Skool launch, the kids, everything I've already mentioned above — and a 25-minute reset in the middle of the day is the cheat code I keep coming back to.
The Uncrowded Move: When you can't add more hours, learn to recover faster. Twenty-five minutes done right beats another hour grinding through fog.
See you again later this week,
Gabriel Judah
P.S. Two reminders, since I went quiet for two weeks: (1) Founding 100 in the Two Hour CEO Skool filled up — enrolment reopens this Friday. Join the waitlist here. (2) The AI Second Brain Blueprint — the same one Hermes now runs on — is free here. I owe you one more letter this week — it's coming.