Pabrai Investment Funds / Junoon · value investor in the Buffett-Munger mold — low-risk / high-uncertainty bets, quality compounders, deep concentration — running synthesis of his appearances, with per-transcript breakdowns and a stock index.
The set-and-forget alternative to an S&P 500 index fund and his answer to "what would you do with $1,000?" — "no leverage, lots of cash, very good management, very deep moat, very boring, very hated and unloved."
Don't value on PE (~40×, distorted by non-cash goodwill charges); on free cash flow it's "nowhere near" that. Worth ~10-17× FCF for the ~2-3% organic growth alone, plus a second engine redeploying cash into acquisitions at ~4-5× FCF (a ~20-25% reinvestment rate). His #1 management team — "kick-ass DNA," shareholder-oriented, disciplined on capital return.
His hold-the-compounder exemplar: never sell "a Costco or a Coke or a Visa or a MasterCard or Amex… unless they're egregiously overpriced." Calibration: ~50× trailing has never been egregious for Costco; ~250× trailing normalized would be.
His flagship idea and #2 management team/individual across the portfolio, purely on the track record. "The WeChat of Kazakhstan" — a super app cash-flowing $2B a year at five-to-seven times cash flow with a dividend yield "approaching 10%," run by a rockstar CEO who owns ~40-43%. The Turkey build-out is a free moonshot: if it fails you still make 2-3×; if it works the upside is open — "Heads I win, Tails I win."
His hold-forever Turkish emblem — the largest warehouse builder/landlord in Turkey (~1.2M m², 99% inflation-indexed blue-chip leases). Bought at a ~$16M cap vs ~$800M liquidation value; now ~$1.5B. Owns 40%: "we're just going to own it forever."
Core of the three-player memory oligopoly — "providing the pickaxes in a gold rush," near-impossible to enter (patents + "black magic" fab know-how). Hold, don't chase: "own it? don't sell; don't own it? don't buy — the party has only just started." Regrets selling his own stake.
A "fabulous business" and export powerhouse, but squeezed by Korea's demographic/labor-cost headwind — "cheaper to build a car in Alabama than in Korea," so production keeps moving abroad. Quality operator, structural drag.
One of three memory makers dominating the KOSPI; a once-brutal 20-player commodity war consolidated into a protected three-player oligopoly with a real moat. Same hold-don't-chase framing as SK Hynix.
His cleanest too-hard-pile demonstration: the -63% drawdown doesn't matter — without high conviction on minimum cash flows over 5-15 years, "we move on." "Adobe may be a no-brainer for one person and a too-hard pile for another. To me… it just goes in too-hard pile."
Berkshire's ~$30B position was Buffett's own call (a front-row seat via GEICO's ad spend, plus a pre-IPO visit from Sergey and Larry) — but it only fits because his universe is tiny, and Google is "now a company with debt" and high capex for the first time. Observer's stance, not a pick.
The US leg of the three-player memory oligopoly; management says even with all patents and engineers they're unsure they could rebuild a burned-down fab at the same cost ("black magic"). He owned it, broke his never-sell rule and regrets it; hold-don't-chase.
His costliest mistake of omission: owned ~1% of Ferrari inside his 2012 Fiat Chrysler stake, sold at the ~2016 IPO; it later grew to a ~$100B cap. "I'd have about a billion more if I hadn't done that stupid thing" — the lesson to never sell the multibagger.
Split verdict: the business is "phenomenal" and Elon is "superhuman" — "do not short a superhuman" — but as an investment it goes in the too-hard pile. Earlier he named it the crowd's next "shiny object" after Bitcoin and AI.
Korea's premier skincare name — done extremely well, but the moat is "a little bit shallow": constant rival attack and customers always wanting the next, better treatment. A great product without a durable brand moat "may not make it."
"Somewhat risky as an investment": everything emanated from Jobs, gone a long time ("nothing new since he left"), plus form-factor risk — the pocket "brick" will be replaced by something more ergonomic, "probably some guy in a garage," more likely not Apple.
Automatic pass on governance: with the founder gone, "40 or 50% of the shares that they buy back end up in the pockets of the managers." Excessive comp is a total exit, never a smaller position — "why would you want to be slightly in bed with a crook?"
In one line:Few bets, large bets, infrequent bets — placed only where the odds approach no-brainer territory, and everything else goes on the too-hard pile, used aggressively ("everything goes in the too hard pile; it's only the anomalies that don't"). Buy outstanding businesses run by outstanding, shareholder-aligned managers on the right yardstick — free cash flow, not headline PE — structure every bet so "heads I win, tails I don't lose much" (better: tails I win), then hold ideally forever. On the market as of Aug 2026: the S&P is "at the worst case not a no-brainer, and more likely ridiculously overvalued" and AI is unplayable — so BRK.B is his index substitute and KSPI (Kaspi.kz) his flagship idea, alongside Constellation Software and Turkey's Reysas.
Few, large, infrequent bets — anything murky is an automatic pass. Bets are only placed where the odds are heavily in his favour, approaching a no-brainer; "anytime things become murky or debatable, it's an automatic pass." There is no half-size compromise, and no obligation to hold a view on what he'll never buy ("why ask the address of a home when you're never going to visit?" — the S&P, the passive-bubble question, private credit).
The index isn't a no-brainer here — buy Berkshire instead. He'd normally tell a young investor to dollar-cost average into an index, but not at this S&P. BRK.B is the substitute: ~40% cash, no leverage, good listed + wholly-owned businesses, "fairly priced or underpriced but probably not overpriced," plus dislocation optionality under Greg Abel ("a double in a few years"). Don't distribute the cash — a dislocation "might wipe out all that cash." Buffett-vs-Abel is the wrong worry; size is the 800-lb gorilla.
AI is a too-hard pile — even the pickaxe sellers. "Two or three players make out big time and then there's a lot of carcasses on the roadside," and nobody knows which is which. Hyperscalers are forced players (Zuckerberg: "whether the bet works or not, we have to play"), and headline capex is inflated ~4-5× by input prices, not volume — so MU and the memory oligopoly, despite a genuine "black magic" fab moat, still fail the only question that matters: where is this in three to five years?
Heads I win, tails I win. Value the provable core alone against the price; if the core is worth 2-3× on its own, the expansion is a free option and the wait is paid by a dividend. KSPI is the archetype: a Kazakh super app at 5-7× cash flow yielding ~10%, plus a Turkey moonshot into a market 8× larger — "if the moonshot doesn't work at all you make two or three times your money." Turkey's Reysas (40% owned, ~$16M → ~$1.5B) is the older emblem.
Source it simply, then test it simply. Read every write-up on Value Investors Club "till something hits me in the head with a 2 by 4"; the write-up covers the company and the operator. Then the filter: explain it to a 10-year-old in four sentences, and only swing at a watermelon — "there are no called strikes," so never invest anywhere until you're "5,000% all in."
Management integrity and capability are non-negotiable — judged on the track record. Ignore the forward plan; read the last 5, 10, 20 years (he's never spoken to Kaspi's CEO). Excessive comp is a total exit, never a smaller position — NVR hands managers 40-50% of its buybacks: "why would you want to be slightly in bed with a crook?"
Never sell the compounder unless it's EGREGIOUSLY overpriced. Deep moats are built by accident and are rare (Visa, Mastercard, FICO, Moody's, Amex, Ferrari); don't sell "a Costco or a Coke or a Visa… unless egregiously overpriced" — ~50× trailing never has been, ~250× normalized would be. His stated regret: "20% Ben Graham, 80% Fisher-Munger," realized only 7-8 years into a 32-year career.
Transcripts
One dated page per appearance — each has its full stock table, talking points, and the saved transcript. Newest first.