Skip to main content
The Frontier File · Issue #10 · July 14 2026

Macro Gave Permission.
Execution Chose The Winners.

One concentrated frontier-tech portfolio, documented in public. Inflation cooled, yields eased, and the market finally exhaled—then immediately reminded everyone that a friendly macro tape cannot save a weak business.

“The number opened the door. The companies still had to walk through it.”

+0.9%
Nasdaq close
duration caught a bid
+4.9%
Micron rebound
trimmed, not abandoned
-25%
IBM reality check
macro cannot hide execution
Cold Open

The quiet before the number

There is a particular kind of silence before a CPI print when you own exactly the kind of companies a hot number can punish. The account looks calm. The quotes move a little. Everyone pretends they are waiting rationally. Meanwhile, every long-duration stock in the book is standing near a trapdoor connected directly to the ten-year yield.

I did nothing. That was the plan.

Then the number came in softer than feared. Yields eased. The Nasdaq climbed 0.9%. Micron bounced almost 5%. For about five minutes, the whole tape felt simple again: inflation blinked, growth breathed, the frontier reopened.

And then IBM fell 25%.

That was the better headline. Not because IBM matters more to my portfolio than CPI, but because it clarified the market I actually own. Macro gave every growth company permission to rally. Execution decided who deserved it.

Macro gave everyone permission. Execution decided who deserved it.

That is the story of Issue #10. Not “CPI was good.” Not “AI is back.” Something more useful: the market is still willing to pay for the future, but it is getting much less patient with companies that cannot produce evidence on the way there.

The Thesis Board

Where my head is at

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

Memory & Storage MU · STX
Still the cleanest evidence-backed thesis in the book. Pricing remains strong, supply remains tight, and both positions sit directly in the path of AI infrastructure demand. The boring hardware keeps acting less boring every week.
Quality AI Infrastructure NVDA · GOOGL · BWXT · RKLB
The shared trait is not hype. It is visible execution: cash flow, contracts, backlog, capacity, customers, or hard-to-replicate assets. This is increasingly the part of the frontier I trust most.
SPCX large enough · still unproven
No adds. The index catalyst is behind it, the position is already meaningful, and the business still needs to earn the valuation in public. Twelve percent is enough exposure to be right without volunteering for more pain if I am wrong.
Rates / Duration SOFI · JOBY · ACHR · QBTS
Cooler inflation widened the runway. It did not finish the work. Lower yields help the valuation math; they do not certify aircraft, fix unit economics, or turn a seed into a core holding.
Power & Physical AI BWXT · OKLO · CENX
Still the next research lane. AI capex is becoming a power, cooling, transmission, storage, and materials story. I want the companies with real scarcity or backlog—not every ticker that learns to say “data center.”
Oil / Geopolitics the thermostat
A soft CPI can be reversed by an energy shock. Oil stayed elevated because Iran risk stayed live. The inflation story improved today, but the Strait of Hormuz still has a hand near the dial.
The Day in One Tape

Relief, with conditions

MarkerClose / ReadMove
S&P 500 soft CPI + bank strength
7,543.59
+0.4%
Nasdaq duration led
26,107.01
+0.9%
Dow IBM swallowed the rally
52,508.27
+0.02%
June CPI headline cooled
3.5% YoY
below fear
Micron memory shortage bid
$983.12
+4.9%
IBM execution warning
-25%

The day did not say “buy everything.” It said the discount rate matters—and then asked every company for receipts. That is a healthier market than a blind melt-up, even if it makes stock-picking less forgiving.

The Portfolio Report

The whole book, through the new lens

One important note: the portfolio marks remain based on my latest supplied ledger rather than a connected brokerage statement. The shape is more trustworthy than the exact closing total.

SPCX
~12%
No adds · prove it
MU
+250%+
0.79 sh · right-sized
STX
Winner
Storage scarcity
RKLB
Leader
Execution anchor
GOOGL
Core
Platform + cash flow
NVDA
Core
AI compute
BWXT
Core infra
Nuclear + defense
CENX
Industrial
Materials + power
ACHR
Growth
Milestone-dependent
JOBY
Growth
Certification watch
OKLO
Seed
Keep small
QBTS
Seed
Proof still needed

The trim did not become wrong

I could feel the regret arrive almost immediately when Micron jumped. Half the shares, half the bounce. The old position would have made more money today.

But that is not how the decision gets graded. I did not trim MU because I knew the next candle would be red. I trimmed it because one company had become too capable of controlling the entire account. I kept half. I kept the thesis. I reduced dependence.

A disciplined trim is allowed to look early. That is usually the cost of discipline.

The portfolio's real personality

The pattern is becoming too obvious to ignore: my best ideas are not necessarily the loudest ones. They are the companies sitting at a real bottleneck with visible execution.

Scarce infrastructure plus receipts. That is the portfolio's emerging center of gravity.

Core
Visible cash flow or execution
Growth
Milestones can expand value
Seeds
Small until proof arrives
1 rule
Excitement never sets size
Three Moves / Watches

No heroics required

1. Freeze SPCX sizing
No averaging down, no adding because the index inclusion “should” help, and no emotional rescue mission. The position already matters. Let the business and tape do the next piece of work.
2. Let MU and STX run without rebuilding the old imbalance
The memory thesis strengthened. Keep the exposure. Do not chase MU back toward 20% because the trim feels early. Bullish thesis, controlled size.
3. Use ASML and TSMC as proof filters
Watch orders, advanced-node demand, packaging constraints, capacity, pricing, and hyperscaler spending. The next question is not whether “chips go up.” It is whether the physical AI buildout remains capacity-constrained and economically real.
The Catalyst Calendar

The market gets more evidence immediately

Jul 15
PPI, Beige Book, ASML, and major financial/industrial earnings
Rates + chips
Jul 16
TSMC earnings — demand, capacity, advanced nodes, and AI capex
Critical
Next 2 weeks
Watch SPCX for a genuine base, not a one-day relief bounce
My watch
Jul 28–29
FOMC decision — softer CPI versus oil-driven inflation risk
Macro
Ongoing
Iran / Hormuz headlines and oil's effect on inflation expectations
Tail risk
The Hot Take

“A soft inflation print can save a multiple. It cannot save a weak business.”

That was the whole day in one sentence. The Nasdaq rallied because cooler inflation lowered the discount rate applied to future earnings. IBM still lost a quarter of its value because the company-specific story cracked.

Macro decides the weather. Execution decides whether the roof leaks.

That is why I want to keep leaning toward infrastructure companies with real scarcity, contracts, customers, capacity limits, and backlog—and keep aspiration-priced positions small enough to survive the days when the market asks for receipts.

Buddy, Real Talk

The dangerous part was not the trim

The temptation this week is to tell myself I should have kept all the Micron. That is the easy regret because the missing upside is visible.

The harder regret is less visible: I finally created almost a thousand dollars of flexibility and immediately spent nearly all of it on the rocket. The trim was risk management. The redeployment was appetite. They happened in the same click, but they were not the same decision.

I still think SPCX can work. I also think I skipped the boring pleasure of having cash.

That is probably the most “me” lesson in the whole file. I am getting better at cutting concentration. I am still learning not to fill every empty space with a new adventure.

No shame. No spin. Just the next thing to improve.

The Bottom Line

The market got oxygen. Now the companies need lungs.

Inflation cooled. Yields eased. The Nasdaq rallied. The frontier got a better weather report.

But oil stayed hot, geopolitical risk stayed alive, and IBM reminded the market that a friendly macro tape cannot protect a weak operating story. From here, the winners need more than duration. They need evidence.

Memory still has scarcity. Seagate still has demand. Rocket Lab still has execution. Power still looks like the next bottleneck. SPCX still has something to prove. And I still have to learn that cash is allowed to exist without immediately becoming a position.

The tenth issue does not end with a prediction. It ends with a standard:

Every large position needs receipts. Every seed needs a milestone. Every trim deserves enough time to prove why it existed.

Macro opened the door. Execution chose the winners. I am trying to build a portfolio that deserves to walk through.

See you next Sunday, buddy. — Alex

§ 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.