Have you noticed how every conversation about artificial intelligence eventually slams into the same unglamorous problem? Heat. Not the metaphorical kind. Actual, expensive, infrastructure-melting heat. After a summer that cooked large parts of the United States and Europe, air conditioning stopped feeling like a lifestyle upgrade and started feeling like basic infrastructure. That is the backdrop for a stock I have watched for years: a century-old climate company that suddenly sits at the intersection of household comfort and the AI buildout.
The Quiet Company Sitting Between Homes And Hyperscalers
The business I am talking about is Carrier Global. Florida-based, more than a hundred years old, and still one of the biggest heating and cooling names on the planet. Residential systems. Heat pumps. Giant commercial plants. The unsexy kit that keeps offices, hospitals, apartments, and now server halls from cooking themselves.
We first bought the shares in early 2021 near the mid-thirties. The stock more than doubled. We trimmed when it approached the low eighties in 2024. Now it is hanging around the high fifties again, and that pullback is exactly why it is back on the desk. In my experience, the market loves a clean story and then panics the moment the story gets a little messy. This one got messy. The underlying demand did not vanish.
Here is the simple version. As the number one residential HVAC player in the Americas and roughly number three in commercial HVAC in the same region, the company is wired into two slow-moving forces that do not need a press release to keep going: people want cooler rooms, and data centers need colder rooms. One of those forces is cyclical. The other looks structural. Together they give you more than one way to be right.
Why Data Centers Changed The Cooling Conversation
Pack denser chips into a hall and you do not just raise electricity bills. You raise the thermal load until ordinary air handling starts to look inadequate. Hyperscalers know this. That is why cooling has jumped from a facilities footnote to a board-level constraint. Carrier’s data center orders jumped more than 300 percent last quarter. That is not a rounding error. That is a demand shock arriving in a company that already knows how to manufacture, install, and service large systems at scale.
When compute density rises, cooling stops being a support function and becomes part of the product.
I do not need that 300 percent print to repeat every quarter for the thesis to work. I do need the industry to keep spending on halls that cannot run without serious thermal management. On that point, the capex plans still look loud. Billions keep getting allocated. Chips keep getting hotter. Retrofit work follows new builds. None of that sounds temporary to me.
There is a catch, of course. Data center cooling is competitive, project-based, and lumpy. A single quarter can look heroic and the next one can look merely fine. Anyone buying this name as a pure AI lottery ticket is doing it wrong. The interesting part is that the same factory network and service bench that serve office towers can chase server halls without rebuilding the company from scratch.
The Part Of The Story That Does Not Need Artificial Intelligence
This is where I get more comfortable. I do not need the data center boom for the investment to pay off. Residential HVAC in the Americas grew 9 percent in the second quarter after a long inventory hangover. Industry volumes are still below recent peaks. That gap is not a guarantee. It is optionality. When distributors finally stop working off excess boxes, replacement demand has room to show up in reported growth rather than just in anecdotes from contractors.
Commercial HVAC has a different rhythm. Buildings age. Codes tighten. Owners delay replacement until the old plant becomes a liability. Then they spend. Carrier has been taking share in important markets while that replacement cycle grinds forward. It is not glamorous. It is also not a fad.
Then there is Europe. Air conditioning penetration is still lower than what Americans take for granted. One brutal summer does not rewrite culture overnight. Several brutal summers start to. I have found that consumer behavior around comfort shifts slowly, then all at once, especially when heat stops being a one-week inconvenience and becomes a public-health story. Heat pumps sit in that same conversation, tied to efficiency rules and energy-price anxiety as much as to temperature.
- Residential recovery after the inventory reset
- Commercial replacement of aging systems
- Gradual European adoption of cooling and heat pumps
- Data center thermal load from denser compute
Four engines. Different speeds. If one sputters, another can still pull. That is the feature, not a footnote.
A Company That Got Leaner After The Spin
Carrier left United Technologies in 2020 as a broader industrial package than the firm you see today. Since then management has cut costs, sold the fire-and-security business, sold commercial refrigeration, and bought a European heating platform in Viessmann. The point of that shuffle was focus: climate and energy solutions, not a museum of adjacent product lines.
Focus only matters if it shows up in cash. Last year the company generated about $21.7 billion in sales. Close to 30 percent of that came from parts and services rather than brand-new boxes. I like that mix. Equipment sales swing. Service work tends to stick around after the install crew leaves. A global footprint across roughly 150 countries and about 47,000 employees is a lot of surface area for that aftermarket engine.
Management is guiding toward roughly $2 billion of free cash flow this year. Buybacks have been aggressive. In the second quarter alone, dividends and repurchases returned about $640 million to shareholders. That is not a hobby. That is a capital-return program running while the portfolio is still being tightened.
A simpler company with rising cash conversion is easier to underwrite than a conglomerate with a good slide deck.
Is every portfolio move perfect? Of course not. Large European acquisitions bring integration risk, currency noise, and a different competitive map. Selling businesses can leave a temporary hole in reported sales even when the remaining mix is healthier. Markets punish that hole first and notice the cleaner story later. That lag is often where the price becomes interesting.
What The Valuation Is Actually Saying
The stock trades near 18.4 times forward earnings with a free cash flow yield around 5 percent. Expected earnings growth over the next three years sits roughly in an 11 percent, 15 percent, then 13 percent pattern. You can argue about the exact decimals. The shape matters more: mid-teens growth against a multiple that is no longer priced like a momentum darling.
Shares are down about 20 percent from the recent high. That is the whole setup in one sentence. The market already took something away. The question is whether it took away too much relative to cash generation and the demand stack I described above.
| Lens | Snapshot | Why It Matters |
| Forward multiple | About 18.4x earnings | Not cheap like a deep value wreck, not euphoric either |
| Cash yield | Near 5% FCF | Gives you a floor while you wait on growth |
| Growth path | Low-to-mid teens EPS | Supports the multiple if execution holds |
| Drawdown | Roughly 20% from highs | Creates a clearer risk-reward than the peak |
In a downside case I can sketch a move back toward the high thirties, call it 36 percent below recent levels. That is not a pleasant number. It is also not a mystery number. It roughly rhymes with the original purchase zone from 2021. If housing-related HVAC stays soft, data center awards slip, and multiple compression hits at the same time, you can get there. I am willing to live with that left tail because the right tail has several independent drivers, not one fragile narrative.
How Cooling Demand Actually Shows Up In Financials
People talk about cooling as if it were a single product. It is not. A residential split system, a rooftop unit on a big-box store, a chilled-water plant, and a purpose-built data hall solution do not behave the same way in an income statement. Lead times differ. Margin profiles differ. The service attach rate differs.
Data center work can juice orders before it juices earnings. That timing gap trips investors every cycle. You see the headline, you buy the stock, then you wait two or three quarters for revenue recognition and installation schedules to catch up. If you cannot tolerate that lag, this is the wrong idea.
Residential is the opposite problem. It is closer to the consumer, closer to housing turnover, closer to distributor inventory. When the channel is stuffed, great end-market weather still produces ugly shipments. When the channel is lean, even a normal summer can look like a beat. The 9 percent Americas residential print after the correction is the kind of number that tells you the pipeline is breathing again, not that the cycle is finished.
Commercial sits in the middle. Projects slip. Specs change. But once a building owner commits to replacing a plant, the aftermarket relationship can last a decade. That is why I keep coming back to the near-30 percent parts-and-services mix. It is the ballast.
Heat Pumps, Policy, And The Slow Rewiring Of Buildings
Heat pumps are easy to overhype and easy to dismiss. Both reactions miss the point. In markets where gas prices spike or building codes squeeze emissions, electrified heating stops being an ideology and starts being a procurement decision. Carrier’s European heating exposure, including the Viessmann piece, is a bet that this procurement decision keeps spreading even when subsidy headlines wobble.
Policy will not move in a straight line. Incentives get redesigned. Election cycles scramble timelines. Installer capacity remains a bottleneck in more than one country. I still think the direction of travel favors companies that can sell both conventional comfort equipment and higher-efficiency systems without needing a new brand for every regulation.
Perhaps the most interesting aspect is not the pump itself. It is the installed base. Every unit in the field is a future service call, a parts order, a controls upgrade, a replacement cycle. A company with reach across 150 countries does not need every region to boom at once. It needs enough regions to be in some phase of the cycle at any given time.
What Could Go Wrong Without A Disaster Movie Script
Let us be adults about the risks. A sharp housing slowdown can delay residential replacement even after inventories normalize. Commercial construction can stall if financing costs stay painful. Data center customers can push projects to the right if power interconnects slip, which they often do. Integration of a large European heating business can take longer and cost more than the first-year slides implied.
- Channel inventory rebuilds too slowly and residential growth fades after one good quarter.
- Data center awards stay lumpy and the multiple never gets the AI premium investors secretly want.
- Free cash flow misses the $2 billion zone because working capital or integration absorbs the difference.
- A broader industrial derating knocks 18 times earnings down toward mid-teens regardless of execution.
None of those require a recession movie. They only require ordinary disappointment. That is why the high-thirties downside sketch exists. If you cannot look at a 30-plus percent drawdown and still sleep, size the position accordingly. Position size is the real risk tool here, not a clever slogan about megatrends.
How I Think About Timing After A Twenty Percent Slide
Buying a name you already owned, trimmed, and now like again is a psychological mess. You remember the eighties. You remember selling some. You feel either brilliant or early, depending on the last tick. Ignore that noise if you can. The relevant comparison is not the high. It is the cash you get from here and the earnings power three years out if the mid-teens growth path is even roughly intact.
A 5 percent free cash flow yield is not a screaming bargain in isolation. Combined with buybacks and a business that just spent years shedding distractions, it is acceptable. Combined with a data-center order book that just went vertical, it is better than acceptable. Combined with residential volumes still below prior run-rates, it starts to look like a stacked set of options rather than a single-theme wager.
I am not arguing that every dip in an industrial name is a gift. Some dips are the market telling you the cycle peaked. This dip looks more like digestion after a rerating, plus a reminder that HVAC is still tied to housing and construction, not only to silicon.
The Aftermarket Is The Unsung Hero
Investors get hypnotized by new equipment. Contractors live on the other side of that obsession. Filters, compressors, controls, seasonal checkups, emergency calls in July. That stream does not make viral clips. It does make a less violent earnings path.
Nearly a third of sales from parts and services is the kind of mix you want when you are underwriting a company through an AI headline cycle. Headlines fade. Compressors still fail. Chillers still need techs. If the install base keeps growing because of data halls and European heating systems, the service bench grows with it. That compounding is quieter than a 300 percent order print and, frankly, more durable.
Scale helps here. Forty-seven thousand people and a presence in most of the world’s climate zones is not romantic. It is a logistics network. When a commercial customer needs a part on a Tuesday, brand reputation is only half the sale. Availability is the other half.
Reading The AI Boom Without Becoming A Tourist
Tourist capital chases the chip designer, then the utility, then whatever vendor has the most photogenic factory tour. Cooling vendors sit later in that parade, which is both a blessing and a curse. Blessing: you can still find a multiple that is not fully unhinged. Curse: you will be accused of stretching the AI label onto an old industrial.
Fine. Stretch it a little. Just do not pretend a residential furnace business is a semiconductor proxy. The honest framing is narrower. Some share of incremental cooling demand is now coming from compute density. That share is growing from a small base at a very high rate. The rest of the company still has to earn its keep the old way: sell systems, service systems, take share, squeeze costs.
I’ve found that the cleanest way to hold a name like this is to split the thesis in two columns. Column one: traditional HVAC cycle plus self-help. Column two: data center thermal demand. You only need column two to be real, not miraculous. Column one does the heavy lifting in the bear case. Column two does the heavy lifting in the bull case. That division keeps you from turning every order print into a personality test.
Cash Returns Versus Reinvestment
Aggressive buybacks can be a tell. Sometimes they mean management sees the shares as cheap. Sometimes they mean management has run out of high-return projects. With this company, I read the repurchase pace as a bit of both and not in a cynical way. The portfolio has already been pruned. Organic growth opportunities exist, especially in climate solutions and large applied systems. There is still room to send surplus cash back without starving the growth file.
Dividends plus buybacks totaling hundreds of millions in a single quarter is a statement about confidence in the cash engine. Watch working capital. Watch whether the $2 billion free cash flow target survives a messy project mix. If cash holds up while earnings grow in that 11 to 15 percent band, the 18 times multiple does not need a new religion to be justified.
Simple hold framework: Cash conversion stays near plan Residential volumes keep recovering off a low base Data center awards remain elevated even if the growth rate cools Buybacks continue without leverage drama
Miss two of those four at once and the stock can revisit the pain zone. Hit three of four and the current price starts to look like a pause rather than a peak failure.
A Note On Competition And Share
Nobody owns cooling. The Americas residential lead is meaningful and still not a monopoly. Commercial rankings shift by segment and geography. Data center thermal is crowded with specialists, controls firms, and other large industrials who smelled the same opportunity. Share gains are earned in spec meetings and service response times, not in essays.
That is why the “number one and number three” framing matters only as a starting point. Scale gets you into the bid. Execution keeps you there. If Carrier is truly taking commercial share while the replacement cycle runs, you should see it in mix and in backlog quality, not only in adjectives on an earnings call.
I would rather own a scaled incumbent learning the data center spec than a pure-play that still has to build a global service bench. That preference is personal. It is also how I have watched industrial adjacency trades play out when the end market gets fashionable. The boring giant often keeps more of the profit than the narrative suggests, provided it does not trip on integration.
Weather, Culture, And The European Angle
Americans treat central air as default. Large parts of Europe still treat it as optional, even after summers that make that stance look stubborn. Optional becomes less optional when nights stop cooling off and offices become unusable. That shift will not print as a single hockey-stick quarter. It will print as a multi-year lift in penetration, with heat pumps riding the same wave wherever policy and power prices cooperate.
Currency will slap the reported numbers around. That is the tax you pay for a real international footprint. Look through it when you can. The strategic question is whether European comfort demand is structurally higher than the last decade’s run-rate. After the recent heat, I think the answer leans yes, with plenty of room for a messy middle.
Putting A Price On Multiple Ways To Win
The phrase gets abused. Here it is specific. Win one: residential restocking and replacement. Win two: commercial share and aging-plant spend. Win three: European cooling and heating mix. Win four: data center thermal systems and the service that follows. You do not need a grand slam. Two solid singles and a walk still move the equity if cash keeps arriving and the share count keeps shrinking.
At around $58, after a double, a trim, and a give-back, the risk-reward is cleaner than it was in the low eighties. Cleaner is not the same as riskless. The left tail toward the high thirties is real. The right tail is a company that compounds mid-teens earnings, retires stock, and keeps collecting a larger rent on the world’s rising need to move heat out of rooms and racks.
The stock does not have to become an AI mascot. It has to keep turning climate equipment and service into cash while a new end market grows up beside the old ones.
That is the whole pitch, stripped of theater. Hot summers made the residential case intuitive again. Dense compute made the commercial-and-beyond case louder. A leaner portfolio made the cash case easier to believe. A 20 percent drawdown made the entry case less greedy than it was a year ago.
If you buy businesses rather than headlines, this is a climate-and-comfort compounder with a newly visible data center kicker. If you buy headlines rather than businesses, you will overpay on the next order spike and panic on the next housing print. I know which habit I prefer. The ticker will keep testing both.
None of this is a recommendation tailored to your balance sheet, your tax lot, or your sleep schedule. It is a field note on a company that sells the unfashionable machinery behind two very fashionable problems: rooms that are too hot and computers that cannot think if they overheat. Sometimes the best way to sit near a boom is to own the plumbing.
