AEA Capital Research · Strategy Briefing
Q3 2026 Strategy Briefing
A fixed Q3 strategy snapshot: macro positioning, the semiconductor/AI-infrastructure thesis, portfolio positioning at the review date, and a personal read on the quarter ahead.
September update
This briefing preserves its quarterly snapshot instead of silently rewriting history. For the live September 10 portfolio, current prices, the SNDK policy breach, and the next decision dates, read the September checkpoint →
Macro Positioning
The latest available macro readings through the September 18, 2026 close. Market levels and slower official series are dated separately so the table does not imply they share one timestamp:
| Metric | Level | Read |
|---|---|---|
| 2Y Treasury | 4.67% (Sep. 16) | The curve remains positively sloped, but the 10-year yield returned to 5.00% Friday, keeping financial conditions tight after the Fed’s rate increase. |
| 10Y Treasury | 5.00% (Sep. 18 close) | |
| Fed Funds Target Range | 3.75–4.00% | FOMC unanimously raised the range 25 basis points on September 16, 2026. |
| Core CPI (YoY) | 2.4% | August 2026 reading, down from 2.5% in July — still above the Fed's 2% target. |
| HY Credit Spread (OAS) | 2.71% (Aug. 25) | Last verified ICE BofA snapshot retained with its date; not represented as a live September 18 reading. |
The long end above or near 5% is no longer a permissive backdrop for a growth-tilted, low-income book. It raises the discount rate applied to future cash flows and makes rate sensitivity a near-term risk alongside the positioning issues on Page 4. Full context is maintained on the homepage's Yield Curve & Credit monitor.
Semiconductor & AI-Infrastructure Thesis
This remains the book's dominant single theme — 29.80% of the portfolio sits in Semiconductors alone (SNDK, AMD, MU, INTC, AVGO, ARM), with more spread across AI/Cloud Infrastructure and Power & Utilities. The standing thesis across the five largest names in the theme, quoted directly from each security page:
SNDK — largest position, 9.48% of book
"Sandisk is one of the five largest NAND flash suppliers globally, vertically integrated through a manufacturing joint venture with Kioxia, and now free to price and invest on its own since its 2025 spin-off from Western Digital. Storage is an under-discussed leg of the AI buildout." Full thesis →
MU — 5.61% of book
"Micron is one of the largest memory and storage chip makers globally... still trades at a reasonable 23x trailing earnings despite being up sharply on cost — cheap relative to how central high-bandwidth memory is to the AI buildout." Full thesis →
AMD — 7.43% of book
"AMD is a leading designer of CPUs and GPUs for PCs and data centers, and is emerging as the closest thing to a credible second source for AI accelerators through its MI-series line." Full thesis →
INTC — 4.14% of book
"Intel remains the CPU share leader in both PC and server markets... but the real thesis is a turnaround bet: whether Intel Foundry can land real external customers and catch process-node execution up to leading-edge competitors." Full thesis →
NBIS — 1.60% of book
"Nebius is a vertically integrated cloud provider for AI and HPC, operating its own data centers across Europe and the US — real infrastructure, not just a story about future AI demand." Full thesis →
Portfolio Positioning
No trades this quarter constitute a reallocation — positioning below is the current, unchanged state of the book, not a proposed change to it (a personal view on possible near-term changes is on Page 4).
Return contribution — today's weight × total return — is concentrated in a small number of names, led by SNDK (+22.71 pts), META (+11.97 pts), AMD (+11.74 pts), NBIL (+11.31 pts), and MU (+11.12 pts). Full ranked list and methodology caveats on the Holdings page.
Q3 Outlook — Personal View
Everything on the first three pages is checkable data. This page is different — it's my own read heading into Q3, stated as opinion because that's what it is.
I'm leaning more aggressive, not more cautious, heading into Q3. The macro backdrop on Page 1 is permissive, the core semiconductor/AI-infrastructure thesis hasn't broken, and I'd add to the book's highest-conviction names if the right setup showed up — I'm not looking to de-risk from here.
The scenario I'm most focused on isn't sector-specific bad news — it's a capital-expenditure digestion cycle across the Magnificent Seven. If a hyperscaler at Meta's scale begins actively monetizing excess compute capacity rather than continuing to build it out, that's a signal AI-infrastructure spending outran near-term demand. NVIDIA's own recent multiple compression is the template I'm watching for: consensus-beating earnings no longer move the stock the way they once did, because the market has shifted its attention from "how fast is this growing" to "is this capex spend sustainable." If that pattern migrates downstream to the memory suppliers this book actually owns, SNDK and MU could face the same growth-without-re-rating dynamic — real earnings growth the market simply stops rewarding once the capex narrative cools.
Nothing below is reflected in the Holdings page yet, and none of it is a decision — it's what I'm actively thinking through heading into Q3. SNDK's weight has swung on both sides of its 10% IPS cap purely on price — it fell under the line on a pullback, and the August rally has carried it back to 9.48%, close enough that another good week puts it over again. It stays flagged Under Review on Holdings rather than treated as settled: a cap that resolves itself by price movement isn't the same discipline as trimming into it deliberately, and this is exactly the position where that distinction keeps mattering. I'm still evaluating rotating some capital into enterprise software and adjacent growth names not fully represented in the book today — adding to NOW, and taking a first look at names like CRM and BE — alongside a possible add to AMD, which I still see as the more credible long-term second-source story in AI accelerators versus the memory names that would be most exposed if the capex-digestion risk above plays out. If any of this becomes real, it'll be reported in the next letter and reflected on the Holdings page the moment it happens, not before.
Risk Disclosures & Sources
AEA Capital Research is an educational project by Aydin Ali and does not constitute financial advice or fiduciary investment management. Nothing in this briefing is a recommendation to buy or sell any security. Page 4 in particular is explicitly labeled as personal opinion, not analysis — treat it accordingly.
Where each page is sourced from
- Page 1 (Macro): September 16–17 Treasury market reporting; August 2026 CPI from the U.S. Bureau of Labor Statistics; the September 16 FOMC statement; and the last verified ICE BofA HY OAS snapshot, explicitly dated August 25. Full detail on the homepage.
- Page 2 (Thesis): Quoted verbatim from each position's own Investment Thesis field — see individual security pages for full bull/base/bear cases and sourcing.
- Page 3 (Positioning): Computed from real position weights and returns via Blossom; full methodology and caveats on the Holdings page.
- Page 4 (Outlook): My own stated opinion as of the compile date above, not derived from any external source.
What this briefing is and isn't
This is a reading aid and a personal outlook statement, not a new disclosure or a trade confirmation. If a number on Pages 1–3 ever conflicts with the page it's sourced from, the source page is correct and this briefing is stale — treat it as a snapshot compiled on the date above. This document is rebuilt each quarter; prior quarters are not currently archived separately.