The Legendary Debate

Warren Buffett: leans back and folds his hands "Charlie and I have a simple test — if you gave us a billion dollars and said 'go compete with Copart,' could we win? I've been thinking about this one for a while. You'd need to acquire hundreds of parcels of land near every major metro area in America — land that's increasingly impossible to zone for salvage operations. You'd need to build a global buyer network spanning 190 countries that took Copart two decades to assemble. And then you'd need to convince State Farm and GEICO to hand you their wrecked cars when they already know Copart's auction gets them higher recovery values. I don't think a billion would do it. I'm not sure ten billion would do it. So let's start there — is this moat real, and more importantly, is it widening or narrowing?"

Charlie Munger: "The moat is real, and I'll tell you exactly why by inverting the question: what would have to happen for Copart to lose? An insurance carrier would need to say, 'I'm going to accept lower recovery values on my total-loss vehicles, anger my policyholders with slower cycle times, and build my own auction platform from scratch — all to avoid paying Copart's fees.' That's irrational behavior, and in my experience, insurance companies are among the most rational economic actors in the world because their entire business is pricing risk correctly. The more interesting question is whether IAA under RB Global's ownership can narrow the gap. And here's what I notice — Copart just told us their U.S. insurance ASPs grew 9% year-over-year excluding catastrophe effects, in a quarter where Manheim indices were flat to down. That means the platform itself is creating incremental value through buyer liquidity, not riding a rising vehicle-value tide. That's the network effect compounding in real time, and it's very difficult for a number-two player to replicate because liquidity begets liquidity — the same physics that made eBay nearly invincible in its core categories for fifteen years."

Dev Kantesaria: "I want to challenge this toll booth characterization directly, because precision matters in how we categorize businesses. A true toll booth — Visa, FICO, Moody's — has a characteristic that Copart lacks: there is literally no alternative pathway. Every credit card transaction must flow through a payment network. Every mortgage must pull a FICO score. But an insurance company processing a total loss can choose Copart or IAA. That's a duopoly, not a toll booth. Now, duopolies can be excellent businesses — Visa and Mastercard are a duopoly — but the critical difference is whether the two players compete on price or coexist on value. In payments, Visa and Mastercard don't compete on interchange rates. In salvage auctions, Copart and IAA compete directly for carrier contracts, which means margins are ultimately constrained by competitive dynamics. The 80-90% market share claim in the thesis needs scrutiny — IAA processes meaningful volume, and RB Global has the resources to invest in their platform. I'm not saying Copart isn't a good business. I'm saying calling it an 'inevitable' toll booth overstates the structural position and makes you overconfident about terminal margins."

David Tepper: "Dev, I hear you on the taxonomy, but you're letting the perfect be the enemy of the very good. Let me put this in practical terms — I've watched dozens of duopolies over my career, and the ones that work for investors share one trait: the industry structure makes it irrational for either player to start a price war. In salvage auctions, Copart and IAA don't compete primarily on fees — they compete on recovery values for the insurer. That's a fundamentally different competitive dynamic than, say, airlines competing on ticket prices. When Copart wins a contract from IAA, it's because they demonstrated empirically — Jeff Liaw said they have 'before and after returns data' — that their auction generates more money for the carrier. That's value-based competition, and it tends to be stable because the leader's advantage compounds through the network effect. The more buyers Copart has, the higher the prices, the more carriers switch, the more buyers come. IAA can't break that cycle by cutting fees — they'd need to somehow attract more buyers, which is a chicken-and-egg problem they've been failing to solve for twenty years."

Mohnish Pabrai: "I want to bring up something nobody's mentioned yet — the physical infrastructure moat, which I think is actually more durable than the network effect. Copart owns its land. They don't lease, they own. And here's the critical nuance: zoning for salvage operations near urban areas has become increasingly restrictive over the past two decades. Every year that passes, it gets harder — not easier — for anyone to replicate Copart's yard network. This is the same dynamic that made Republic Services and Waste Management such compounders — try getting a new landfill permitted near a major city. It's essentially impossible. So even if some well-capitalized competitor wanted to enter salvage auctions, they'd face a decade-long process of acquiring and permitting land parcels that Copart assembled over thirty years. That's not a moat you can swim across — it's a moat that gets wider every year purely through regulatory and land-use dynamics, without Copart having to do anything clever."

Robert Vinall: "Mohnish raises the land point, and I want to connect it to what I think is the most underappreciated aspect of this business: the reinvestment runway. Copart isn't just milking an existing franchise — they're actively expanding into continental Europe, the Middle East, and Asia, where salvage auction infrastructure barely exists. Think about what that means mathematically. In the U.S., total-loss frequency hit 24.2%. In most European markets, that figure is in the low teens because the infrastructure to efficiently auction damaged vehicles simply doesn't exist yet. As Copart builds yards and buyer networks in these markets, they're essentially recreating the U.S. playbook with a two-decade head start on any local competitor. The capex looks elevated right now — $569 million in fiscal 2025 — but this is the kind of investment spending that creates permanent competitive advantages, not maintenance capex keeping the lights on. The question I'm wrestling with is whether international margins can eventually approach U.S. levels, and I think the answer is yes, but on a 7-10 year timeline, not 2-3."

Pulak Prasad: "I want to apply an evolutionary lens here, because Copart's survival history is genuinely remarkable and tells us something about organizational fitness. This company made the single most consequential strategic decision in its industry when it migrated to online-only auctions in 2003 — years before competitors, in an era when the industry thought physical auctions were sacred. That decision wasn't obvious at the time. It was a genuine adaptation to a changing environment, and it required the organizational courage to cannibalize a working business model. Companies that can do that — Netflix pivoting from DVD to streaming, Amazon pivoting from bookseller to cloud provider — tend to have an adaptive capacity baked into their culture that's extraordinarily rare. But here's the question that keeps me honest: does Copart still have that adaptive gene, or was the 2003 decision a one-time stroke of genius by a founder who's no longer running the business? Jeff Liaw mentioned deploying AI across document processing, driver dispatch, and a total-loss decision tool, which suggests the culture remains innovative. But I'd want to see more evidence before concluding this is a permanent organizational characteristic rather than a historical one."

Warren Buffett: "Pulak raises a crucial point about adaptability, and I want to connect it to Guy Spier's 'away from desk' test. Could I own this business for twelve months without checking the stock price, reading an earnings call, or following any news? For most businesses, the answer is no — I'd worry about competitive disruption, management mistakes, or secular decline. With Copart, I genuinely believe the answer is yes. The physical infrastructure isn't going anywhere. The insurance industry isn't going to stop totaling cars. The buyer network compounds on its own through simple economic gravity. That's the hallmark of a business you can own while you sleep. Now, Dev, I want to come back to your point about this not being a true toll booth —"

Dev Kantesaria: "Warren, let me be precise about my concern. I acknowledge this is a high-quality business — the operating margins, the capital-light model, the network effects are all real. My issue is with the insurance volume cyclicality and what it reveals about inevitability. Copart just reported U.S. insurance units down 10.7%. The CEO attributed this to consumers 'paring back coverage' and 'softer claims activity.' When your volume depends on consumer decisions about insurance coverage levels — decisions driven by macroeconomic pressures, carrier pricing cycles, and regulatory changes — you don't have inevitability. You have cyclicality dressed up in a very nice business model. Visa doesn't see transaction volumes decline 10% because consumers are 'paring back' their spending — the payment simply must flow through the network. That distinction matters enormously for long-term compounding, because cyclical volume declines create earnings volatility that makes intrinsic value harder to compound and harder to assess."

Charlie Munger: "Dev, I think you're conflating cyclicality with fragility, and they're very different things. See's Candies is cyclical — chocolate sales dip in recessions and spike at Christmas. But the competitive position never wavers. What matters isn't whether volumes fluctuate quarter to quarter — of course they do, every business on earth experiences demand variation. What matters is whether the competitive position strengthens or weakens through the cycle. And every piece of evidence suggests Copart emerges from downturns stronger. They used the COVID period to accelerate land acquisitions. They used the current volume softness to win new carrier accounts with empirical returns data. The share count issue is worth discussing — it went from 933 million in fiscal 2020 to 967 million now — but that's about capital allocation decisions, not business quality."

Warren Buffett: "Let's pivot to the financial history now, because the numbers tell a remarkable story. Revenue has compounded from $872 million in 2011 to $4.65 billion in 2025 — a fourteen-year CAGR of roughly 12.7%, without a single annual decline. Net income went from $166 million to $1.55 billion over the same period — an 18% CAGR. And here's what strikes me most: operating margins expanded from 30% to 37% while the business nearly quintupled in revenue. That's operating leverage in a platform business, exactly what you'd expect if the moat is real. Each incremental vehicle processed through an existing yard and existing buyer network drops disproportionately to the bottom line."

Robert Vinall: "The ROIC trajectory is the number that demands explanation, though. It peaked at 28.9% in 2019 and has declined to 16.2% today. On the surface, that looks like a deteriorating business. But when you decompose it, the story is entirely different. Equity went from $1.45 billion in 2019 to $9.2 billion in 2025 — a sixfold increase — while net income only tripled. The denominator exploded because management retained virtually all earnings and let cash accumulate to $5.1 billion. Strip out the excess cash, and the operating business is still earning returns in the high twenties on the capital actually deployed in salvage operations. This is a capital allocation problem masquerading as a return-on-capital problem, and it's exactly the kind of thing that separates sophisticated analysis from headline-scanning."

Mohnish Pabrai: "Robert, that's the critical insight. And it connects to what I find most interesting about the current setup — management has finally started deploying that cash. They've repurchased $500 million in stock year-to-date in fiscal 2026. At roughly $33 per share, that's around 15 million shares, about 1.5% of the float. Now, is that transformative? No. But the share count trajectory tells me something important. Shares went from 933 million in 2020 to about 967 million today — net dilution of about 3.6% over five years from stock-based compensation. The $38 million in annual SBC is actually quite modest relative to a $32 billion market cap — that's about 12 basis points. The dilution came more from option exercises and issuance. If buybacks at $500 million per year can offset that and start reducing the share count, you shift from 12-13% earnings growth to 13-15% EPS growth. That's a meaningful compounding difference over a decade."

David Tepper: "Let me cut through the elegance and talk about what matters right now. The stock dropped from $60 to $33 in roughly six months — a 45% decline. The market went from pricing this as a premium growth compounder at 35x earnings to a 21x cyclical-concern business. That kind of re-rating creates opportunity for people like me who focus on what's actually happening in the business versus what the stock price implies. Revenue grew 9.7% in fiscal 2025. Net income grew 13.9%. Free cash flow hit $1.23 billion — a record. The business is performing. What changed is the insurance volume cycle, and I've seen enough cycles in my career to know that the recovery may not be predictable in timing, but it's predictable in direction. Consumers who drop collision coverage eventually get it back — either because their lender requires it, because they have a scare, or because insurance rates stabilize and affordability returns."

Pulak Prasad: "David, I want to push back on the certainty of that reversion. You're right that insurance volume cycles have historically mean-reverted. But this cycle has a feature that previous ones didn't: cumulative premium inflation of 30-50% over three years. That's not a normal insurance pricing cycle — that's a regime change in the cost of auto insurance. When premiums go from $150 per month to $225, some consumers don't just raise their deductible temporarily — they structurally change their behavior. They drive less, they keep older cars longer, they move to states with lower insurance requirements. And the longer-term existential question that nobody wants to discuss is ADAS — advanced driver assistance systems. Total-loss frequency has risen from 15.6% to 24.2% over a decade. Beautiful tailwind. But what happens when automatic emergency braking, lane-keeping assist, and eventually autonomous features start reducing accident frequency? That's not a 2027 problem, but it's absolutely a 2035 problem, and if we're claiming to think in decades, we need to grapple with it."

Dev Kantesaria: "Pulak, that's exactly my point. When I evaluate a business for inevitability over a 10-to-20-year horizon, I need the demand driver to be structurally locked in. Visa's demand driver is the secular shift from cash to digital payments — it's been going in one direction for fifty years and has decades of runway left globally. Copart's demand driver is car accidents. And there is a plausible, investable scenario where car accidents decline materially due to technology. I'm not saying it's certain — maybe a 15-20% probability over the next decade — but that's enough uncertainty to keep it outside my framework. I admire the business enormously, but I need inevitability, not probability."

Warren Buffett: "Now let's talk price. At $33.39, you're paying about 21 times trailing earnings of $1.60 for a business that has grown earnings 18% annually for fourteen years. The free cash flow yield is about 3.8% on reported FCF of $1.23 billion. The enterprise value, if you credit the net cash position, is roughly $27 billion against $1.9 billion in EBITDA — about 14x EV/EBITDA. Those are reasonable numbers for a high-quality compounder, but not obviously cheap. I'd want to own this business, but I'd want to start buying around $30, which gives me a more comfortable margin of safety as the insurance volume cycle works itself out."

David Tepper: "Warren, with all due respect, waiting for $30 when the stock is at $33 is the kind of discipline that costs you more than it saves. If insurance volumes inflect positive next quarter — and the leading indicators suggest the ex-CAT decline is narrowing — this stock goes to $38-40 before you can blink. I've seen it happen a hundred times: value investors demand one more leg down, the catalyst arrives, and they spend the next two years watching the stock compound without them. At 21x earnings with a record FCF quarter behind us and $5 billion in cash on the balance sheet, I'm buying today. The asymmetry is clear — maybe $6-7 of downside to $27 if volumes deteriorate further, versus $7-9 of upside to $40-42 on recovery. That's a trade I take every time."

Mohnish Pabrai: "I'm with David on this one, but let me frame it differently. At $33.39 with trailing EPS of $1.60, I'm paying 21x for a business I think can earn $2.00 or more within two years if volumes normalize. That gets me to 17x forward earnings on a normalized basis for a business with 36% operating margins, zero debt, and a compounding network effect. Now, do I see a scenario where this triples? At $32 billion market cap, tripling means $96 billion — aggressive for a salvage auction business, I'll grant you. But I don't need a triple. I need a reasonable probability of 50-70% upside over three years with limited downside, and I think that's exactly what I have. The land portfolio alone is probably worth $8-10 billion at replacement cost. That's my floor."

Robert Vinall: "I'm in the 'Buy Lower' camp at $30. My reasoning is specific to the reinvestment math. At $33.39, the FCF yield is 3.8%. The risk-free rate is 4.3%. I need the FCF growth rate to be meaningfully above the gap to justify holding a riskier asset than treasuries. Copart's historical FCF growth has been 15%+ annually, which more than justifies it — but right now, in the teeth of a volume decline, I can't underwrite that growth rate with confidence. At $30, the FCF yield rises to about 4.1%, and I'm buying at a price that works even if growth moderates to 10% for the next few years while the cycle resolves. That 10% discount from here is the price of my uncertainty about timing, and I think it's reasonable for a business I intend to own for a decade."

Pulak Prasad: "I require $28. That's my price for accounting for the ADAS tail risk that nobody else is willing to quantify. Think of it as an evolutionary insurance premium — I'm paying less today because the competitive environment for this species could change materially in 10-15 years. At $28, I'm buying at roughly 17.5x current earnings for a business with genuine survivorship characteristics. If ADAS doesn't materially reduce accident frequency, I've bought a wonderful compounder at a significant discount. If it does, I've protected myself with a lower cost basis. The Darwinian approach demands that I always account for environmental change, even when the current environment seems perfectly suited to the organism."

Dev Kantesaria: "I'm avoiding the stock entirely. Not because I think Copart is a bad business — it's clearly excellent. But my framework requires category alignment, and Copart doesn't pass. The insurance volume cyclicality, the long-tail ADAS risk, and the existence of a direct competitor in IAA place this outside my 'inevitable outcome' category. I need businesses where I can say with near-certainty that in ten years, the world will still be paying their toll. With Copart, I can say it's probable, but not certain, and probable isn't sufficient for how I construct portfolios. Warren and Charlie are right that this is a wonderful business. But there are wonderful businesses inside my framework and wonderful businesses outside it, and I've learned that discipline about category boundaries is more important than being right about any individual stock."

Warren Buffett: surveys the room "Let me try to pull this together, because we've had a genuinely substantive debate that revealed real fault lines. On the qualitative side, we have near-unanimous agreement that Copart's competitive position is exceptional — the combination of a buyer liquidity network effect, owned physical infrastructure that's increasingly impossible to replicate, and a consignment model that eliminates inventory risk creates something close to a permanent competitive advantage. Charlie's inversion test is compelling: there is no rational economic actor who would try to displace Copart, because every pathway to competition requires massive capital expenditure with uncertain returns against an entrenched network effect. Six of seven of us agree this is a business worth owning.

The financial evidence supports the qualitative story. Fourteen consecutive years of revenue growth, fifteen consecutive years of net income growth, operating margin expansion from 30% to 37% — this is the track record of a genuine compounder, not a cyclical business riding a favorable environment. The ROIC decline from 29% to 16% is a capital allocation story, not a business quality story, and Robert's decomposition of the cash-adjusted returns made that clear.

Where we genuinely disagree is on two questions: cyclical timing and structural risk. David and Mohnish are willing to buy today at $33.39 because they see the insurance volume headwind as temporary and the risk-reward as asymmetric — and their logic is sound. If volumes recover within 12-18 months, this stock re-rates quickly. The four of us in the 'Buy Lower' camp — myself, Charlie, Robert, and Pulak — believe the quality warrants ownership but the price doesn't yet provide adequate margin of safety for the uncertainty in volume recovery timing. We'd begin accumulating between $28 and $31. Dev stands apart, not disputing the quality but applying a framework that categorically excludes businesses with this level of volume cyclicality and the long-tail ADAS risk that Pulak rightly identified.

The majority position — five of seven in favor of ownership at some price — reflects our conviction that Copart is a rare business: a physical-world toll bridge with software-like economics, run by competent management who are finally deploying their cash pile productively. We'd rather be patient and pay $30 than eager and pay $33, but we'd rather own it at $33 than never own it at all."