I keep coming back to the same odd scene. A room full of people arguing about chips, models, and who trains the next system first, while the actual constraint sits a few miles down the road in a substation that is already booked out. If the servers cannot get power, the story is just a brochure. That is the lens I use when a fast market call lands on Quanta Services, Planet Labs, and Duke Energy in the same breath. Three tickers. Three completely different jobs. One of them builds the wires. One photographs the planet every day and just put more hardware in orbit. One sells regulated electricity and has to compete with government paper that still pays more. None of them is a slogan. All three are easier to misread if you only catch the headline.
What follows is not a transcript and not a tip sheet. It is a slower pass over the same three names, with the election noise, the yield math, and the speculative hangover left in the open. I have found that the useful part of a lightning-round comment is rarely the verb. It is the condition attached to the verb. Still a buildout. Done going down. Not quite a four percent coupon yet. Those conditions are where the real work starts.
Why The Boring Contractor Became The Data Center Conversation
Quanta Services does not design language models. It does not sell cloud credits. It builds and maintains the electric arteries that make those businesses physical. Transmission lines. Substations. Interconnections. Underground work. The unglamorous middle of the power system. For years that profile kept the stock in a contractor bucket, the kind of name income investors skimmed and growth investors ignored. The last two years broke that habit, and not because the company reinvented itself. Demand reinvented the customer list.
Hyperscale operators are pouring extraordinary sums into data center campuses. Public estimates for combined spending by the largest platforms this year have been framed around 735 billion dollars. That figure is not a Quanta revenue line. It is the size of the room. A slice of that money becomes concrete, switchgear, high-voltage interconnection, cooling plant, and site work. Quanta has spent years buying the specialist shops that do exactly those jobs, including electrical and mechanical platforms aimed at mission-critical facilities. The pitch, stripped of marketing, is simple. If the load shows up, somebody has to connect it, and self-perform craft labor is the scarce input.
The numbers behind that pitch are no longer theoretical. Second-quarter revenue was reported at about 9.56 billion dollars, up roughly 41 percent from the prior-year quarter. Trailing twelve-month revenue through June sat near 32.9 billion. Management lifted full-year 2026 revenue guidance into a range of roughly 39.3 to 39.7 billion dollars, with a record backlog around 53.4 billion. About three-fifths of that backlog was expected to convert within twelve months. Electric revenue in the quarter jumped more than 40 percent, and the segment margin widened to about 11.5 percent. That is not a story stock living on a slide. It is a contractor whose order book got ahead of the narrative.
The chip is the celebrity. The interconnection is the appointment you cannot skip.
I still think the buildout case is intact, and I say that with the election sitting right in the middle of the calendar. Political seasons have a way of freezing permitting talk, utility commission tone, and the mood around large capital projects. Headlines start to sound like nothing gets built until the votes are counted. Sometimes that is theater. Sometimes a real pause shows up in booking cadence. The distinction matters. A pause in sentiment is not the same thing as a cancelled load. Data center power requests do not dissolve because a campaign ad ran in a swing state. They wait, they get redesigned, they move a county over. The physical need remains.
What The Backlog Actually Says
Backlog is the adult version of a growth story. It is not a promise of margin, and it is not immune to cancellation, but it is harder to fake than a total addressable market slide. Quanta’s book near 53 billion dollars, up from the mid-40 billions earlier in the year, gives the next several quarters a floor that pure software names do not have. Management has also pointed to a larger transmission pipeline, with some internal research framing a jump toward 170 billion dollars of prospective transmission work. Treat that as a pipeline, not a contract. Pipelines leak. Even a leaky one of that size changes how you underwrite a contractor.
There is a second layer that gets less airtime. Acquisitions of specialist platforms, including names such as Phalcon, Enerfab, and related craft businesses, were expected to add on the order of 1.2 to 1.4 billion dollars of revenue and 120 to 140 million of adjusted EBITDA in 2026. That is bolt-on scale, not a transformation. It does widen the company’s ability to self-perform the awkward jobs around technology load centers, the ones that used to get subcontracted away. In my experience, contractors who keep the craft in-house keep the schedule. Schedule is what the hyperscale customer actually buys.
Free cash flow guidance in a band around 2 to 2.5 billion dollars for the year is the other adult metric. Growth that does not throw off cash is a different animal. Quanta is not a royalty on artificial intelligence. It is a project business with working capital, weather, and labor inflation. Cash generation at that level is what lets the backlog story survive a multiple reset.
Election Noise Versus Project Noise
Here is where I part company with the cleanest version of the bull case. Election years do slow things. Not always the steel. Often the paperwork. A utility that wants a new line through three counties will pick its moment. A data center developer that needs a certificate will not volunteer to become a campaign prop. If you only watch the stock, that hesitation looks like demand disappearing. If you watch interconnection queues and signed load agreements, it looks like a queue.
Perhaps the most interesting aspect of Quanta right now is that the company has already raised guidance in the middle of that noise. That is not proof the noise is fake. It is evidence the booked work is large enough to carry the year even if new awards wobble for a quarter or two. After the political calendar clears, the base case is more building, not less, because the load did not get smaller while everyone was arguing. I like that setup more than I like any single price target. Price targets age in a week. A 53 billion dollar book does not.
- Record revenue and a raised full-year range near 39.5 billion dollars at the midpoint.
- Backlog around 53 billion dollars, with a large share expected to bill inside a year.
- Electric margins expanding as utilization and large transmission work improve the mix.
- Craft labor and self-perform capability as the real constraint, not a lack of customers.
- Political hesitation as a timing risk, not an obvious cancellation of the load.
None of that makes the stock cheap by habit. Contractors that start trading like growth stocks eventually have to earn the multiple every quarter. A miss on utilization, a sticky labor cost, or a delayed interconnection can knock the narrative back into the industrial bin overnight. I would rather own the thesis with that risk written down than pretend the multiple is a gift.
How Quanta Sits In A Portfolio, Not A Headline
If you already own the chip designers and the cloud platforms, Quanta is a different bet on the same cycle. It fails in a different way. A model that disappoints does not cancel a substation. A permitting delay does not obsolete a training run. The correlation is real and incomplete, which is the useful kind. I have found that investors who only buy the celebrity layer of a theme end up with one risk, repeated five times. The contractor layer is lumpier, less loved, and often the piece that still has a backlog when the celebrity multiple compresses.
Position size should respect that lumpiness. This is not a bond proxy. Project timing moves quarterly prints. A sensible approach is to treat Quanta as a core infrastructure holding inside a growth sleeve, not as a trade you need to be right about by the next jobs report. Dollar-cost entry around political headlines is reasonable if you believe the load is durable. Chasing a single up day because a commentator said the word buy is how people turn a good business into a bad entry.
A plain way to underwrite Quanta: Demand: data center load plus grid replacement Proof: backlog and raised guidance Constraint: craft labor and permits Threat: multiple compression if growth normalizes Timing risk: election-year hesitation, not vanished need
That little map is not a model. It is a reminder of what you are actually buying. You are buying execution against a power shortage, with a workforce the company has spent a decade assembling. Management has said, in various forms, that self-perform work is the edge. Roughly four-fifths of the work stays inside the firm. That is a cultural fact as much as a financial one. Cultures like that do not appear in a single capex cycle, which is why late entrants struggle to copy the model when the phone starts ringing.
The Margin Question Nobody Wants To Sit With
Revenue growth is the easy applause line. Margin is the argument. Electric operating margin moving from about 10.1 percent to 11.5 percent in a year is the kind of shift that changes earnings faster than sales. The company has talked about a 10 to 12 percent band on the utility side of electric work, with the top of the band showing up when large transmission jobs and full utilization arrive together. That is a narrow window. It is also a window they have already touched.
Can it widen further? Maybe, if the mix keeps tilting toward bigger transmission and technology load centers, and if craft utilization stays tight without wage inflation eating the gain. I would not underwrite a straight line to the top of the band and then beyond. Project businesses revert. Weather, claims, and a single difficult job can hand back a year of mix improvement. The honest base case is stable-to-better margins on a much larger sales base, not a software-style operating leverage story. Earnings estimates for 2026 have been marked up toward the mid-teens per share, with another step higher penciled in for 2027. Those are estimates. They move. They are still the scoreboard the market is using.
Compare that with equipment makers who sell turbines and transformers. Different bottleneck. One company is capacity-constrained in factories. Quanta is capacity-constrained in people and permits. Both can win. They do not win for the same reason, and they should not be sized as if they were the same stock. I prefer the execution story when the factory story is already priced like a shortage that never ends. That is a preference, not a law.
What Could Make The Buy Wrong
A thesis without a failure mode is a slogan. Quanta’s failure modes are ordinary and serious. Hyperscale spending could pause if returns on the current wave of clusters disappoint. Interconnection reform could move slower than load growth, stranding projects in queues. Labor could get tight enough that margins stall even as revenue rises. A large acquisition could integrate badly. Valuation could simply be ahead of the next two years of cash, which is a failure of price rather than of business.
The election is the failure mode people reach for first, and it is the one I trust least as a standalone kill switch. Policy can change depreciation rules, tax credits, and the tone of federal agencies. It rarely erases a signed need for power in a region that is already short. If you are waiting for a perfectly quiet political backdrop before you look at a grid contractor, you may be waiting through the part of the cycle when the backlog is built. I would rather accept a messy tape than demand a clean one.
| Lens | What To Watch | Why It Matters |
| Backlog conversion | Share billed inside 12 months | Turns the book into reported sales |
| Electric margin | Hold above 11 percent | Shows mix and utilization are real |
| Guidance | Any cut to the 39 billion range | First hard sign the pause is more than mood |
| Cash | Free cash flow near the 2 billion floor | Keeps the growth from being cosmetic |
| Labor | Craft hiring and wage commentary | The actual capacity constraint |
If those five lines stay intact through the next two prints, the data center framing holds. If guidance comes down because customers delayed, the stock will rerate before the long-term need changes. That gap between the quarter and the decade is where patient capital either earns its keep or loses its nerve. I have lost nerve on names like this before, usually a quarter early. It is an expensive habit.
Planet Labs After The Bubble And The Launch
Planet Labs is the odd one in this trio, and that is why it is worth the time. The company photographs Earth, day after day, and sells the change. Agriculture. Forestry. Maritime traffic. Energy infrastructure. Civil agencies. Defense and intelligence customers. The product is not a rocket. The product is a record of what moved. Will Marshall built that company as co-founder and chief executive, and the reputation that travels with him is of an engineer who stayed with the operating problem instead of the keynote. I have a soft spot for that kind of founder. Soft spots are not a model. They do explain why some investors stay through a drawdown that would have cleared a no-name ticker.
The stock did what speculative vehicles do. It ran hard when anything orbital was treated as a call option on the future, then it gave a large share of that run back. A 52-week range that has stretched from just above 10 dollars to above 50 tells you the emotional range, not the business range. By early October the shares were trading in the mid-teens, and they had a strong session after the company confirmed a fresh launch. Twenty satellites went up on a rideshare mission. Eighteen SuperDoves to refresh the daily imaging flock. Tanager-2, a hyperspectral craft aimed at methane, water quality, and even mineral signatures. And a prototype tied to a larger partner’s in-orbit computing experiment, carrying custom artificial intelligence accelerators into space for the first time in that program.
That is a good day. It is not a new company. Annual revenue is still in the neighborhood of a few hundred million dollars, essentially all of it from imagery and the analytics wrapped around it. Gross margin above 50 percent says the data business can be attractive once the fleet is in place. It does not say the stock cannot overshoot again. The question a buyer has to answer is whether the washout already did the hard part.
A satellite stock that has already fallen out of a speculative cloud is not automatically cheap. It is simply done being a costume.
A portfolio note, not a price target
My read is closer to the optimistic side than the cynical one, with a caveat I will not sand down. The core flock is a real product with real customers, and replenishment launches are how that product stays real. Hyperspectral methane detection is a credible second act if regulators and energy operators keep paying for emissions visibility. The in-orbit computing prototype is the part that can reignite the speculative premium. If accelerators can run inference in space, draw power from the sun, and shed heat, a new revenue line appears that has nothing to do with selling another photo of a port. If the prototype is only a prototype, the stock has to live on the imaging business. Living on the imaging business is fine. It is just a smaller dream.
I think the worst of the air-pocket is likely behind it, which is a different claim from saying the shares cannot revisit the lows. Done going down is a trader’s sentence. The investor’s sentence is that the business survived the costume phase. Planet has now put hundreds of spacecraft on orbit across dozens of launches. That operating history is the asset. The costume was the multiple. Separating the two is the whole job.
Who Actually Pays For A Picture Of The Earth
It helps to be concrete. A commodities desk wants to know whether a mine is moving ore. An insurer wants a before-and-after of a floodplain. A defense customer wants persistence without owning the spacecraft. An agriculture buyer wants crop stress before the local report. None of these customers need a keynote. They need the picture to show up, the archive to be comparable, and the analytics to be boringly reliable. That is a less romantic business than the launch video, and it is the one that pays the bills.
Government work is both the stabilizer and the concentration risk. Budgets slip. Programs get recompeted. A company that looks diversified across civil, commercial, and defense can still have a quarter that hinges on one award. I do not treat that as a reason to avoid the name. I treat it as a reason to size it like a specialty data business, not like a utility. Quanta can disappoint and still have a 50 billion dollar book. Planet can disappoint and the narrative snaps back to cash burn and launch cadence. Different sport.
- Separate the daily imaging franchise from the hyperspectral option.
- Treat the in-orbit computing test as upside, not as the base case.
- Respect the drawdown from the highs without assuming the low is a floor.
- Watch launch cadence and customer mix more than the one-day pop.
- Size the position for volatility that a contractor backlog does not have.
There is a human detail worth keeping. Founders who have already lived through one full boom-and-bust in public markets tend to talk differently the second time. Less destiny, more fleet math. If the commentary stays on satellites commissioned, bands captured, and contracts renewed, the company is in operator mode. If it drifts back into total addressable universe language, the costume is being put back on. I would rather buy operator mode in the mid-teens than destiny mode at the old highs.
Duke Energy And The Paper That Pays More
Duke Energy is the income name in the set, and the argument around it is almost embarrassingly simple. The stock has been heavy. The dividend has not been cut. Intermediate and longer government yields have been high enough to give every utility a fight. When a five-year Treasury is discussed in the neighborhood of 5 percent, and the ten-year has recently traded through 5 percent as well, a regulated utility yielding under 4 percent has to justify itself with growth, safety, or both. Coupon alone will not do it. That is not a moral judgment on the company. It is arithmetic.
Around the start of October, Duke traded near 114 dollars, with a forward dividend around 4.34 dollars a share. Depending on whether you use a trailing or forward convention, the yield sits in the high 2s to the high 3s. A figure near 3.8 percent forward is a fair working number. The five-year average yield is in a similar neighborhood, a little under 4 percent. So the stock is not screamingly cheap on its own history. It is cheaper than it was when yields were lower and the shares sat well above recent levels. A 52-week low not far under 111 dollars means buyers are not being asked to catch a falling knife from the moon. They are being asked to decide whether something close to the recent floor is enough.
I want a 4 percent yield before I get excited. Right now the forward yield is a bit shy of that. The gap is not huge. It is enough to explain why the stock has struggled while risk-free paper pays more. You can bridge a half-point gap with dividend growth and a regulated earnings path. You cannot pretend the gap is not there. Anyone who buys Duke today is explicitly choosing the operating business over the Treasury. That choice can be right. It should be conscious.
Yield gap, roughly: Treasury near 5% minus Duke forward yield near 3.8% = about 1.2 points you must earn back in growth, stability, or price recovery.The bridge is not imaginary. Management has reaffirmed 2026 adjusted earnings guidance in a band of about 6.55 to 6.80 dollars a share and has talked about 5 to 7 percent annual adjusted earnings growth through 2030. The payout sits around 65 percent of trailing earnings, which is comfortable for a regulated utility and far from a strain. The quarterly dividend was lifted recently, and the company has a long streak of annual increases. Customer growth has been positive. Economic-development load under electric service agreements has been measured in multiple gigawatts, which is the quiet way a utility participates in the same data center and industrial buildout that makes Quanta interesting. Different risk. Same electrons.
Is it down enough to buy? For a long-horizon income account that wants a Carolinas-centered regulated franchise, yes, with the yield caveat attached. For someone who can roll five-year government paper and sleep, the Treasury still wins the pure income bake-off. I do not think those are contradictory sentences. They are two different jobs. Duke is a business with a growing rate base, storm risk, regulatory lag, and a dividend that has historically crept higher. The Treasury is a contract. Mixing them up is how people end up angry at a utility for doing exactly what a utility does.
When A Utility Is Allowed To Lose The Yield Contest
There is a version of this market where every dividend stock must out-yield the Treasury or be sold. That version is tidy and often wrong. A utility can lose the spot yield contest and still win a five-year total return if earnings compound, the dividend inches up, and the multiple mean-reverts when rates ease. It can also lose both contests if rates stay high and the regulator gets stingy. The path is not knowable. The setup is describable.
Duke’s beta is low. The dividend safety profile that outside screens assign it is strong, which matches a payout ratio that is not stretched and a history of payment that runs back decades. Yield attractiveness is only average, which matches the math above. Low drama, middling coupon, credible growth. That is the product. If you need drama, this is the wrong aisle. If you need a power company that is also signing large new loads, it belongs on the list, bought with a limit rather than a market order on a green day.
I would rather add on a day when the yield kisses 4 percent than argue that 3.8 is close enough and therefore identical. Close enough is how slips happen. A limit near prices that produce that 4 percent handle is a cleaner rule than a feeling that the stock has suffered enough. Suffering is not a valuation method. The coupon gap is.
Three Names, Three Jobs, One Calendar
Put them on the same page and the lightning-round logic gets clearer. Quanta is a buy if you believe the data center and grid buildout survives the election pause, and the backlog says that belief is not a leap. Planet Labs is a buy if you think the speculative premium has already been wrung out and the fleet plus a couple of options is enough, with the launch week as evidence the machine is still flying. Duke is a buy if you want the regulated compounder and you can live with a yield that still trails government paper, ideally waiting for a 4 percent entry rather than forcing it.
They do not hedge each other in a textbook way. A rates spike hurts Duke directly and can compress Quanta’s multiple. A risk-off tape hits Planet harder than either of the other two. A data center slowdown hits Quanta’s new work and, with a lag, Duke’s load growth, while leaving Planet mostly alone. The only shared theme is that all three are physical. Wires, pixels from orbit, and kilowatt-hours. After a year of arguing about software margins, physical businesses feel almost restful. Restful is not the same as safe.
| Name | Role | What Has To Be True | Entry Temperament |
| Quanta Services | Data center and grid buildout | Backlog converts and permits thaw after the vote | Own the cycle, do not chase the verb |
| Planet Labs | Daily Earth data, plus options | Imaging franchise holds and the washout sticks | Smaller size, wider swings |
| Duke Energy | Regulated income and load growth | Earnings path offsets a sub-4 percent yield | Prefer a 4 percent yield handle |
If I had to rank conviction rather than excitement, Quanta would sit first, because the evidence is in the book and the guidance, not in a launch video. Duke would sit second for accounts that actually need income, with the yield discipline attached. Planet would sit third in size and first in optionality. That ranking is personal. It will look foolish if the prototype in orbit becomes a product line and the contractor multiple compresses. Rankings are for sizing, not for pride.
A Practical Way To Hold The Set
Here is a structure I would actually use, not a fantasy allocation. Quanta as the largest of the three, funded from money that was going to chase another chip name. Duke as a staged buy, first slice now if you need the exposure, second slice only if the yield reaches 4 percent or the shares revisit the recent low with the dividend intact. Planet as a satellite sleeve position, small enough that a trip back toward the low does not change your month, large enough that a rerating is not irrelevant. Revisit after the election, not because the election decides physics, but because the excuse for delay will have expired. If awards and interconnections still stall with the excuse gone, the Quanta thesis needs a rewrite. If they resume, the pause was what it looked like.
Taxes, account type, and existing concentration matter more than any of the sentences above. A taxable account that already owns a basket of utilities does not need another regulated name at a 3.8 percent yield just because the tape looks tired. A retirement account that has no grid exposure might. I cannot see your book. I can tell you the mistake I see most often, which is treating three mentions in one segment as a basket you must buy together. They were mentioned together because the questions arrived together. The businesses did not merge.
The Mood Versus The Meter
Markets in an election autumn develop a taste for fatalism. Nothing will get permitted. Rates will never ease. Speculative names are permanently broken. Fatalism is a mood. The meter is the backlog, the launch log, and the dividend coverage. Quanta’s meter still reads expansion. Planet’s meter reads a company that can still put twenty spacecraft up and talk to them. Duke’s meter reads a covered dividend and a yield that has not yet reached the level where income buyers historically stop haggling. Moods change faster than meters. That lag is the entire opportunity, and also the entire way people get stubborn and wrong.
I do not need all three to work for the afternoon to have been useful. One durable infrastructure compounder, sized properly, earns the attention. The satellite name is a reminder that drawdowns after bubbles can be entries if the product survived. The utility is a reminder that yield is a price, not a personality. Keep those separate and the lightning round becomes a research list. Blend them into a single hot take and you will be disappointed by at least one of them for reasons that were visible on day one.
So the short version, after all the long way around. Quanta still looks like a data center and grid buildout worth owning, with the election as delay rather than cancellation, as long as the book keeps converting. Planet Labs looks like a real operator that has already taken the speculative round trip, and a fresh launch is a better reason to look than a better reason to chase. Duke is close to interesting and not quite at the coupon I want, because government paper near 5 percent is still the cleaner income instrument until the yield gap narrows or the growth shows up in the total return. That is the whole argument. The rest is patience, position size, and a refusal to let one fast afternoon do your underwriting.
None of this is a recommendation sized to your life, your taxes, or your timeline. Prices move. Guidance gets revised. A yield that looks shy of 4 percent on a Friday can be through it by the next ugly rates day. Check the filings, check the book, and decide which of the three jobs you actually need filled. The wires, the pictures, or the coupon. Pick the job first. The ticker is the easy part.
