The whole thing started with $5,000 and a question.
What happens if a normal retail investor stops pretending he is a mini hedge fund, stops chasing every shiny object with a ticker, and instead builds a concentrated book around the places where the future is actively under construction?
Not the “future” in the conference-panel sense. I mean the actual pipes: chips, memory, storage, satellites, launch infrastructure, energy, fintech rails, robotics, quantum, biotech. The unsexy plumbing under the headlines. The stuff the world has to buy before the keynote demo becomes reality.
So I opened the account with $5,000 on March 31, 2026. By the Day 62 snapshot, it was sitting around $7,179 — up about 43.6%. That is a strong start. It is also early enough that bragging would be goofy. Two months does not make a genius. Sometimes it makes a guy with good timing and a keyboard.
The point is not “look how smart I am.” The point is: here is the process, here are the rules, here are the receipts, and here is where I might be wrong.
The three rules that run the book
1. Seasoning, not the steak.
Sub-$5 names and pre-revenue moonshots can exist in the portfolio, but they do not get to be the meal. A $200–$500 seed position is research with a heartbeat. A giant allocation into a cash-burning science project is how you turn ambition into confetti.
2. Catalysts over vibes.
I do not buy because something feels ready to run. I want something specific: earnings, filings, launches, trial readouts, regulatory milestones, IPO windows, contract cycles. A catalyst has a date. A vibe has a Pinterest board.
3. Cheap and distressed are different things.
A beaten-down quality company can be interesting. A company with going-concern smoke, debt issues, and cash running on fumes is not “undervalued.” It is a trap with a ticker symbol and a cute little hat.
The operating principle.
Aggressive, but disciplined. Big swings are fine. Blindfolded swings are how you donate money to strangers with nicer boats.
The early scoreboard
The early MVPs were the names that sat under real demand instead of just a good story: MU, RKLB, STX, NVDA, and GOOGL. Memory, launch, storage, AI compute, and the largest digital toll roads on Earth. Not exactly a sleepy little picnic basket.
The laggards were not necessarily failures. SOFI, OKLO, and ARM were still on the board because the thesis had not broken. There is a difference between “not moving yet” and “wrong.” The hard part is knowing which one you are holding before the market charges tuition.
The most important move was not buying. It was trimming. MU got trimmed after running hard because one position had grown too large for the account. That was not doubt. That was discipline. If a single name can ruin the month, it is no longer a position. It is a hostage situation.
| Bucket | Names | Read-through |
|---|---|---|
| MVPs | MU, RKLB, STX, NVDA, GOOGL | The enabler layer worked. Infrastructure beat decoration. |
| Laggards still held | SOFI, OKLO, ARM | Slow starts, but thesis intact pending catalysts. |
| Exited | PATH, YPF | PATH was the wrong sector/time. YPF did its macro-hedge job, then the setup cooled. |
| Planted | ARM, OKLO, SERV | Semiconductor IP, nuclear-for-AI, and the first physical-AI seed. |
Where my attention is going
The themes are not equal. Some are firing now. Some are next-wave seeds. Some are dangerous little dragons that deserve a tiny cage and a respectful distance.
| Theme | Status | My current read |
|---|---|---|
| AI Infrastructure | Firing now | The backbone: semis, memory, storage, power, data centers, networking. |
| Space Economy | Catalyst-heavy | Public space names can benefit from the SpaceX halo, but dates and filings need verification. |
| Quantum | Real, volatile | The theme is increasingly validated, but sizing has to respect the violence. |
| eVTOL / Air Mobility | Milestone-driven | Huge upside narratives, pre-revenue reality. Patience and position size matter. |
| Nuclear AI Power | Next-wave seed | Compute demand eventually hits the wall called electricity. |
| Physical AI | Early innings | Robots are moving from demo stage to workflow stage. Not fully investable everywhere yet. |
| Speculative Biotech | Seasoning only | Binary outcomes, real upside, no hero sizing. |
A good portfolio needs a trash can.
The biggest lie in retail investing is that every idea deserves a place in the portfolio. It does not. Some ideas deserve a watchlist. Some deserve a note. Some deserve to be admired from a safe distance like a Florida raccoon holding a switchblade.
This is why exclusion matters. I want the report to say what got rejected and why. If a stock is cheap because the business is deteriorating, I do not need a lower price. I need a better business. Cheap is not a strategy. Cheap plus survival plus a catalyst can be a strategy. That middle word is doing a lot of work.
The bubble bears may be right about valuation and wrong about timing.
I am not pretending AI infrastructure is cheap everywhere. It is not. Some of the charts look like they were drawn by a caffeinated falcon.
But valuation alone does not pop the bubble. Cancellations do. Demand cracks do. Order books rolling over do. Capex getting pulled does. As long as the hyperscalers are spending, memory is tight, storage demand is real, and the compute arms race is still escalating, I am more worried about being too early to panic than too late to notice.
I get cautious when the orders start getting cancelled. Not when the comment section gets dramatic.
A public lab notebook for frontier-tech investing.
The Frontier File is where I track the account, the thesis, the catalysts, the mistakes, and the evolving map of the sectors I think matter most over the next decade.
It will not be polite stock-market soup. It will be direct, opinionated, occasionally wrong, and hopefully useful. The promise is not perfection. The promise is receipts.
Next issue: the actual thesis board, updated catalyst calendar, and the names I think are doing real work underneath the AI headline trade.