Contrarian Quality at GQG Partners – Rajiv Jain | Capital Allocators with Ted Seides

ELI5/TLDR

Rajiv Jain built GQG from scratch in 2016 and somehow grew it to $160 billion in a decade where almost every active manager bled assets. His one trick: own “quality” — but he defines quality as barriers to entry, not whatever Wall Street is currently in love with. Right now that means he owns almost no AI chips and no Mag 7, and instead owns energy, utilities, steel, tobacco, and emerging-market banks. He thinks the AI boom is real technology but terrible economics, and he’d rather be early and dull than crowded and exposed.

The Full Story

A trader before he understood what a stock was

Jain started in 1980s India, matching dividend checks against his father’s physical stock certificates. Then he just started trading — “classic punting, not much thought.” The cold-call origin story is almost comically pre-internet: fresh out of University of Miami, no contacts, he picked up the CFA directory and dialed senior CIOs because he figured mid-level people were too busy. One of them answered, said “nobody calls me, what’s your story,” and gave him his break. He bought a fax machine and kept a spreadsheet of who was interested.

The formative detail isn’t the trading. It’s a boring job doing export paperwork in Delhi — bills of lading, where one mistake gets the whole document rejected by the bank.

“Things have to be a little more precise manner than simply putting in a model or you grow at 15% next 5 years and we all live happily ever after.”

That allergy to the happily-ever-after spreadsheet runs through everything he says.

Crisis as a teacher

Jain has a knack for starting jobs right before things explode. He joined Vontobel as an EM co-manager on October 31, 1994 — six weeks before the Tequila crisis. Became EM portfolio manager in January 1997, six months before the Asian crisis. Became CIO in January 2002, and promptly watched 75% of clients fire the firm.

Each crisis sanded down a belief. The Asian crisis killed his faith in top-down country models — the macro call didn’t work; only the bottom-up balance-sheet work kept him alive. So top-down became a risk-off switch, never a reason to buy:

“If Chinese growth is good and inflation is good, you don’t buy China because of that. You still need valuations and corporate earnings.”

2008 taught the deepest lesson. He’d smartly exited all his bank exposure in early 2007 — then loaded up on energy, believing the “decoupling” story. Lehman fell, the market briefly recovered, and then energy melted by October, down more than half in weeks. He’d seen the financial crisis coming and still got wrecked, because he didn’t connect the dots on how it would ripple. The takeaway:

“Relative is fine in an up market but over the long run if you don’t have absolute returns nobody needs you. You don’t pay bills with relative performance.”

Hiring people who can’t agree with you

The most distinctive thing GQG does flows directly from 2008. Jain noticed that the mortgage crisis had been on the cover of BusinessWeek a year and a half early — the press saw it, Wall Street was “in complete la-la land.” So he started hiring journalists — former investigative reporters — as analysts whose job is to take the opposite side by default. Half the research team is journalists.

The clever part is the incentive design. If you pay people to agree, they’ll agree.

“You want to structure the compensation where actually by default they cannot agree with you.”

It generalizes into his whole team philosophy. He deliberately did not bring his old Vontobel team to GQG, because a team that grew up with you for 15 years will never disagree with you. He hired hedge-fund people with 10-15 years of long-short scars instead — maximum chance they’d push back. No personal single-stock trading allowed; his own net worth sits in the same vehicles clients buy. Fees deliberately below median (“a Costco model”). And he split the jobs cleanly: he runs ~20 investment people, his partner Tim runs the other ~220 that are the actual business.

Quality means barriers to entry — and it’s forward-looking

GQG stands for Global Quality Growth, so the obvious question is what “quality” means. His answer is unfashionable: not brands, not margins, not “compounders” — barriers to entry.

“If this building was on the beach here… there are no commercial office space in Fort Lauderdale on the beach. That’ll be extremely valuable.”

Crucially it’s forward-looking quality. A business can move along that spectrum. Energy pipelines have become high-barrier — try getting one approved. Steel plants in Europe — “your grandkids might get approval.” Meanwhile semiconductors are getting lower quality, because the Chinese are ramping capacity, and his rule is blunt: if the Chinese are a competitor, they’ll overproduce and kill you. Software is mostly low-barrier — how many software companies have survived 30+ years? Microsoft, maybe Oracle.

Cyclicality, he insists, is not the enemy. Every business is somewhat cyclical. What kills you is paying a high multiple on peak margins of a cyclical — he’s done it, with Japanese names that went 5% margins → 25% → back to 5%, where he’d paid 50x earnings.

The AI bet against the consensus

This is the loud, uncomfortable part. GQG owns almost nothing in semis or Mag 7 today, having once had 40%+ in tech, and Nvidia was the single biggest profit contributor in firm history. It’s cost them — “painful last 12 months.” The thesis:

“The cumulative capex of all these mag companies in their history is $1.5 trillion… now they’re talking about 3 trillion in just 3 years… you’re spending a trillion dollars a year and the revenue on AI talk about maybe 70 80 billion.”

The mechanics he points to: Google’s clean free cash flow shrank to ~$10bn as capex outran cloud revenue; cloud itself is now low-quality (200+ providers, customers shop on price); digital advertising is near saturation; Nvidia’s “free cash flow” is partly recycled into investments in its own customers. And the demand looks subsidized — “if Starbucks starts selling coffee at 25 cents, there’ll be shortage of Starbucks coffee.” His verdict: powerful technology, bad economics, and time is not a friend.

He’s careful not to be dogmatic. They exited tech in 2021, then re-entered in 2023 when they saw ChatGPT’s early impact. They have no problem buying back in — they just need to see a true killer app and real enterprise adoption, and right now hallucinations are the blocker. Goldman, he notes, ran AI systems in parallel for years because “you can’t say oh it may be 80%.”

Where he’s actually putting money

Energy (double-digit free-cash-flow yields at $75-80 oil, no one wants to add capacity). Utilities (the world has underinvested in power; 8-10% visible EPS growth, faster than the S&P). Tobacco — over five years Altria outperformed Meta, because cash generation compounds. Emerging-market banks — a Brazilian family-owned bank at ~7-8x earnings with a 7% dividend that’s earned 15% real ROE for 30 years. The unifying screen is a 9-11% expected return assuming no multiple expansion. If the math needs the multiple to hold up, he passes.

Don’t blow up

Position sizing is borrowed from credit analysts: how would S&P rate this? A monoline E&P company can never get a AAA, so it can never be a top position — Exxon can be large, Occidental (which he likes more) can’t, because it’s narrower and riskier. Sizing tracks stability, not conviction. The whole portfolio is built around one rule: don’t blow up.

“What is not acceptable is market down 40 and we down 43, we outperform.”

Portfolios are concentrated — US book ~30-35 names, top 10 about half. He’s underperformed for 12 months but hasn’t lost money, still compounding mid-teens, and he’s quick to remind that losing money in down markets is the real crisis, not lagging in an up market.

The closing wisdom

His pet peeve is dogmatism — “I try to avoid those people because I know they’ll convince me with facts.” He talks to his 87-year-old father two or three times a week (“some things you only learn with age”). And the recurring theme of a long career: humility.

“The biggest thing is that you begin to appreciate how little you know. You become humbler because the conviction level actually goes down.”

Key Takeaways

Quality = barriers to entry, and it’s forward-looking — a business can become higher- or lower-quality over time (energy pipelines up, semiconductors down).
Top-down macro is a risk-off switch, never a buy signal. Good macro doesn’t justify buying; bad macro can justify caution.
Pay analysts to disagree. GQG staffs half its research with former investigative journalists, compensated on their independent calls, not on agreement.
Hire people who can’t think like you — Jain deliberately avoided his old team and recruited 10-15-year hedge-fund veterans for maximum dissent.
Absolute returns over relative. “You don’t pay bills with relative performance.” Underperforming is survivable; losing money in a down market is not.
Position sizing mimics a credit rating. A narrow/monoline business can never be a top weight regardless of conviction; sizing tracks stability, not enthusiasm.
Cyclicality isn’t low quality — but paying a high multiple on peak-cycle margins is the classic way to lose your shirt.
The AI math: ~$1.5T cumulative Mag-7 capex historically vs. ~$3T proposed in three years, against maybe $70-80bn of AI revenue. Powerful tech, bad economics.
Subsidized demand isn’t real shortage — “if Starbucks sells coffee at 25 cents, there’ll be a shortage.” Test demand at proper pricing.
Lower the hit rate on purpose. A high hit rate sets the bar too high and makes you miss multibaggers; the best ideas are the ones you doubt, “because if I have doubt, the world has doubt too.”
You only learn a stock by owning it — being underwater in a name sharpens thinking; he’s lost money in every area over time, which makes him a better analyst on each.
Cash compounding beats narrative: over five years Altria (declining-volume tobacco) outperformed Meta.
GQG facts: founded 2016, ~$160bn AUM, listed in Australia in 2021 (75% insider-owned), four core products, US book 30-35 names with top 10 ≈ half, four PMs with equal votes, only one team meeting a week.

Claude’s Take

This is a genuinely good listen, and the score reflects substance over polish. Jain is the rare manager whose process descriptions aren’t just retrofitted to recent performance — the “hire journalists who can’t agree with you” structure, the credit-analyst sizing, the absolute-return discipline all hang together as a coherent system that predates his current positioning.

The BS filter does need to come on for the AI bear thesis, though. He’s talking his book — GQG is down a year of relative performance precisely because of this call, and a long interview gives him a free platform to explain why the market, not he, is wrong. The capex-vs-revenue framing is rhetorically strong but a touch slippery: comparing cumulative historical capex to forward three-year capex, and current AI revenue to total proposed spend, is exactly the kind of apples-to-oranges that makes any infrastructure buildout look insane mid-construction. That said, he hedges honestly — he re-entered tech in 2023, says he’ll buy back in on a real killer app, and frames it as “bad economics, not bad technology.” That’s a defensible position, not a permabear pose.

What earns the 8 is the meta-level honesty: a manager whose central message is “I know less than I used to, build a team that can prove me wrong, and don’t blow up” is worth more than ten managers with crisp ten-bagger stories. The bits on incentive design (you get the behavior you pay for) and on deliberately lowering your hit rate are transferable well beyond investing. Docked from a 9 only because a chunk of the runtime is a single contrarian macro call delivered with more confidence than the data strictly supports.