Kyle Grieve and Shawn O'Malley reopen a stock pitch Mr. O'Malley made about a year ago and passed on, Comfort Systems USA, and grade the decision now that the shares have gone up roughly fivefold. They work through the backlog, the margin expansion, the acquisition math, the balance sheet, the incentive plan and a rebuilt intrinsic value model, and argue about whether missing a five-bagger was an analytical error or a working investment process doing its job.
Hosts: Kyle Grieve and Shawn O'Malley, co-hosts of The Investor's Podcast, who run the show's intrinsic value portfolio and pitch businesses to each other on air
Published: 29 August 2026 on The Investor's Podcast feed
Apple Podcasts | 1 hr 19 min
✅ Time saved: 44 min
Key Takeaways
A $320 fair value became an $1,800 stock in about a year
Mr. O'Malley had talked about buying below $300 a share
"So I put it in the too hard pile, but I did end up missing out on a five bagger"
The multiple barely moved — the earnings did all the work
P/E in the mid-30s at the time of the original pitch, about 43 times today
Net income nearly tripled in the last 12 months against 2024, and is up almost 10 times from 2020
The backlog was the visible part of the story and was still underestimated
Up more than eight times since 2020, and more than doubled between 2024 and 2025
Mr. O'Malley: "I was pretty much wrong by an order of magnitude about just how much Comfort Systems could benefit from AI capex"
Operating margins went from a tight band near 5% to about 16.5%
The original model had them declining to around 8.5%
Contracting is labor-intensive, so the expansion came without the operating leverage a software business would have
Electrical contracting went from 0.4% of revenue in 2017 to nearly 29%
The segment grew 81% in the past year; mechanical, the larger one, grew about 40%
Technology is now 56% of year-to-date revenue, up from 30% in 2024
The moat is the balance sheet, not the trade
About $24 billion in aggregate contract value across nearly 8,500 projects at the end of 2025, one of them worth about $470 million
Surety bonding capacity keeps smaller contractors out of the bidding
Six deals since the start of 2025, for a combined price of over $540 million
Mr. Grieve backed out an average multiple of around nine times operating earnings
Feyen Zylstra went for about 10 times operating profits, or 0.7 times revenue
The balance sheet carries $1.8 billion in cash against $53 million of long-term debt
A credit facility amended last summer to about $1.1 billion is undrawn
Management is paid on EPS and free cash flow, with the targets published
The EPS target went from about $5.50 in 2023 to $15.75 in 2025
They have taken the full 200% of target, the maximum allowed, for each of the past three years
The rebuilt model gets to about $1,600, below the current price, and neither host is buying
Mr. O'Malley: "buying at these prices feels like really a momentum bet on AI spending"
A Stock Pitch Revisited at Five Times the Price
Mr. Grieve opened episode 842 by saying the show was doing something different: going back to a stock Mr. O'Malley pitched a while ago and passed on. The model at the time assigned a fair value of around $320 a share, and Mr. O'Malley waited for a better entry point that never came.
"I remembered vividly, and yeah, this is going to be a painful episode for me because we really, or I really, missed out on quite the opportunity. My valuation could not have been more off." — Shawn O'Malley
"And the stock today is now sitting at five times its price when, unfortunately, Shawn, you first looked at it, just miles past the top of even the most bullish view that you possibly could have modeled for." — Kyle Grieve
Mr. O'Malley framed the question he wanted the audience to settle: "is our investment process flawed, or is this just part of the reality of having an investment process where you end up excluding things that other people will make a lot of money on, and that's okay"
Mr. Grieve said the show goes back over old pitches to grade the decision rather than the outcome, covering both the winners and the mistakes of omission
He said he speaks for himself, Mr. O'Malley and their co-host Daniel in weighing the process very heavily
Comfort Systems USA is a mechanical, electrical and plumbing installation and service provider based in Houston. Mr. Grieve said management dislikes the serial acquirer label but that it is the most useful one available. Mr. O'Malley said part of what he liked was the established playbook behind over $10 billion in revenue and over 23,000 employees.
The compounding record is what makes the size unusual. With a market cap over $60 billion, the company has compounded revenue at 13% a year and earnings per share at 22% a year for the last two decades
Over the past 10 years those rates rose rather than decayed — revenue growth of 21% and EPS growth of 37%
In the year since the original pitch, revenue is up 33% and EPS up over 50%
Mr. Grieve said businesses at a similar market cap usually see growth plateau or decline, and Comfort's numbers say the opposite: the business is getting better
Why Cyclical Businesses Are the Hardest Ones to Value
Mr. Grieve said he agrees with the instinct to avoid cyclicals, because he cannot get conviction on when a cycle turns or what normalized earnings across the cycle look like. His illustration was blunt: a cyclical business could make $2 billion one year, lose $1 billion the next, and make $10 billion the year after.
Mr. O'Malley agreed cyclicals are harder to value, then made the counterpoint: that unpredictability means the market is more likely to misprice them periodically, which creates opportunity for anyone who genuinely knows the industry
The rule of thumb he gave listeners inverts the usual reading of a P/E. At the top of a cycle, earnings rise explosively and Wall Street refuses to give full credit for a temporary boost, so the multiple looks cheap while the stock is expensive
At the bottom, the earnings denominator is depressed and the multiple looks expensive at the point of best value
"when cyclicals are cheap, they're actually expensive. And when they're expensive, they're probably actually cheap" — Shawn O'Malley
The Original Decision, and the Difference Between a Mistake and a Miss
Mr. O'Malley walked back through the call. He estimated fair value at about $320 a share, talked about buying below $300, had low conviction because the earnings are cyclical, and shelved it.
"So I put it in the too hard pile, but I did end up missing out on a five bagger." — Shawn O'Malley
At the time of recording the share price is almost $1,800
He allowed for the possibility that this is simply what a disciplined process costs: "every now and then you're going to get these things where the casino falls in love with a stock and it flies to the moon"
He said that kind of move is not necessarily the sort of fundamentals Graham, Munger and Buffett would find attractive
Mr. Grieve pointed out that the haggling over entry price was irrelevant in hindsight, since Mr. O'Malley was debating 290 against 320: "Instead of maybe a 4 to 5 bagger, you would have been looking at a 5 to 6 bagger."
Mr. Grieve noted he never bought the business either
Mr. O'Malley drew the distinction the whole episode rests on: "I think there's a major difference, though, in making an analytical mistake versus making what is maybe a correct and pragmatic call as an investor by honoring your own margin of safety" and then having it overrun by an unknowable swing in demand
He said resisting the urge for self-flagellation was the more productive use of their time, and added that this framing is what lets him sleep at night
Mr. Grieve said he has made enough investing mistakes to fill an entire show
The Original Thesis: GDP-Rate Organic Growth Stacked on a Serial Acquirer
Mr. O'Malley laid out the thesis as he built it. Assuming no further acquisitions, he thought organic growth might run 2% to 4% a year, roughly the level of GDP growth, because demand for mechanical and electrical engineering services should be steady and might beat GDP slightly.
The rest was reinvestment. Since 2015, Comfort Systems has earned a return on incremental invested capital well north of 20%
If a company could reinvest 100% of earnings at those rates and stack organic growth on top, earnings should compound at well more than 20% a year
The catch he built into the model was diminishing returns. Companies rarely find enough opportunities to absorb all their earnings at 20%-plus, so only so much capital goes to work each year at a satisfactory rate
Shareholder returns in a serial acquirer come from three parts: the organic growth of the underlying companies, the share of earnings reinvested into acquisitions multiplied by the return on those investments, and buybacks or dividends
Comfort has two reinvestment lanes, electrical contracting and mechanical contracting, with mechanical over 70% of revenue
The question he got wrong was maturity. He asked whether the company was running out of attractive businesses to buy and whether the law of large numbers meant it needed bigger or more numerous deals to move the needle
"And so that is where I ended up being quite wrong. I thought that the business was much more mature than at least it's proven to be in the last year."
The Backlog Grew Eightfold, and That Was the Visible Part
Mr. Grieve said he is a major fan of the serial acquirer model, which he described as a compounding machine when run well, and put Comfort Systems in the category of boring, high-quality businesses investors like to talk about and rarely own.
He said the boringness is the reason most investors skip them, and offered his own case: "I had the same problem with owning Micron, which unfortunately 20x from where I first bought it"
The AI angle turned a mundane business of two or three years ago into one of the hottest names in the market
The show owns hyperscalers including Alphabet and Amazon, so the hosts see the hundreds of billions being spent on AI each quarter
The backlog is the clearest measure of what that spending did. It has grown by over eight times since 2020 and more than doubled between 2024 and 2025
Mr. Grieve put the counter-case to Mr. O'Malley: looking at it in 2025, a fair-priced boring business had already received a massive tailwind and might already be swept up in hype and overvalued on temporary earnings
Mr. O'Malley said that was exactly his thinking, and that he had seen the backlog double over a single year for a business that had been slow-growing and plain
The AI Capex Cycle Was Underestimated by an Order of Magnitude
Mr. O'Malley said he recognized AI would be a net positive and could lift earnings power for years, because once a data center is built it needs routine HVAC maintenance to keep functioning, and new buildings become customers for many years.
HVAC service revenue on data centers is higher margin than Comfort's other contracting work, because cooling very expensive and advanced computer chips is specialized
He said everyone has been shocked by the scale of big tech's data center capex, naming Meta, Alphabet, Tesla, Amazon and Microsoft, with Apple to a lesser extent
He looked at the business not long after the so-called DeepSeek moment, when it seemed large language models could be trained at a fraction of the cost — and spending kept growing regardless
"So I was pretty much wrong by an order of magnitude about just how much Comfort Systems could benefit from AI capex."
His explanation is the shape of the curve: exponential growth is hard to model confidently, because a slight error makes the margin of error in an intrinsic value enormous, unlike underwriting a more linear industry
He said it remains an open question what share of these inflated earnings is sustainable, and allowed that it might be a genuine paradigm shift with stickier earnings
Operating Margins Broke Out of a Two-Decade Band, and Nobody Knows the New Normal
Mr. Grieve said the original model projected operating margins declining steadily over five years to around 8.5%, from around the 10% range at the time. Instead they have expanded to 16 and a half percent.
From 2005 until 2022 the margins sat in a tight range around 5%; since 2020 they have skyrocketed to nearly 17% today
That leaves the question the whole valuation turns on: what are normalized operating margins going to be?
Mr. O'Malley said the expansion is hard to explain and harder to believe will last. This is not software or manufacturing, where economies of scale translate into operating leverage
The criticism of the regional-contractor roll-up model is that growth requires buying more contracting businesses, and they are very labor-intensive
"it's very one-to-one of, if you want another dollar of earnings, you need to spend a dollar to acquire that" — with variable costs and more hiring rather than fixed costs spread across a wider revenue base
Two unknowns therefore sit side by side: what share of earnings is artificially inflated by data center spending that may not last, and what margins look like on the other side of it
Where the 5x Came From: The Multiple Barely Moved
Mr. Grieve said margin improvement is part of the story but not the whole of it, and walked through the arithmetic of a business that went up about five times in a single year.
The P/E was somewhere in the mid-30s when the original episode came out and sits at about 43 times today
He said that multiple expansion is nice but nowhere near enough to explain a 5x without large earnings growth
Revenue: $7 billion in 2024, $9.1 billion in 2025, and $11.2 billion in the last 12 months
Since fiscal 2024, revenue has compounded at 34%
Net income has nearly tripled in the last 12 months against 2024 earnings, and is up by almost 10 times from the 2020 number
Mr. O'Malley said these are not results you often see in a public company: "It seems like more of what you would expect from a startup where they're doubling revenue every year in their infancy. And yet we're talking about what should be a pretty boring contracting business."
The growth was partly telegraphed through the reported customer backlog, which makes the trade look obvious in hindsight
"order backlogs do tell you where the business is going, if everything goes as hoped, but it's also not necessarily contractually guaranteed income either" — Shawn O'Malley
He said he underestimated both how much the backlog would grow and what share of it would convert into real sales dollars, and that AI spending has proven far more durable than he or most people would have anticipated 12 months ago
The growth is not evenly spread. Mechanical, the larger and more traditional HVAC segment, accelerated to about 40% over the last year; the electrical engineering services segment grew 81%
The Electrical Segment Was a Rounding Error in 2017 and Is Now the Growth Engine
Mr. Grieve went back through the revenue mix. In 2017 the electrical segment accounted for just 0.4% of revenue, and it has risen very sharply since to become nearly 29%.
No single acquisition did it. He contrasted Comfort with QXO, a business the show looked at recently that scaled up on only three acquisitions
Comfort has steadily made more acquisitions since 2022, deploying about $319 million over the last year across a handful of smaller deals
The deal pace stepped up. Mr. O'Malley had noted an average of about one and a half to two deals a year at the time of the pitch; in 2025 it went to around four
Two of them came after the pitch: Feyen Zylstra and Meisner Electric
Mr. O'Malley said the deals are made from a position of strength, priced with discipline and, as in most of their deals, funded almost entirely from cash with little need for debt.
The Feyen Zylstra purchase price was about 10 times operating profits, or 0.7 times revenue, which he called reasonable
FZ adds electrical contractors supporting data center buildouts and brings more exposure to healthcare
"Hospitals have huge HVAC needs and they don't really go out of business or have downturns because people are always sick." — Shawn O'Malley
He said he likes the deals because Comfort is sticking to what works and not venturing beyond its circle of competency or making unprecedented deals
Mr. Grieve agreed, commending them for not reinventing the wheel or moving into an adjacent vertical where they have no experience
Same-Store Sales Grew 26%, and That Is the Number That Regresses
Mr. Grieve said most serial acquirers run organic growth in the low single digits. For fiscal 2025, Comfort had about 3.4% from new acquisitions and 26% growth in same-store sales, driven by demand in technology and data centers.
That is the double-edged sword. With historical same-store growth in the 3% to 4% range, he said the current rate looked highly likely to regress toward that mean
If it becomes clear the market has been in an AI spending bubble, he said, the outcome could be more than reversion — a revenue decline for several years
He added that it does not look imminent: this quarter the backlog increased by $1.6 billion, and $1.4 billion of that was same-store growth
Mr. O'Malley explained the term for listeners: same-store sales growth refers to the businesses Comfort already owns, as against anything from recent acquisitions
The bull and bear cases sit on the same fact. If the AI capex comes to fruition, the stock's run is not entirely unjustified and Comfort could be in a new growth paradigm nobody saw at this scale five years ago
The bear case is a stock market crash, AI startups going under, lenders bailing on data center financing, or in his words "maybe Anthropic's IPO doesn't go very well this fall"
In that case much of the backlog evaporates and the jump in revenue and earnings looks like a flash in the pan
He said a lot of the spending only became visible in the last 12 to 24 months, though ChatGPT's popularity gave an early hint
Technology Now Dominates the Revenue Mix, and Texas Is the Next Virginia
Across Comfort's customer categories, Mr. Grieve said one stands out: the industrial segment, historically the largest, whose growth over the last year has been near parabolic at 74%. It accounts for around three quarters of revenue in the first half of 2026.
Technology sits inside that segment, which is why it matters. 56% of Comfort Systems' year-to-date revenue came from technology, up from just 30% in 2024
The company also added another electrical business, Hunt Electric, which he expects to add industrial and technology work
Mr. O'Malley said management's latest presentation flags large project bookings in Texas, North Carolina and Indiana
He has seen the boom firsthand living in Virginia, which he called the data center capital of the U.S.
Texas is the hub of the future, forecast to overtake Virginia by 2030, with "Texas reportedly using about a fifth of its energy capacity to support data centers today"
Texas already leads the U.S. in chip exports, with billions of dollars of further government funding and private equity going into the industry there
The risk he flagged for anyone buying today is local politics. Data center construction and its effect on the energy grid is being protested in many places, and he suspects it becomes an increasingly controversial issue in Texas
He said the search for more favorable regulatory stances is what moved construction out of Virginia in the first place, and Texas will not necessarily stay friendly indefinitely
Copper Was First-Order Thinking; HVAC Technicians Were Second-Order
Mr. O'Malley used the electrification trade to explain what he missed. The grid does need to expand for electric cars and battery-powered goods, and many people have bet on copper, which he called a crowded trade because it is obvious first-order thinking: more electricity demand means more copper to conduct it.
Energy consumption in the U.S. was flat for decades, and per capita consumption is finally growing again thanks to AI and data centers
Comfort Systems is the second-order bet, an idea he credited to Howard Marks: "increased electricity demand leads to more copper demand, but less obviously, it also leads to greater demand for HVAC technicians"
He gave himself no excuse. He knew data centers and EVs were driving electricity demand and therefore demand for electrical engineering services — "And that's because my wife works in renewable energy. So I saw through her work how energy demand had just been exploding."
He still did not connect the dots enough to avoid underestimating how much further the segment could run after its initial growth
Mr. Grieve said second-order thinking needs depth in the area to be possible at all: the need for more HVAC technicians is written on the wall, but nobody who never thinks about HVAC technicians will get there. He called it a circle of competence problem
Management That Sandbags Is Worth More Than Management That Sells
Mr. Grieve took a detour into how he reads management, prompted by the optimism baked into Comfort's original pitch.
A CEO saying they are optimistic about the future means very little, because no CEO would tell shareholders they are pessimistic — "maybe Buffett would do that"
Most CEOs have to watch their words so as not to upset the market, which dulls listeners to bullish management commentary
He was explicit that he does not think Comfort's management is untrustworthy; this is his general view
The edge comes from time. Owning a business for years teaches you how its management speaks, whether they sandbag or run optimistic, and that feel is unavailable to an investor seeing the company for the first time
Mr. O'Malley called Comfort's a high-class management team and said their reserve about the opportunities ahead of them probably limited his own optimism
"businesses that tend to conservatively underestimate their prospects are actually probably in a stronger position than the companies that have to sell investors on just how great their business model is" — Shawn O'Malley
If you know you are in a good position, you let the numbers do the talking, which he said is Comfort's style
He said in hindsight he should have spent more time on the data center and chip manufacturing part of the business
What the Disclosures Actually Tell You About the Next Year
Mr. Grieve said the correlation between data center or chip manufacturing activity and Comfort's mechanical and electrical offerings is obviously there, but he thinks it is nearly impossible to pin down precisely enough to use in an investment context.
The disclosure looks out about a year. The backlog represents revenue expected over the next six to 12 months, and probably understates the real number, because shorter-term projects inside that window are not included in it
A swelling backlog is therefore a good signal that things are getting better
Mr. O'Malley added that management is not aggressive with guidance but recently noted things are likely to keep going well for the remainder of 2026, which he believes
It does not tell them where the backlog ends up, but he reads it as a signal it will at worst be sustained and at best increase significantly
Alphabet's $200 Billion, and an SEC Rule Change That Reads Like a Bubble
Because the intrinsic value portfolio owns hyperscalers, Mr. Grieve used one of them as a proxy for the spending. Alphabet has guided 2026 capex to a range of about $200 billion, and management has said that number is likely to increase in 2027.
He was careful about the limit of the inference: there is no way of knowing whether Alphabet is even a Comfort Systems customer, but as a proxy for data center and chip spending it points the same way
Mr. O'Malley raised something he had read that day: the SEC has made it easier for data centers to sell asset-backed securities
A major subset of data center securitizations no longer needs the disclosures and investor protections similar deals typically require, including risk retention, the rule that an issuer holds some of the debt so its interests align with investors
"And to me, that feels like a sign of a bubble." — Shawn O'Malley
He also pointed to private market giants — Blackstone, Apollo, Brookfield and BlackRock — promising to invest up to $500 billion into data center infrastructure alongside Nvidia
"How does one even fathom $500 billion in spending?"
He said his skeptical impulse on AI spending could not have proven more wrong, so his views come with a grain of salt
Mr. Grieve said easier access to capital is the common feature of the bubbles he has studied, whether the asset is a data center or a tulip bulb
"where people can find cheap money, money is going to go into those sectors"
He assumes the SEC and the government know this and want the sector to grow, and said the problem is that nobody knows whether the sector will earn an attractive return
Rising Margins Say Demand Is Outrunning the Competition
Mr. O'Malley said a tailwind this large should attract competition, and inevitably will. The evidence that it has not yet bitten is in the margins, which are trending up and to the right — meaning demand is growing fast enough to more than offset new competitive pressure.
Expanding margins also rule out the failure mode of a badly run contractor: bidding on any project available, even at a loss, to inflate revenue
He said poorly run real estate and contracting companies do sign a pile of projects that lift revenue, please some investors and unlock executive bonuses, while making the company less profitable over time
Electrical contracting carries longer-term contracts than some other contracting work, so each incremental dollar earned there is arguably worth more because the cash flows are contractually more likely to persist
Mr. Grieve offered a counter-example from a business he owns, unnamed on air, in housing construction
It has had to take on lower-margin projects to keep revenue from falling sharply, and margins came down as a result
He said it is not close to Comfort's quality, but the point generalizes: when bidding dynamics tighten and suppliers become hard to find, contractors can raise project prices and margins rise with them
Low demand is worse for these businesses; the high demand Comfort is exposed to right now is a very good thing
The Moat Is a Balance Sheet: $24 Billion of Contracts Across 8,500 Projects
Mr. Grieve said he assesses a competitive advantage by whether it is widening or shrinking. He agreed with Mr. O'Malley's original view that Comfort's moat is not especially strong, and still does, with one addition.
The advantage Mr. O'Malley had identified was specialization plus a geographic footprint that keeps the existing customer base sticky
The new one is the ability to take on ever-larger projects. At the end of 2025 the company had about $24 billion in aggregate contract value spread across nearly 8,500 projects, and one project was reportedly worth about $470 million
A local or regional competitor without Comfort's balance sheet cannot bid anywhere near that size
Using Hamilton Helmer's framework, Mr. Grieve classified this as a corner resource, and said it is becoming a more powerful advantage as project scale grows
Surety Bonding Is the Barrier to Entry Nobody Talks About
Mr. O'Malley said he found the mechanism confusing on first reading, and walked listeners through it. Comfort has surety bonding capacity, which lets it win larger projects.
A customer needs a guarantee the project will actually be finished. If the contractor walks off the job or goes bankrupt mid-build, the project is delayed or canceled
Surety bonding is a form of insurance against that risk: the contractor promises to finish, and if it does not, it has to pay for the job to be completed
Comfort Systems has no losses to date and is not expecting any
It is both a financial advantage and an intangible one, resting on a trusted reputation. He compared it to Berkshire Hathaway, whose balance sheet and standing let it capture deals others could not
Berkshire "made a killing, helping bail out Goldman Sachs during the great financial crisis", partly on cash and partly because Buffett was seen as a trusted partner
The bonds come from surety companies, not from Comfort itself. Those firms look at how much equity a company has, how much cash is in the bank, how long it has been around, and whether it has a record of completing large projects
Mr. Grieve's version: a small two-truck HVAC company in a tiny town will never get bonded for a $100 million project, and a contractor builds legitimacy as it scales
"no surety company wants to take a risk betting on a smaller operation if they don't have to" — Shawn O'Malley, who said that creates a barrier to entry and lets Comfort get bonded for increasingly large jobs easily
He added the honest caveat that most contractors are not hugely profitable businesses
Process Power, a Company Founded in 1917, and a Decentralized Model
Mr. Grieve said other investors argue that successful serial acquirers have process power, defined as an organization and activity set that enables lower costs or superior products which can only be matched by an extended commitment. In plain terms, it has to be built over time.
Comfort Systems has been publicly traded for over 30 years, and the original business was founded back in 1917
The structure is decentralized, which Mr. O'Malley had originally compared to Berkshire. Headquarters does not call the shots for every subsidiary; responsibility is delegated to lower-level managers closer to the customer
The process improvement Mr. Grieve highlighted is the move toward prefabrication and modular construction, building parts of a project off-site, which helps them finish increasingly complex jobs
Incremental Returns of Nearly 60%, and Six Deals Since the Start of 2025
Mr. O'Malley said he had looked at returns on incrementally invested capital between 2015 and 2024 and found over 25% returns, which he called an exceptional job of allocating capital, and guessed the recent numbers are higher still.
He compared the resistance to the law of large numbers with Alphabet and Amazon using core-business profits to drive growth into cloud computing, while saying he would not put Comfort in the same category
Mr. Grieve updated the calculation. He credited John Huber with popularizing the metric, then extended Mr. O'Malley's analysis: trailing 12 months net income ending in the second quarter of 2026, plus any additional incremental invested capital
The incremental return on capital more than doubled, to nearly 60%
The compounding rate of the business rose to about 39%
Mr. O'Malley kept his caveat attached: much of the organic growth is a tailwind he does not think lasts forever
The deal pace has changed materially. From one and a half to two deals a year at the time of the pitch, Comfort has completed six deals across the mechanical and electrical segments since the start of 2025
"those six deals had a combined purchase price of over $540 million"
The open question is whether that is a new normal or whether the opportunities deplete quickly
Mr. Grieve said he gets giddy when a serial acquirer raises its acquisition pace, but that pace depends on the size of the M&A team and the pipeline
Reverse-Engineering What Comfort Pays: About Nine Times Operating Earnings
Mr. O'Malley raised the risk of a bad deal — wrong synergies or the wrong price — and cited a portfolio holding as a case of restraint.
Netflix passed on buying Paramount Skydance. Mr. O'Malley said he was excited about Netflix acquiring IP like Superman, then added: "I'm glad though that they had the willpower to restrain themselves from making a really, splashy and sexy deal", because it did not fit their financial discipline
Comfort will never generate deals that make headlines, but the point is having disciplined criteria
Mr. Grieve went to the filings for Comfort's acquisition criteria and was disappointed. He said Perimeter Solutions and Lifco spell out what they look for; Comfort does not
He read it as a decision not to hand copycats the blueprint rather than a red flag
So he reverse-engineered it from the last six deals. The company discloses purchase price but not always the revenue of what it buys
He noted approvingly that Comfort uses notes to sellers as part of the payment, so it does not have to front the entire purchase price
For five of the deals the backlog is disclosed, and Comfort states it expects to recognize about 75% of backlog as revenue over the next 12 months
He built a table projecting future revenue from backlog, applied Comfort's own mechanical and electrical segment margins to estimate operating earnings, and got a directionally correct view
The average multiple came out around nine times operating earnings. He called that slightly higher than he usually sees on serial acquirers, and acceptable given the organic growth — a buyer getting zero organic growth should pay a cheaper multiple
He is torn nonetheless: "26% organic growth from an explosion in data center spending is completely unsustainable over the long term"
One hiccup came in the latest acquisition, Summit. Contingent earnout obligations for fiscal 2025 decreased by 62%, driven by a lower earnout expense for Summit after larger changes in the prior year's forecast and results, and as a result of reaching the maximum cumulative earnout target
Mr. Grieve said Summit may have been a little too bullish on the following year's numbers when purchased, and that he is nitpicking given how well everything else has gone
Mr. O'Malley explained earnouts for listeners: bonuses the acquirer pays the acquired company for hitting performance targets, so a decrease looks like falling costs but "It's actually more of a bad thing because it means the new business they acquired is underperforming expectations."
Negative Net Debt, an Undrawn Facility, and Notes Owed to Former Owners
Mr. Grieve revisited the debt profile Mr. O'Malley had flagged as a strength. At the time of the original research the company had about $131 million in debt against $860 million in cash, and had generated over $700 million in operating cash flow over the previous nine months.
The position has since improved. As of the second quarter of 2026 there is $1.8 billion in cash and only $53 million in long-term debt
A credit facility amended last summer raised borrowing capacity to about $1.1 billion, and appears completely undrawn
All the debt currently carried comes from the notes to former owners issued in acquisitions, at interest rates of about 4% to 5.5%
Mr. O'Malley said negative net debt was a real positive for him, invoking Buffett's first rule of investing, not to lose money
His serious point: an equity investor in a bankrupt business loses everything, so good investing is as much about minimizing catastrophic mistakes as finding big winners
"bankers are probably falling over themselves to lend to a company like Comfort Systems with such a high quality balance sheet"
He said the only reason he can see for Comfort to draw on the facility would be a monster acquisition target with a multi-billion dollar price tag, which does not fit a strategy of small regional deals
Mr. Grieve said the business is well below its debt covenants and well protected by the equity and cash on the balance sheet, and Mr. O'Malley closed the topic by quoting Munger: "I had nothing to add"
A Succession Test, and Insider Ownership Drifting Down
Mr. O'Malley said Brian Lane was CEO and president when he first looked at the company, and had clearly set the business up for success. In December 2025 Mr. Lane gave up the title of president to Trent McKenna, previously chief operating officer, who joined the company in 2004.
Mr. Grieve read it as Mr. Lane finding his successor. Mr. Lane is now about 69 years old, and the president title carries enough responsibility to serve as a test of Mr. McKenna
He called culture a heavily underrated part of a competitive advantage and said he is always happy to see internal promotion
He said it is too early to give Mr. McKenna an honest review, and that he would attribute the strong operating metrics since the change more to Mr. Lane
Mr. O'Malley said the stability of the leadership team is something he highly respects and probably explains much of the success
Mr. Grieve's twist on that: the market knows stable management of a good company is valuable, which is part of why the premium is there
The stock currently trades on about 31 times operating earnings, which he called very high but unsurprising given recent performance
Insider ownership does not scream alignment, though he called it adequate: about 1.8% in 2024, down to about 1.2% currently
With founder William George still on the team as CFO, he would have liked a higher figure, but it is not a game breaker
His yellow flag is that the decline suggests management converts options to cash on vesting rather than holding
He would prefer insider ownership stay above 1%, and there is a risk it drops below if selling continues
The charitable read he offered: management may believe now is when they get the most cash from vested options, and may be more inclined to hold once the data center and chip manufacturing cycle ends and shares are more reasonably priced
An Incentive Plan Paid at the 200% Maximum, Three Years Running
Mr. O'Malley said his initial analysis outlined how management compensation at all levels is tied to earnings per share growth and free cash flow, which he called one of the better incentive plans the show has covered.
Mr. Grieve agreed, saying it is probably one of the best he has analyzed, and that base salaries are reasonable for a company with a $60 billion market cap — all executives make between $300,000 and a million dollars
What he respects most is that the targets are published. Most companies he sees keep them internal so management knows what to hit without telling shareholders
"EPS target in 2023 was about $5.50 and in 2025, it's up to $15.75."
The free cash flow target over the same period went from about $204 million to $302 million
He noted the actual numbers are quite a bit higher, so these are targets management holds high conviction in hitting
Management has received the full 200% of target, the maximum allowed, for each of the past three years
Mr. O'Malley said the bonuses were well deserved and the proof is in the numbers, "with shares compounding at over 100% annually since 2023"
The Risks: 40% Recurring Work, and Grieve's Own Losses on Backlogs
Mr. O'Malley said their co-host Daniel had asked good questions about risk when they first looked at the company. The core one was the correlation with the U.S. construction cycle, historically a volatile industry.
The hedge he identified is the installed base. Comfort is not only reliant on new builds — about 40% of the business comes from existing buildings, in renovations, expansions, maintenance and repairs
In a cash crunch some of that can be deferred, but only for so long, which makes it a natural hedge
Mr. Grieve said further diversification into technology helps, but with a caveat: technology has its own cycles, and the company is arguably trading traditional construction cyclicality for data center and chip manufacturing cyclicality
"Once that cycle turns, I can't see how this business maintains its current momentum."
His second risk is personal, and cost him money. He has held two very small businesses with backlogs he thought were large relative to their market caps
Both worked for a short period and then ran into trouble: projects took longer than expected, so revenue booked for one year arrived in 18 months or two years
Or deadlines fell immediately after an earnings report, pushing the revenue into the following quarter
"So my experience with businesses with growing backlogs, to be honest, hasn't been great."
He said Comfort's own revenue, EPS and free cash flow growth have been outstandingly steady, with free cash flow the one lumpy line quarter to quarter
Mr. O'Malley said quarterly swings in working capital are what a construction business produces, and it does not bother him: "since free cash flow has compounded at 22% a year for the last two decades, I have a lot of faith that they'll manage these working capital needs pretty well"
The Rebuilt Model Comes In Below the Share Price, and They Still Aren't Buying
Mr. Grieve rebuilt the valuation. Because the original model was built on 2024 numbers and the share price has appreciated five times, he said it needed more than a light update. Mr. O'Malley noted the model is linked in the episode's show notes and is free.
The rebuild uses trailing 12 months figures from the first half of 2026, with the biggest changes in the revenue base and the margins
The original model had operating margins gently declining from about 10.5%; they are now running at about 16.5%
He was candid that the exercise is harder now than it was then, because he thinks the business is running somewhere around the top of the cycle, and said he would be lying if he claimed any insight into how much longer the AI cycle lasts
The base case assumes revenue compounds at about 13% a year on average, reflecting a slowdown but also a new reality with more growth opportunity from AI than before
He accounts for the multiple coming down once the top of the AI spending cycle is clear by using an exit multiple about two-thirds of the current valuation
He described the inputs as very subjective, aiming to be directionally correct even while being precisely wrong
"So with all these assumptions, I get a fair value of about $1,600 for Comfort Systems, which is a small discount to the current stock price." — and that is before applying any margin of safety to set an entry price
Mr. O'Malley's conclusion is that the intrinsic value genuinely did increase, through external factors and the company's savviness in capitalizing on them, and he would be more excited to own it after this episode than when he first looked
But the range of outcomes is very wide, nobody knows where the AI super cycle leads, and he could not get comfortable at a price that did not hedge downside risks he does not think are priced in
"buying at these prices feels like really a momentum bet on AI spending, being able to continue as it has"
He was direct about the psychology: "I have a lot of FOMO from this one. And I think that clouds my thinking."
Rather than doubling down and chasing at five times the price, he would rather stay on the sidelines and watch for a moment when sentiment turns sharply against the business and the market discounts the quality of the maintenance revenue those data centers will keep generating
"it's okay that this one got away from us. But we don't have to chase it and there are always going to be more opportunities"
Mr. Grieve, who set out to grade the decision, ended up defending it. He said he had not wanted to bash Mr. O'Malley for skipping it, and then found the opposite: "the more and more I understood the business, the more I actually think he made completely the right decision and not buying it, because simply put, there's just so many unknowns in this business"
If either of them were building or funding AI data centers and knew the capex cycle had another decade in it, he said, Comfort Systems would probably look really cheap right now
Neither has any conviction on that call, so skipping the business is the right one
Mr. Grieve closed with a quote from William Thorndike — "The heads of many companies are not skilled in capital allocation. Their inadequacy is not surprising. Most bosses rise to the top because they have excelled in areas such as marketing, production, engineering, administration, or sometimes institutional politics" — and said the rare company gets a CEO who can both run the business day to day and allocate capital well, which is why Comfort Systems under Brian Lane has been such a big winner.
Comfort Systems turned out to be a much better business than either host modeled, and both still think the right move is to leave the five-bagger alone rather than pay a price that only works if AI capital spending keeps compounding.
Products, Companies & Tools Mentioned
Comfort Systems USA (The subject of the episode: a Houston mechanical, electrical and plumbing installation and service provider, over $10 billion in revenue and over 23,000 employees, whose shares have gone up about fivefold since the show passed on it)
Feyen Zylstra and Meisner Electric (The two acquisitions made after the original pitch; FZ went for about 10 times operating profits or 0.7 times revenue and brings electrical contractors serving data centers plus healthcare exposure)
Hunt Electric (A further electrical acquisition Mr. Grieve expects to add industrial and technology work)
Summit (Comfort's latest acquisition, and the one blemish — fiscal 2025 contingent earnout obligations fell 62% on a lower Summit earnout, which Mr. O'Malley said signals underperformance against expectations)
Alphabet and Amazon (Hyperscalers held in the show's intrinsic value portfolio; Alphabet has guided 2026 capex to about $200 billion with more expected in 2027, used here as a proxy for data center spending)
Meta, Tesla, Microsoft and Apple (Named alongside Alphabet and Amazon as the big tech spenders breaking records on data center capex, with Apple to a lesser extent)
Nvidia, Blackstone, Apollo, Brookfield and BlackRock (Named as promising to invest up to $500 billion into data center infrastructure — "How does one even fathom $500 billion in spending?")
QXO (A business the show looked at recently that scaled on only three acquisitions, contrasted with Comfort's many smaller deals)
Perimeter Solutions and Lifco (Serial acquirers Mr. Grieve said disclose their acquisition criteria clearly, unlike Comfort Systems)
Netflix (A portfolio holding that passed on buying Paramount Skydance; Mr. O'Malley was glad they restrained themselves despite the appeal of IP like Superman)
Micron (Mr. Grieve's example of a boring business he owned that went up 20 times from where he first bought it)
Berkshire Hathaway (Cited twice: as the model for a decentralized structure, and as proof that a balance sheet plus trust wins deals, as in the Goldman Sachs bailout during the financial crisis)
The Securities and Exchange Commission (Has made it easier for data centers to sell asset-backed securities, dropping requirements including risk retention — which Mr. O'Malley reads as a sign of a bubble)
Books & Resources Mentioned
Hamilton Helmer's competitive advantage framework (The source of "corner resource," Mr. Grieve's classification for Comfort's balance-sheet-driven ability to bid on projects competitors cannot)
Howard Marks on second-order thinking (The frame Mr. O'Malley used to explain why copper was the obvious trade and HVAC technicians were not)
John Huber (Credited with popularizing return on incremental invested capital, the metric both hosts used to assess capital allocation)
William Thorndike on capital allocation (The closing quote, on how few CEOs rise to the top through capital allocation skill)
The show's Comfort Systems valuation model (Linked in the episode's show notes and free to access, per Mr. O'Malley)
Get the latest market chatter as it happens: