Skip to main content
The Frontier File · Issue #5 · Week ending June 26, 2026

My Stock Won.
Tech Lost.

One retail account. One concentrated frontier-tech thesis. Micron printed the blowout of the year while the tech tape cracked. That is not a contradiction. That is a current shift — and the current tells you where the oxygen is going.

+58.4%
On capital · Day 87
user-reported book
$8,346
Book value · from $5,270
as of Fri close
19 / 20
Positions green
ARM lone red
~21–22%
MU concentration
the live risk
Captain’s Log

The week the tide split

Micron gave me the win, and the market gave me the warning. Revenue exploded, margins went nuclear, guidance cleared the estimate like a flats skiff over clear water — and then tech still sold off. That is the whole issue. Not “AI is fake.” Not “semis are dead.” The market was repricing who gets paid for the AI buildout and who has to pay up.

Micron reported fiscal Q3 revenue of $41.46 billion, non-GAAP gross margin of 84.9%, and guided fiscal Q4 revenue to $50.0 billion ± $1.0 billion. That is not a vibe. That is a cash register with a jet engine bolted to it.

But Friday’s tape mattered too. MU cooled to about $1,132.33 by Friday close, QQQ finished at $706.52, and SPY at $728.99. So I’m cleaning up the newsletter language: the post-earnings glow was real, but the final weekend snapshot should be anchored to the actual Friday close.

The refined thesis

This was not tech losing because AI demand broke. This was tech wobbling because AI demand got expensive. Memory makers collect the toll. Hyperscalers, device makers, and hardware buyers feel the squeeze. My book owns both sides, but it is overweight the side collecting the check. That is the Memory Wedge.

The Thesis Board

Where my head is at

The map before the positions. firing   mixed/watching   next-wave seed   headwind.

The Memory Wedge MU · STX collect · AMZN · GOOGL pay
The cleanest insight of the week. Rising memory prices help the scarce-asset owners and pressure the buyers. This is why a memory stock can win while parts of tech lose.
Concentration Risk MU ~21–22% of book
The thesis got stronger. The sizing got louder. This is no longer a theory discussion — it is the next process test.
Rates / Multiples still the undertow
The Fed/rate backdrop can still compress frontier multiples even when company execution is strong. Great results can still meet a mean tape.
IPO Froth / Frontier Cooldown OpenAI · SPCX · private-market air
Reported OpenAI IPO delay chatter adds pressure to the “frontier valuation” mood. Good. Froth cooling creates better entries for actual infrastructure.
Space / Rate-Sensitive RKLB · PL · ASTS · ACHR · JOBY · SOFI
These remain live but must earn their size through milestones. Space is not seasoning anymore in spots, but the price still has to make sense.
Next-Wave Seeds SERV · OKLO · QBTS
Small, weird, asymmetric. Exactly where they belong. The rule stays the rule: seasoning, not the steak.
The Week in One Tape

Just the numbers, cleaned up

MarkerLevelRead
MU Friday close after earnings week
$1,132.33
still the driver
QQQ tech proxy
$706.52
soft tape
SPY broad market proxy
$728.99
risk-off close
GOOGL Dow inclusion Monday
$337.39
watch flow
AAPL memory-cost price hike narrative
volatile
buyer squeeze
OpenAI IPO chatter frontier sentiment
2027?
froth check

The better framing: money did not reject AI. It started asking which companies have pricing power inside AI. Micron’s quarter says memory has it. Apple’s price hikes say memory buyers are feeling it. That gap is the tide line.

The Portfolio Reef

The whole book, no hiding

MU
+268%
The steak · size now loud
RKLB
+60%
Launch proxy
PL
+51%
Sat imagery
STX
+49%
Memory tailwind
CENX
+46%
Industrial input
BWXT
+39%
Nuclear defense
AEVA
+38%
LiDAR
MDA
+29%
Space systems
GOOGL
+28%
Dow join watch
ASTS
+27%
Satellite cellular
JOBY
+26%
eVTOL
ACHR
+26%
eVTOL
NVDA
+21%
AI compute core
AMZN
+15%
Buyer side of wedge
SPCX
+13%
1 share · froth trimmed
SOFI
+8%
Fintech rails
QBTS
+4%
Quantum seasoning
OKLO
+4%
Nuclear AI
SERV
+2%
Robotics seed
ARM
-2%
Lone red
+58.4%
On invested capital
$8,346
Book value
19/20
Positions green
+268%
MU, the steak
~4×
vs S&P run-rate
Action Deck

Trim the size, not the thesis

The next move is not dramatic. It is maintenance. No victory-lap selling. No panic selling. No market-open cowboy button. Just bring the book back into balance.

Clean center trim~0.29 MU share

At Friday’s ~$1,132 close, that would raise roughly ~$328 and pull MU toward ~17.5% of the reported book.

Acceptable range0.25–0.35 share

Raises roughly ~$283–$396 and moves cash from ~$246 toward about ~$529–$642.

Process ruleLimit order only

Let liquidity come to you. If MU rips away, you still own the steak. If it fades, you avoided selling the hole.

Public accountability line

If next issue still says MU is ~22% of the book, the issue should call it out plainly. The discipline is not the sentence. The discipline is the action.

Catalyst Calendar

We position around dates

Mon Jun 29
GOOGL joins the Dow Jones Industrial Average, replacing Verizon — watch index-flow tailwind versus megacap wobble.
Tailwind
This week
MU concentration cleanup — target ~17–18%, rebuild cash, keep thesis intact.
My move
Thu Jul 2
June payrolls — labor read into a still-hawkish Fed backdrop.
High
Fri Jul 3
U.S. markets closed for Independence Day observed.
Holiday
Jun 29
Honeywell Aerospace spin-off / HONA watch — industrial/aerospace unlock and index mechanics.
Watch
Late Jul
SOFI Q2 earnings — fintech rails get re-tested.
High
The Hot Take

“A blowout that drops tech is not a contradiction. It is the market discovering who owns the tollbooth.”

The bears looked at Micron’s monster quarter and sold pieces of tech. That sounds insane until you see the pressure line. If memory is scarce enough to push Micron’s margins toward 85%, then every company buying that memory has a new cost problem. Apple’s MacBook and iPad price hikes were not a random consumer-electronics footnote. They were the Memory Wedge showing up at the register.

That is the whole reason the book held up. I own the scarce thing. I also own some buyers of the scarce thing, but not enough for the buyer-side pain to drown the seller-side gain. The thesis did not get weaker because tech sold. It got more specific.

Buddy, Real Talk

The win exposed the weak spot

Two issues running, I have flagged that Micron is too big. Two issues running, I have not fixed it. That is not evil. It is just human. Winners make you feel smart, and feeling smart is where sloppy risk management likes to hide in a nice shirt.

The trick is not to punish the winner. The trick is to stop letting one name carry the steering wheel. Micron did its job. Now I have to do mine. Trim a slice, keep the core, rebuild the powder, and let the rest of the portfolio breathe.

That is the honest version: the win is real, the thesis is stronger, and the sizing needs a haircut before the ocean decides to give me one with barnacles attached.

The Bottom Line

Own the scarce thing. Fix the size.

Eighty-seven days. $5,270 deployed. About $8,346 on the board. Roughly +58.4%. Nineteen of twenty positions green. A memory-stock monster win inside a tech tape that got weird fast.

The lesson is cleaner now: the market was not rejecting AI demand. It was repricing the cost of AI demand. Memory suppliers collect. Memory buyers pay. That is the wedge, and this week it showed up in my actual account instead of just a thesis doc.

But concentration is still concentration, even when it is wearing a crown. The right move is not to abandon MU. It is to right-size MU, keep the thesis alive, rebuild dry powder, and keep hunting the next frontier setup with a little less “please don’t sneeze, Micron” baked into the account.

See you next Sunday, buddy. — Alex

Source Notes

Receipts, not vibes

  1. Micron FY2026 Q3 results — revenue, margin, EPS, and Q4 guidance.
  2. Micron + Anthropic strategic agreement — memory/storage supply agreement and AI infrastructure collaboration.
  3. Reuters on OpenAI IPO timing — reported 2027-delay consideration.
  4. Alphabet joining the Dow — June 29 replacement of Verizon.
  5. Nasdaq 2026 holiday calendar — July 3 Independence Day observed market closure.
  6. Reuters on Apple price hikes — memory-cost pressure hitting device makers.
  7. Reuters on Honeywell Aerospace spin-off — June 29 aerospace separation watch.
§ Related reading
§ The frontier continues.

Get the next issue when it drops.

One sharp dispatch from the frontier: what changed, what matters, what I'm watching, what the evidence changed — and what the portfolio is doing about it.

No spam. No daily market sludge. Research notes only. Unsubscribe whenever.