The book did two big things this month. It rotated out of a long-held winner into two new setups, and it doubled down on the flagship into a drawdown I am going to walk through in full rather than bury. Everything below is marked against the July 10 close. New here? Start with Who Is Shawarma Capital, then come back.
Closed this cycle
Three positions came off the book, one a rotation and two fast round-trips.
PGY, +37 percent, rotated. In mid-June I sold Pagaya at $15.68, up 37 percent from my $11.47 basis, and rotated the proceeds into building Ouster and Amprius. Pagaya did nothing wrong. It printed its first clean GAAP profit and the tape finally repriced it. I took the win and moved the capital into two names with nearer catalysts.
BGDE, +14 percent. I bought Big Digital Energy at the first-post level on June 17 and sold at $11.10 into the re-rate a week later. A failed bitcoin miner sitting on scarce, grid-connected PJM power, insiders buying from four dollars to over eight, and a shell repricing from sixteen to forty million dollars. The trade was the re-rate, and it came fast, from the mid-eights through eleven and change and straight back down. That swing is what I stepped aside from. I still cover the name and shipped a Part 2 on the financing tell, and I am watching for a cleaner re-entry.
CCXI, +19 percent. I bought Agility Robotics through the CCXI trust at $16.40 on June 29, as it broke off the ten-dollar floor, and sold at $19.50 into the spike two sessions later. The stock ran from ten to nearly twenty in two weeks and gives back three and four dollars in a day. I booked the move and I am letting the volatility cool before I decide on re-entry.
The held book, by return
Nine positions now. Four up double digits, from a Rubrik near a double to an AmpliTech at plus forty-nine. Ouster roughly at cost. Four underwater, including one of the two June adds and the flagship I just sized up into weakness.
RBRK, +98 percent. Rubrik nearly doubled from my basis. The Q1 fiscal 2027 print beat every metric and raised guidance across the board, subscription ARR up 32 percent to $1.57B, revenue up 39 percent, and the subscription contribution margin widened to 13.2 percent from 8 percent a year ago. The AI control plane thesis is playing out in the numbers. No exit.
IQEPF, +81 percent. IQE signed the multi-year indium phosphide epiwafer supply agreement with Tower Semiconductor for AI data-center optics and settled the old IP dispute in the same stroke. The stock spiked on the print and gave some back into the semis pullback. Position grew through price action, no fresh capital added.
LPKFF, +80 percent. The activist standoff went to a vote on June 4 and management won. The German listing gave back part of the pop into the small-cap pullback and trades near seventeen and a half euros, still a double from my basis. The glass-substrate AI packaging monopoly thesis is intact and Part 4 is live.
AMPG, +49 percent. AmpliTech is a ten percent position I opened on June 1, on its own and separate from the Pagaya rotation. It ran through the seven-dollar target I underwrote it against and pulled back to the high sixes with the microcaps. A twenty-four-year-old RF house turned 5G open-RAN radio OEM with a quantum tail, 48 percent gross margins and zero debt. The first conversion is on the record, FCC and ISED cleared the radio, and the BEAD program money is moving. Return on my June 1 basis, still the widest asymmetry in the book. Note it is a different company from Amprius below.
OUST, +6 percent. Ouster is now a full ten percent position, built through June and funded by half the Pagaya rotation. The Pentagon widened the China lidar ban to the top two suppliers, the Rev8 factory qualified for federal infrastructure dollars under Build America Buy America, and the company raised two hundred million into strength. Blended basis around forty-one dollars, trading in the low forties. A conviction position now.
BMM, flat. Blue Moon Metals sits right on my basis after the consolidation. Springer in Nevada is still the only permitted US-domiciled ammonium paratungstate hub, and Nussir in Norway carries the EU Critical Raw Materials strategic designation. It is one of the two largest positions in the book, roughly a fifth of it, and I am still building toward the target weight. The probability-weighted expected value against a six-and-a-half-dollar print keeps the asymmetry wide.
MRLN, -17 percent, doubled down. The one I owe you the straight version on, and the biggest position change this month. Merlin is my highest-conviction name, and when the PIPE resale registrations went effective and pushed the stock into a thin float down to a fifty-two-week low near four-thirty, I treated the supply as a gift and added about a hundred thousand shares into the forced selling. That takes Merlin back to the top of the book at around thirty percent and pulls my blended cost down from $6.03 toward the low fives, so the mark is nearer minus seventeen than the minus thirty it would show on the old basis. The move was supply, not story. The C-130J critical design review cleared with USSOCOM, the program moved into aircraft integration on the hundred-million-plus IDIQ ceiling, the defense budget the thesis points at keeps inflecting, and none of the seven kill criteria have been met. I built Merlin World so you can walk the fleet yourself.
CODA, -17 percent. Coda Octopus is still below my basis after the microcap drawdown. Real-time 3D sonar for defense and offshore energy, profitable and debt-free, with defense primes including Anduril moving into the channel. Thesis intact, and the plan is to size back in on dips.
AMPX, -29 percent. Amprius is the other half of the Pagaya rotation, and it is underwater. I built it to a ten percent position in June around sixteen dollars, and the silicon-anode complex has since sold off hard with the rest of the microcaps, taking it to the low elevens. The operating case has not broken. Amprius is the only Western silicon-anode pure-play, Q1 revenue was up two and a half times year over year to $28.5M, it printed its first twenty percent gross margin, and it raised the FY26 guide to a $130M floor with positive adjusted EBITDA. Defense and drone customers include Anduril, Skydio, and Switchblade. Q2 earnings on August 6 are the next read. I show it red because it is red.
How the book is weighted
After the add, Merlin re-anchors the book near thirty percent, with Blue Moon Metals and IQE right behind it around a fifth each. AmpliTech, Ouster, and Amprius each sit near ten percent from the June builds. The winners keep their weight, the losers stay sized to conviction, and nothing is dressed up.
What I am watching into August
MRLN, the next SOCOM C-130J task order window and the Q2 print with backlog, the first hard read on whether the loss narrows on plan now that I am sized up.
AMPX and OUST, both report Q2 on August 6, the first look at each balance sheet after their raises and the first checkpoint on the two new adds.
BGDE and CCXI, both on re-entry watch once the near-term volatility settles.
BMM, continued build toward the target weight as the Springer and Nussir permitting and financing steps land.
The full theses
Every position, in one place:
See you at the end of the month.
Long all held positions. The three above are closed. Not investment advice.
Hecking gnarly brah
You do good work, and I really appreciate your thorough analysis of these companies. I already had owned $MRLN and since I found you guys on Substack, I have since bought $AMPG and $TRT. I looking into becoming a member at some point, thank you!