← All reports THE JEFF REPORT CASE FILE 029 · GLOBAL · 16 JUN 2026
Jeff·Financier
Model basket
Portfolio Construction · AI Supply-Chain Basket · 16 Jun 2026

The $10,000 Chokepoint Portfolio

Sixteen companies that own a physical chokepoint in the AI hardware stack — optics, cooling, tungsten, rare earths, laminates — each valued the full way: size, price-to-sales, margins, growth, market share, the KPIs that actually move it, and the recent run-up. Then one question answered: for an illustrative $10,000 model sleeve split across them, what's the best allocation, and why.

NAMES VALUED · 16 INVESTABLE FROM USD BROKERAGE · 11 PRICES · last close, 13–15 Jun 2026, USD
01 — The Recommendation

The best $10,000 split

This is a hypothetical model allocation, not a live position — drag any slider to build your own; the donut and total update instantly. The default below is the recommended weighting. Everything in this basket has already run 100–1,500% in a year, so the build leans on the profitable, cash-generative toll-booths and uses the scarcity-miners as smaller, convex satellites.

$10,000
allocated
Total allocated$10,000
FIG. 029-A · model allocation, illustrative only

Only the 11 names buyable from a standard USD brokerage carry a live weight. The four mainland-China A-shares and the Hong Kong fibre name are valued in full below but excluded from the dollar allocation — see the access note.

Why this shape. Two-thirds sits in profitable, AI-demand toll-booths whose pricing power is contracted, not hoped-for — VRT (a $15B backlog you can underwrite), COHR (the 6-inch InP monopoly), LITE (the EML monopoly), GLW (the fibre backbone) and AVC (≈70% of AI cold plates, and the cheapest multiple in the group). The remaining third is the diversifying scarcity sleeve — PPTA (trading near book NAV, fully funded, the cleanest risk/reward), plus ALM / MP / LYC / UUUU for the tungsten and rare-earth chokepoints that move on different catalysts than optics. Rogers (ROG) is left at zero — a real PTFE moat, but its data-centre revenue is ~18 months away and it has already run +153%; revisit nearer $110–120.
02 — The Screener

All sixteen, side by side

Click any column header to sort. Market caps and prices in USD; P/S on trailing sales; 1-yr is the total return into the recent run-up.

Ticker Company Price Mkt cap P/S Rev gr. 1-yr Conv. Verdict Access $ split

Access: YES standard USD brokerage · LTD via Taiwan/ASX/HK or OTC with friction · NO mainland A-share, not practically buyable.

03 — Optics & Cooling

The profitable toll-booths

The core of the book: businesses with real revenue, expanding margins, and demand that's contracted out to 2028. Click any card for the full valuation, KPIs, the run-up, and the recent tape.

04 — Ex-China Materials & Miners

The scarcity sleeve

Lower correlation to the optics trade — these move on export controls, mine commissioning, and government offtake. Higher variance, sized as satellites. Includes Rogers (defense-grade PTFE), held at zero on valuation.

05 — The Leverage-Holders

Valued, but you can't easily buy them

The Chinese bottleneck names have the strongest fundamentals in the whole set — Innolight and Eoptolink run 30–37% net margins. But they're mainland A-shares (or a +1,459% Hong Kong fibre name), not practically buyable from a USD brokerage. They're here for completeness, with the accessible proxy that captures the same cycle.

The access reality — why these sit at $0 in the split

Mainland China A-shares (Shengyi, Envicool, Innolight, Eoptolink) trade in renminbi on the Shanghai/Shenzhen exchanges. In principle they're reachable through Northbound Stock Connect, but in practice a standard USD retail account can't place a normal buy: it needs Connect explicitly enabled, RMB settlement, and — for the ChiNext names — additional eligibility. There are no ADRs. YOFC's Hong Kong line (6869.HK) is reachable via an IBKR HK route, but it's up ~1,459% in a year and trades below its own analysts' targets.

The cleaner move: their exposure is already in the book. Owning COHR + LITE captures the same optical-transceiver cycle that drives Innolight/Eoptolink; VRT captures the cooling cycle that drives Envicool; GLW captures the preform cycle that drives YOFC. The theme comes through in dollars, with liquidity and disclosure that can be trusted. For direct A-share beta, a China-tech ETF (e.g. CQQQ) holds several of these at small weights.