Investment Council Debate: PAYC (Paycom Software)

PHASE 1: BUSINESS QUALITY & MOAT

Warren Buffett: adjusts his glasses "Software businesses fascinate me because the best ones print money — and Paycom certainly prints money. But here's the question I keep coming back to: is this a toll bridge that every mid-market employer must cross, or is it just another vendor that happens to be sticky? Because those are two very different businesses, and they deserve very different prices. Charlie, you've been studying this one — what do you think kills it?"

Charlie Munger: "What kills it? I can name three scenarios, and none of them are crazy. The first is Rippling. And I don't mean Rippling as a generic 'competitor' — I mean Rippling as the first company to build a unified single-database HCM platform from scratch, without the legacy of acquired modules, and extend it into IT device management, spend management, and corporate cards. That is the specific architectural advantage that has protected Paycom for twenty-five years, and Rippling is replicating it while broadening the surface area. The second scenario is more insidious: AI democratizes the automation features that differentiate Paycom — Beti, GONE, IWant — and within eighteen months, ADP and UKG have equivalent capabilities, shifting the battleground to distribution and brand trust, where ADP's forty-year reputation with HR directors dominates. The third is organizational: Richison has been CEO for twenty-eight years, holds the CEO, President, and Chairman titles simultaneously, filed six officer departure 8-Ks in five months, and tried a co-CEO experiment that failed publicly. That's the kind of leadership concentration that works brilliantly until it doesn't — and when it doesn't, there's no backup plan. Now, none of these scenarios are guaranteed. But I can't dismiss any of them either, and that's what prevents me from buying at today's price."

Dev Kantesaria: "Charlie's framing is right, but I want to challenge the room on a more fundamental question: does Paycom actually pass the toll booth test? When I own Visa, every electronic transaction on Earth must pass through their network — there is no alternative path. When I own Moody's, every bond issuer needs that rating or pays a 30-to-50 basis point penalty. Can payroll processing for a 300-employee manufacturing company in Houston occur without Paycom? Obviously yes — ADP, Paychex, UKG, Rippling, Gusto, and dozens of regional providers process payroll for the other 95% of the addressable market. Paycom has 5% market share after twenty-five years of operation. That is not a toll booth. That's a well-run competitor in a fragmented market. The 91% retention rate is good — but it's not the 97-99% that I see in my portfolio. The 9% annual churn means roughly 3,500 clients walk away every year despite genuine switching costs. That leakage rate would be unthinkable for FICO or Visa."

Robert Vinall: "Dev, I think you're applying the wrong framework and reaching the wrong conclusion from the right observation. Not every compounding business needs to be a monopoly. In my moat hierarchy, I distinguish between tier-one moats — cost savings and network effects that actively benefit the customer — and tier-two moats — switching costs that lock customers in. Paycom clearly has a tier-two moat: once a company has spent ninety days configuring tax jurisdictions across 10,000 taxing authorities, loading employee data, training every worker on Beti's self-service payroll verification, and integrating benefits elections — the cost of switching is measured in months of operational disruption, not dollars. The boomerang effect Richison described on the earnings call is the proof: clients who left for cheaper alternatives came back because they discovered the integration headaches were real. That doesn't happen with commodity products. Where I agree with you, Dev, is that this moat is defensive — it protects territory but doesn't compound it. Five percent market share after twenty-five years tells me the moat prevents client losses but doesn't accelerate client gains. That's a different compounding profile than Visa's."

Pulak Prasad: "I want to bring the evolutionary lens to this. Paycom has survived twenty-eight years of technology cycles — the on-premise era, the first wave of cloud migration, the smartphone era, and now the AI transformation. Revenue grew 14% during the 2020 pandemic, which was the worst GDP contraction in a generation, because employers cannot stop paying their workers regardless of economic conditions. That crisis resilience is not accidental — it reflects a non-discretionary demand floor that has no equivalent in consumer technology or cyclical industries. But survival and thriving are different concepts. The growth deceleration from 30% to 9% to a guided 6-7% is consistent with a species that has filled its ecological niche — adapted perfectly to the U.S. mid-market, 50-to-5,000 employee segment, but unable to expand beyond it. The 9% churn is the metabolic cost of maintaining the colony. If new logo acquisition falters — and the three-month sales retraining introduces real execution risk — the population could begin shrinking."

Mohnish Pabrai: "Pulak, let me push back on the ecological niche framing. A species that occupies 5% of a habitat hasn't filled its niche — it's barely scratched the surface. The more relevant question is whether there's a structural limitation in how fast Paycom can expand within that habitat. And here the transcript gives us a clue that nobody in this room has mentioned: Richison said that 'consumers and clients oftentimes have a more difficult time of digesting full solution automation.' Think about what that means. The product's most important advantage — the automation that eliminates 90% of payroll labor — is difficult for prospects to understand in a ninety-day sales cycle. That is a genuine go-to-market ceiling, and it explains why a twenty-five-year-old company with a clearly superior product has only 5% penetration. The three-month sales retraining is Richison's attempt to break through that ceiling, but it's a risky bet: you're pausing the sales engine at exactly the moment growth is decelerating."

David Tepper: "Mohnish, I hear what everyone's saying about moat quality and go-to-market ceilings, but you're all debating whether Paycom is a great business, and I think you're missing the more important question: is it a mispriced business? I don't need Paycom to be Visa. I need the stock to be cheaper than the business warrants. And right now, this thing has dropped 43% from its highs because growth-mandate funds mechanically sold when revenue growth dropped below their 15% minimum threshold. That's not an analytical process — it's portfolio construction rules creating forced liquidation. The demand side is fine: Richison said there's no change in prospect willingness to buy, bookings were up in 2025, retention improved to 91%, and a record number of boomerang clients returned. The reflexivity here is virtuous, not destructive — unlike a bank run where lower stock price actually impairs the business, Paycom's lower price just makes buybacks more accretive. But I'll save the numbers for phase two."

PHASE 2: FINANCIAL EVIDENCE

Warren Buffett: "Let's turn to the numbers. The ten-year financial record here tells an interesting story — revenue from $329 million to $2.05 billion, gross margins steady at 83% the entire time, and ROIC settling into a 24-32% range for the past five years after peaking above 50% during the capital-light scaling phase. That ROIC normalization isn't moat erosion — it's the natural denominator expansion as capitalized software development and client fund balances grow. But what catches my eye is the cash flow picture, because there's a genuine discrepancy in the data that demands honesty. The fiscal.ai reported free cash flow for FY2025 is $67.7 million — which on a $6.85 billion market cap implies barely a 1% FCF yield. ROIC.ai's OCF-minus-capex methodology shows approximately $394 million in TTM FCF, or $7.04 per share, which tracks cleanly with the decade-long trajectory from $0.47 to $5.99. The difference appears to be short-term investment purchases flowing through investing activities, but until we see several more quarters confirming the higher figure, I want to anchor my valuation on GAAP earnings rather than FCF alone."

Charlie Munger: "Warren's right to flag that discrepancy, and it connects to a broader point about earnings quality. The FY2024 stock-based compensation figure was negative $23 million — a $153 million swing from the prior year's $130 million. That anomaly inflated FY2024 GAAP EPS to $8.77, which the market then treated as 'peak earnings.' When FY2025 came in at $8.35, the narrative became 'declining earnings.' But strip out the SBC anomaly and normalize FY2024 to something like $6.20-$6.50 in EPS, and underlying profitability actually improved substantially in FY2025. The market is anchoring on the wrong baseline. However — and this is important — I cannot tell you with precision what normalized SBC looks like. It was $95 million in 2022, $130 million in 2023, negative $23 million in 2024. That volatility makes clean earnings normalization genuinely difficult, and I want to be compensated for that uncertainty."

Dev Kantesaria: "Let me add the owner-earnings lens, because this is where the bull case gets more complicated than it appears. If you take ROIC.ai's TTM FCF of $394 million and subtract normalized stock-based compensation of, say, $110 million — using the 2019-2023 range while excluding the anomalous 2024 — you get owner earnings of approximately $284 million. Divide by the $6.85 billion market cap, and the owner-earnings yield is roughly 4.1%. The ten-year Treasury is at 4.3%. So an investor buying Paycom today is earning barely the risk-free rate on true economic earnings. Now, the business is growing, and buybacks at current prices add accretion — but that 4.1% starting yield is thin for a business with a narrow moat and genuine competitive uncertainty from Rippling."

David Tepper: "Dev, you're penalizing the company twice. SBC is already expensed in the GAAP income statement — that's how you get to the $8.35 EPS. Then subtracting it again from FCF is double-counting the dilution. The honest measure is GAAP earnings, which already reflect SBC as a cost, or FCF without subtracting SBC again. At $8.35 GAAP EPS and $124.82 stock price, you're paying 14.9 times — the cheapest this stock has ever traded. And the adjusted EBITDA story is even more striking: management guided $950-$970 million for FY2026, which at 44% margins represents genuine operational improvement. The real question isn't whether the business is cheap on today's economics — it obviously is for a 25% ROIC zero-debt software company. The question is whether growth stabilizes at 7-9% or continues toward 4-5%, and I think the weight of evidence favors stabilization."

Robert Vinall: "David, the question of whether GAAP EPS or owner earnings is the right metric matters enormously for the compounding math, and I don't think it's double-counting. GAAP expenses SBC as a non-cash charge that depresses net income, yes — but it doesn't capture the actual share dilution that reduces your ownership claim on future cash flows. The correct way to think about it is: FCF represents cash available to shareholders, but SBC creates new shareholders who claim a piece of that cash. Owner earnings — FCF after SBC — captures both effects. That said, your point about the 14.9x GAAP P/E stands. For my 15% return hurdle, I need to see 7.7% earnings yield plus 2.5-3.0% net buyback accretion plus 1.2% dividend plus 3-4% EPS growth from revenue and margins. That math works at $108. At $125, I get 12-13% expected return — competitive, but below my hurdle."

Mohnish Pabrai: "The capital allocation history is the elephant in the room that nobody's wrestled with honestly enough. The zero-acquisition discipline over twenty-eight years is genuinely rare and admirable — it preserves the single-database architecture that is the entire moat's foundation. But management spent $1.125 billion on cumulative share buybacks at an estimated average price of around $375 — three times today's stock price. That's roughly $750 million in destroyed shareholder value from poor timing, while management described themselves as 'opportunistic buyers.' If you're going to rely on the buyback machine for compounding math — and everyone in this room is — you have to acknowledge that the historical track record of buyback execution is objectively terrible. Now, at $125, the math changes dramatically: each $370 million deployment retires 3 million shares instead of 1 million. But I need the price lower to be confident the compounding actually materializes."

PHASE 3: VALUATION & VERDICTS

David Tepper: "I'm buying today at $124.82. The setup is classic: forced selling by growth funds that hit their portfolio construction limits created a 43% drawdown in a zero-debt, $2 billion recurring-revenue business with 25% ROIC. The asymmetry runs about 2-to-1 — maybe 15-20% downside to $100-$106 in a bad scenario, 35-45% upside to $170 if FY2026 EPS normalizes toward $9.50 at 18x. Management has beaten initial guidance by 1-2 points in consecutive years, retention is improving, and Q4 recurring revenue accelerated to 11.3% against a tough comp. I'd add aggressively below $100."

Warren Buffett: "At $124.82, I think the stock is approximately fairly valued on today's economics — 15x times $8.35 gets you right there. There's no margin of safety at this price. I'd want $110 or below, which gives me 13x normalized EPS and a 6.4% FCF yield before buyback accretion. At that price, I'm comfortable even if growth disappoints and Paycom becomes a 5-6% grower, because the passive return from FCF yield plus buybacks plus dividends still delivers mid-teens returns."

Charlie Munger: "I'd wait for $105 — 12.5x normalized earnings. I need a price that works even if all three of my death scenarios partially materialize simultaneously. At $105, survival alone produces satisfactory returns. I don't need growth; I need the paychecks to keep going out."

Dev Kantesaria: "I want $115 — approximately a 5% FCF yield entry using the ROIC.ai measure. The architectural moat is real, and if IWant's adoption trajectory sustains and pushes retention toward 92%, this could evolve from a well-run competitor toward something with mild toll booth characteristics. But today, 5% market share after 25 years and 9% churn disqualify it from the toll booth category. I'm intrigued enough to start small."

Robert Vinall: "My 15% hurdle rate is non-negotiable, and $108 is where the math clears. At that price, 7.7% earnings yield plus 2.5% buyback accretion plus 1.2% dividend plus 3-4% growth gets me to 15.5%. The IWant catalyst could shift the moat from tier two to tier one — switching costs to genuine cost savings — and if retention reaches 92%, I'd revise fair value to $150 and increase conviction significantly."

Mohnish Pabrai: "P/E of 15.5x clears my gates, market cap of $6.8 billion is in range. But the asymmetry at $125 is only about 1.5-to-1 — I need 2-to-1 minimum, which means $100. At $100, downside to $80-$85 on 10x trough earnings is 15-20%, upside to $133 on 16x normalized is 33%. The boomerang clients tell me product superiority is real. The buyback timing tells me management's capital allocation instincts are not."

Pulak Prasad: "I need $105 — the widest margin of safety in the room, because my evolutionary framework assigns 30% probability to the structural stagnation scenario where EPS compresses to $7 and the multiple falls to 12x, producing an $84 stock. At $105, that downside is only 20% versus 24% upside in the base case. The non-discretionary demand floor makes permanent capital loss from $105 extremely unlikely, and that's what matters most to me."

Warren Buffett: "Let me ask the room one final question that cuts to the heart of this. Would you be comfortable owning Paycom for twelve months without checking the price, without reading an earnings call, without seeing a single headline?"

David Tepper: "Yes — the payroll checks go out regardless. But I wouldn't do that with most of my positions."

Charlie Munger: "No. The sales execution pivot, the C-suite turnover, and the Rippling competitive dynamic all require monitoring. This is not a set-it-and-forget-it business at this stage of its evolution. It might become one again in two years, but not today."

PHASE 4: SYNTHESIS

Warren Buffett: surveys the room "Let me try to pull together where this discussion has landed, because I think we've identified something genuinely interesting even as we disagree on price.

On the qualitative side, we broadly agree on three things. First, payroll processing is about as non-discretionary as any business I've ever studied — 160 million American workers need a paycheck every two weeks, calculated correctly across 10,000 tax jurisdictions, and that demand doesn't disappear in recessions. The 2020 stress test proved this when revenue grew 14% during the worst economic contraction in a generation. Second, the single-database architecture is a real structural advantage — not a toll booth, as Dev rightly points out, but genuine switching costs that manifest in 91% retention, improving, with clients who leave for cheaper alternatives actually coming back. That boomerang effect is rare and meaningful. Third, Richison's twenty-eight-year organic-only strategy — zero acquisitions, zero debt — reflects the kind of long-term discipline we prize, even as the C-suite turnover and failed co-CEO experiment introduce legitimate succession risk.

Where we disagree is on two critical questions. David believes the 43% drawdown was mechanical — forced selling by growth funds hitting portfolio construction limits — and that the reflexivity is virtuous because lower prices make buybacks more accretive. He's buying today at $125. The rest of us acknowledge that thesis but want more margin of safety, because we're less certain about whether growth stabilizes at 7-9% or continues decelerating toward 4-5%. Charlie's inversion reveals four death scenarios — Rippling competition, AI commoditization, founder departure, and sales execution failure — none dismissible, none certain. The 5% market share after twenty-five years and the 9% churn rate both suggest that product superiority hasn't translated to market share dominance, and that's genuinely troubling for a long-duration compounder thesis.

On price, six of us would buy between $100 and $115 — a range that represents 13x to 15x EPS and provides a 10-18% margin of safety from our fair value estimates of $128-$140. David alone would buy today. The consensus is clear: this is a good business at approximately fair value, and it becomes a compelling investment with another 10-12% decline. The leading indicator is Q2 2026 client count growth — above 4% confirms the growth stabilization thesis; below 3% confirms the terminal deceleration narrative. Until that data arrives in November, patience is the correct posture for most of us. But David's point about the reflexive buyback math at depressed prices is not wrong — and if you have a different risk tolerance, buying here with a 3% position is defensible."