The one-paragraph version

Imperial Brands is one of the simplest ideas in the market right now. It is a mature tobacco business that throws off a lot of cash, returns nearly all of it, and keeps buying back stock at a rate that is hard to ignore. At the current run-rate, the company is set to return roughly £2.7bn to shareholders: about £1.27bn of dividends plus a £1.45bn buyback, against a market value of about £18.6bn at 2,443p. That is roughly a 14.6% shareholder yield. The shares are down hard in 2026, yet on 9 September CEO Lukas Paravicini said Imperial remains fully in line with FY26 guidance and made clear that the lower share price makes the buyback more effective. That is enough for me to stop waiting and publish the idea now.

Why this one has my attention

I am not buying Imperial because I think cigarettes are suddenly exciting. I am buying it because the capital allocation is. The company does not need to become a growth story for this to work. It only needs to keep turning a slowly declining profit pool into a large amount of cash, and then use that cash intelligently.

Most investors look at Imperial and stop at the first line: declining volumes, regulation, tobacco, not very fashionable. Fair enough. But if you stop there, you miss the part that matters. At today’s valuation, this company is shrinking itself fast enough that the share count can do a lot of the heavy lifting.

This is not really an earnings story. It is a denominator story.

Henry Singleton understood the basic idea better than almost anyone. At Teledyne he bought back enormous amounts of stock when he thought it was cheap. Buffett later reduced the lesson to one sentence: what is smart at one price is dumb at another. Munger put it even more memorably: “Pay close attention to the cannibals.” Buybacks are only as good as the price paid. At today’s valuation, Imperial looks much closer to the intelligent end of the spectrum.

The flywheel

This is the heart of the whole thing, and I think it is more powerful than many shareholders realise.

First, the buyback reduces the share count. That part is obvious.

Second, a smaller share count makes the dividend cheaper to fund in total, even if the dividend per share goes up. Imperial has already been showing this. The payout per share has risen, but the total dividend bill has barely moved because there are fewer shares left to pay.

FY22 → FY25: dividend per share 141.17p → 160.32p (+13.6%); total dividend cash £1.325bn → £1.314bn, essentially flat.

Third, the cash that no longer has to be spent paying dividends on retired shares stays inside the machine. That cash can help fund the next buyback. If the share price is still cheap, the next buyback retires an even bigger slice of the company. Then the dividend bill gets lighter again. And around we go.

Fewer shares → lower aggregate dividend cost → more buyback capacity → fewer shares again.

What the last few years already show

The nice thing is that you do not need heroic assumptions to see the mechanism. It is already there in the historical numbers.

From FY22 to FY25, dividend per share rose from 141.17p to 160.32p. Free cash flow moved from £2.562bn to £2.7bn, with a wobble in between but no real impairment to the cash engine. The share count has been shrinking by roughly 4.5% to 5% a year. Meanwhile the announced buyback stepped up from £1.0bn to £1.1bn, then £1.25bn and now £1.45bn.

That combination matters. Shareholders receive more per share. The company does not have to increase the total dividend bill much to do it. And the buyback itself has been getting larger. This is exactly the kind of setup I like: not flashy, just quietly compounding in the background.

The buyback has increased every year since the programme began. Source: Imperial Brands company results.

How long until I own the whole thing?

This is the calculation that made me want to write the piece in the first place.

Take 762.8 million shares outstanding, a share price of £24.43, and free cash flow of £2.7bn. Now make the assumptions deliberately boring. Assume free cash flow never grows. Assume the dividend per share rises by just 2% a year. Assume everything left after the dividend goes into buybacks. And assume the share price never rerates.

There is no multiple expansion in that model. No rescue from a heroic NGP success story. No assumption that management suddenly discovers a second growth engine. It is a very plain setup.

Even then, the arithmetic gets silly quite quickly. The falling share count makes the dividend bill lighter. The lighter dividend bill leaves more room for the buyback. The buyback retires a bigger percentage of the remaining company. At a constant 2,443p, the toy model runs out of shares during FY2035.

Except mine.

I am obviously not claiming that I will literally become the last shareholder of Imperial Brands. The point is to make the mechanism visceral. At today’s price, this setup cannot just continue indefinitely without forcing some other variable to move. Either the shares rerate, the buyback changes, the cash flow deteriorates, or management allocates capital elsewhere. But the current combination of cash, valuation and buybacks has a very short fuse.

Flat £2.7bn FCF, 2% annual dividend growth, all residual cash used for buybacks, and a constant 2,443p share price. Starting share count: 762.8m, the latest disclosed figure excluding treasury after the 8 September 2026 repurchase. The model runs out of shares during FY2035. Illustration, not forecast. Source: Author calculation.

Why I do not mind a weak share price

This is the part of the thesis that still feels counterintuitive, even after you understand it. I own the shares, so instinctively I want the price to go up. But Imperial is also buying every day on my behalf.

At around 3,300p, a £1.45bn buyback retires roughly 44 million shares. At 2,443p, the same £1.45bn retires roughly 59 million. Nothing has to improve in the operations for the buyback to become meaningfully more effective. The market is simply offering the company more of itself for the same money.

That is why I am not praying for a quick rerating. A cheap price is not the enemy of this thesis. For a while, it is the fuel.

More importantly, I no longer have to infer whether management thinks about the share price the same way. At the Barclays Global Consumer Staples Conference on 9 September, Paravicini said the lower price makes the buyback more impactful because Imperial gets more shares back for the same money. He noted that roughly 20% of the share count had already been retired since 2021 and expects this year’s programme to remove roughly another 6%. He also reiterated the evergreen buyback and strongly signalled another significant programme next year, while making clear that the Board still sets the amount annually.

Is Imperial actually leaning into the sell-off?

One thing I wanted to check was whether the buyback was behaving the way a rational cannibal should. It is easy to say that a lower share price makes a repurchase more valuable. It is another thing to keep buying when your own stock is falling.

So far, the answer is encouraging. From April through July, Barclays spent about £5.5m per trading day on Imperial’s behalf, at a weighted average purchase price of roughly 2,780p, and bought about 196,000 shares per trading day. In August, as the shares fell into the 2,600s and then the 2,500s, average spend rose to £7.5m per trading day and shares bought rose to about 287,000 a day. The average purchase price was about 6% lower, while cash deployed per day was roughly 38% higher and shares bought per day about 46% higher.

I would not overstate this. Barclays is executing the second tranche under an irrevocable, non-discretionary mandate, and daily purchases also depend on market volume and preset limits. There is another reason not to overread the next few weeks: only about £115m remains to be spent before the end of October, so the daily run-rate can fall materially from here without telling us anything about management’s view of the shares. Still, August is exactly what I want to see from a buyback programme when the stock gets cheaper.

There is an even cleaner way to see the benefit. The first £725m tranche bought 23.1m shares at an average 3,138p. The second tranche had already bought 22.4m shares by 8 September with roughly £115m still to spend. If that remaining cash were deployed around 2,443p, the same £725m would retire about 27.1m shares, roughly 17% more stock for exactly the same money.

Source: Imperial Brands RNS announcements; author’s calculations. April-July daily averages use total spend and total shares bought divided by all trading days in the period. Second-tranche figures are updated through the 8 September purchase. Remaining cash is assumed spent at 2,443p for illustration.

This is why I care much more about the price at which Imperial buys its shares than whether the share price makes me feel good today.

Is the business good enough to feed the machine?

This is the only question that really matters. Buybacks do not save a bad business. If the underlying cash engine is collapsing, all you are doing is concentrating yourself into a worse asset.

So far, Imperial still looks good enough. In FY25, Tobacco and NGP net revenue grew 4.1% at constant currency. Tobacco price/mix was up 5.4%, more than offsetting the volume decline. More importantly, the volume decline has actually been improving for three straight years: -7.1% in FY23, -4.0% in FY24 and -1.7% in FY25. Free cash flow was £2.7bn and cash conversion was about 97%.

That is all I need. I do not need a miracle. I need a business that remains cash-generative and reasonably stable while management keeps shrinking the share count at sensible prices.

Why the market is sulking in 2026

The first half of FY26 was weak, and I do not want to hand-wave that away. Tobacco and NGP net revenue grew only 1.8% at constant currency and adjusted operating profit only 0.6%. Australia has been messy because illicit tobacco is damaging the legal market. The US has been tougher at the value end. NGP is moving in the right direction, but not in a clean straight line. There is also the Delaware litigation bill and a less friendly currency backdrop.

All of that explains why the market is unimpressed. What it does not prove is that the capital-return story is broken. In the same September discussion, with the financial year almost finished, Paravicini said Imperial remains fully in line with FY26 guidance. He pointed to the Australian step-down beginning to lap, a more stable tariff backdrop and the usual second-half flow-through of pricing as reasons for confidence in the H2 acceleration. The standing guidance remains 3% to 5% adjusted operating profit growth at constant currency, at least high-single-digit adjusted EPS growth and at least £2.2bn of free cash flow. If that holds, the machine still has plenty to eat.

I had originally planned to wait for an October pre-close before publishing this. I no longer think that makes sense. Management has just addressed the central questions in public and, as I write this, Imperial’s financial calendar does not yet list an October trading statement; the next scheduled results announcement is 17 November. Imperial has historically issued an early-October pre-close, so one may still be added. But I am not going to sit on the thesis waiting for a date that is not set and may only be a short update anyway.

One of the scarier summer risks has also eased. The original German excise proposal contemplated a roughly €2 increase and an accelerated September implementation. Paravicini said the September increase is now off the table and the discussion for January is around €0.50 to €1.00. A €1 increase could still hurt volumes, but this is a much more manageable setup than the one investors were worrying about a few months ago.

The self-help story is getting more tangible too. Management now expects the Langenhagen exit and Taiwan disposal to deliver around £100m of annualised savings once completed in July 2027, with manufacturing excellence adding another roughly £25m on an annual basis by the end of this year. I am not putting those savings into the toy model. I like that they make the flat-FCF assumption harder, not easier, to disappoint.

NGP is helpful, but it is not why I own the stock

I do not need Imperial to win the global race in next-generation nicotine. In my head, that business is still optionality, not the thesis. Later in the conversation, Paravicini described a deliberately disciplined approach: wait until a category is established, enter where Imperial has a route to market, and do not waste shareholder money trying to create a market from scratch. European vape is already profitable. The recent Helwit acquisition cost an initial £39.8m and more than doubles Imperial’s existing modern-oral share in Sweden, while Black Buffalo gives it another targeted route into US oral nicotine. Great if those assets compound. I do not need them to rescue the valuation.

There is also a US duty-drawback opportunity that should begin to contribute meaningfully in the second half of FY27 and have a full-year effect in FY28. Management did not quantify it, so I assign it no value here. That is the kind of upside I prefer: visible, potentially useful, and absent from my base case.

What could go wrong

The bear case is not mysterious. The flywheel stops working if free cash flow starts shrinking faster than the share count. That could happen if pricing power weakens, if excise or regulation bites much harder than expected, if the Australian problem spreads, or if Imperial has to give up too much margin to defend its US position. It could also happen if management keeps buying back stock but the stock turns out not to be cheap.

Those are the risks I care about. Not the share price falling. The cash flow falling.

What I still want to see

The thesis does not need another trading statement to exist, but three things still matter.

• The FY27 buyback amount. Management has now strongly signalled another significant programme. It has not promised £1.45bn again, let alone a progressive increase. At today’s valuation, that number still matters enormously.

• The full-year cash numbers. I want FY26 free cash flow, leverage and cash conversion to confirm that the operating business is still feeding the machine.

• The operating detail. I want the US and Germany share dynamics to look controlled, Australia to stop getting worse, and NGP to keep moving toward profitability without swallowing capital.

The verdict

This is why I like the story. I do not need Imperial to become a different company for this to work. I do not need it to out-innovate Philip Morris, and I do not need some heroic terminal value. I need a business that keeps generating decent cash, a management team that keeps returning it intelligently, and a share price low enough that those repurchases create real value for the people who stay.

At 2,443p, Imperial is not far from that sweet spot. On the forward annualised dividend run-rate, the yield is about 6.8%. The buyback is roughly 7.8% of the current market value on its own. The dividend pays me while I wait. The buyback increases my ownership while I wait. And, slightly perversely, the market’s boredom helps rather than hurts me.

The September conference gave me enough to stop waiting. Management says the cash guidance still holds and the buyback remains an important lever. What I want now is the FY27 number.

I own the shares. My share count is not going down.

Disclosure

I own shares in Imperial Brands. This article reflects my personal research and opinions and is not investment advice. The projection above is a thought experiment designed to illustrate the mechanics of a shrinking share count; it is not a forecast.

Source note: Operating figures and capital-return figures are drawn from Imperial Brands annual reports, FY25 results and HY26 results. September management comments are paraphrased from CEO Lukas Paravicini’s 9 September 2026 Barclays Global Consumer Staples Conference webcast on Imperial Brands’ investor-relations site. Daily buyback figures, current share count and programme progress are drawn from Imperial Brands RNS announcements; buyback charts are the author’s calculations. Market price used: 2,443p, the 9 September 2026 close. Shares in issue excluding treasury shares: 762.8m after the 8 September purchase. Second-tranche spend through that purchase: approximately £609.9m, leaving approximately £115.1m.

Disclosure: I initiated a position in Copa Holdings (NYSE: CPA) shortly before publishing this article. Forward estimates in this article are my own.

In the second quarter, Copa Holdings (NYSE: CPA) paid 85% more for jet fuel than a year earlier. It does not hedge. Operating margin fell from 21.7% to 8.7%.

Then it raised fares and kept growing.

Yield rose 8.7%. Capacity rose 16.5%. In July, load factor reached 89.7%. The shares, meanwhile, fell more than 20% from their June high to $126.45 at the end of August.

If you think you already know how this airline story ends, look at July.

I bought Copa for one reason: the current fuel shock is testing the Panama network in real time, and so far the network is repricing faster than the market seems to believe. The risk is equally simple: 46 incoming MAX deliveries could turn a good airline into an ordinary one.

THE INVESTMENT IN ONE SCREEN

• Quality: 22.6% operating margin in 2025 - more than three times the global airline industry - and a three-year ROIC averaging 16.7%.

• Fuel test: Roughly 64% of the higher guided fuel cost is already reflected in guided unit revenue.

• Valuation and risk: The middle of my 2027 range produces roughly $17-$21 of EPS, or about 6-7x earnings; 46 incoming MAX deliveries could push capacity beyond what the network can absorb.

I listened to eight consecutive earnings calls, from Q3 2024 through Q2 2026, and read the filings, traffic releases and Investor Day material alongside them. The calls matter because last year’s ROIC tells you what the old fleet earned. They tell you how management behaves when Boeing is late, Argentina gets over-served, fuel doubles and forty-odd aircraft are still coming.

A seller can make a coherent case. Copa’s recent margins benefited from a supply-constrained post-Covid region. Fuel has now delivered the first serious stress test in years and quarterly margins collapsed into single digits. Management responded by raising 2026 capacity guidance from 11-13% to 14-15%. That is exactly how good airline economics usually end: strong returns invite capacity, capacity weakens pricing, and the industry gives the economics back.

I think that reading misses what is happening underneath the headline margin.

The quality is real. The cash cost of growth is too.

Copa earned $818.9 million of operating profit in 2025 on $3.62 billion of revenue, a 22.6% operating margin. Applying the company’s 13.4% effective tax rate gives $709 million of NOPAT. I define invested capital as equity plus borrowings plus aircraft lease liabilities and, deliberately, I do not subtract the cash pile. On average 2024-25 invested capital of $4.73 billion, I get a 2025 ROIC of about 15%.

Even allowing for those basis differences, the gap is hard to miss. Copa finished 2025 above LATAM and IAG, and its 22.6% operating margin was a little more than three times the IATA estimate for the global airline industry.

Margins are only the first test. The harder question is whether those profits survive the capital required to produce them.

Copa does not rank first, and that is useful context. LATAM screens higher, while IAG and Ryanair are also excellent operators. What stands out is Copa’s floor: on this deliberately conservative measure, ROIC never fell below 15% in any of the last three years.

Keeping cash in invested capital matters. Netting out liquidity pushes ROIC above 20%, but an airline cannot sensibly run without a large buffer. Fuel shocks, recessions, airport closures and aircraft groundings are not theoretical risks. Some of that cash is operating capital in disguise.

ROE was stronger again: $671.6 million of net income against average equity of $2.57 billion, or about 26%. This is not leverage pretending to be quality. Copa finished 2025 at 0.6x adjusted net debt/EBITDA.

Cash conversion is less glamorous. Operating cash flow was $1.15 billion in 2025, but aircraft purchases, pre-delivery payments and intangibles consumed about $953 million. Free cash flow after all aircraft investment was only $197 million by my calculation. In 2024 it was about $501 million and in 2023 about $443 million.

The next aircraft matters more than last year’s ROIC.

Copa can still compound value while consuming capital. It simply has to earn a high enough return on the capital it keeps putting back into the fleet. That is a harder test than pointing to one year’s accounting margin, and it is the test 2026 is starting to answer.

The moat is the hub

Calling Copa the Panamanian flag carrier is technically correct and economically misleading. Panama is too small to explain the airline. The product is connectivity between other cities in the Americas through Tocumen.

Source: Copa Holdings, 2025 Investor Day.

Panama lets Copa aggregate thin demand across a network that has been built up over decades.

Copa says the hub offers nonstop or one-stop service across more than 5,000 city pairs. At Investor Day, management showed that about 80% of the origin-destination combinations in its network have fewer than 20 passengers a day in each direction. Those markets are usually too thin to support convenient nonstop service.

Panama solves the density problem. Copa can collect passengers from dozens of origins, connect them through one banked hub, and redistribute them across dozens of destinations. The aircraft are replaceable. The accumulated network is not.

A third-party OAG schedule analysis of about 120,000 Copa flight frequencies makes the pattern more concrete. In the 9M26 schedule, 69% of routes were Copa-exclusive. Among the 67 routes where frequency was increasing, 82% were exclusive and Copa’s ASM share was about 88%. On the routes where capacity was being cut, its share was closer to 53%. Competitor capacity was also falling on the contested routes where Copa was adding seats.

Those route statistics are useful; the calls show how management behaves when conditions change. Argentina became one of Copa’s strongest markets, competitors piled in, and yields softened. Pedro Heilbron did not argue that the moat made supply irrelevant. He said Copa would not be growing there as much as it had over the prior year. That is the behavior I want from an airline manager: move the airplane.

Panama also keeps the operating model simple. At June 30, all 131 aircraft in the fleet were Boeing 737 variants. One narrowbody family covers most of the Americas from a centrally located, sea-level hub. Crew training, maintenance, spare parts and scheduling are simpler than at a multi-fleet network airline.

That simplicity shows up in ex-fuel CASM: 5.8 cents in 2025 and 5.7 cents in 2Q26 despite the 16.5% capacity increase. Copa gets much of the revenue opportunity of a connecting network while retaining a relatively simple narrowbody cost base.

Panama’s tax regime helps too. Income from Copa’s foreign operations is generally outside Panamanian income tax, while traffic originating or terminating in Panama is taxed under the local rules. The consolidated effective tax rate was 13.4% in 2025. That tax treatment amplifies the returns already created by the network.

Independent research on overlooked businesses and mispriced securities.

Subscribe free to receive the next deep dive.

What the next aircraft cost

This is where the story becomes dangerous. At Investor Day, Copa showed 46 MAX deliveries scheduled from 2026 through 2029. In April 2026 it added another 40 firm MAX orders plus 20 options for 2030-34. An order book is not a moat. It is a claim on future capital.

Across the last eight calls, the near-term deliveries look less frightening. In Q3 2024, Boeing delays forced Copa to plan less growth than management thought demand could support. By Q1 2025, management described the airline as “tight in capacity.” When deliveries caught up, much of the 2026 growth was not speculative route creation: management said about half came from the full-year effect of capacity already added in 2025, around 10% from new destinations, and the rest mainly from extra frequencies and utilization.

Management has since raised the 2026 plan to 14-15%, so the old 11-13% decomposition cannot simply be copied forward. What matters is that management re-explained the higher number after the raise rather than changing the story. Boeing delivered earlier, utilization improved, and demand was still there.

Fleet flexibility is better than the gross delivery count suggests. Older 737-700s can be retired instead of sent through expensive heavy checks. Leases can roll off. Delivery slide rights exist. Aircraft can be sold. Copa’s Investor Day downside case showed how an 8% planned fleet CAGR through 2029 could fall toward 2.5% if demand weakened.

Newer aircraft also improve the economics. MAX 8s drive most of the near-term capacity increase in the OAG schedule work and burn materially less fuel than the 737-800s they replace or supplement. Gallons per ASM are improving even while the headline fuel price gets worse.

Management has aircraft coming. It also has exits.

That does not guarantee discipline. The 2030-34 order tells us management believes the runway is long; it does not tell us management is right. But so far the evidence looks more like delayed growth being unlocked than airplanes searching for somewhere to fly.

The fuel test

Copa does not hedge fuel. In 2Q26 the all-in fuel price jumped from $2.32 to $4.28 per gallon. Fuel expense more than doubled, and operating margin fell into single digits. The pass-through mechanism is therefore visible in almost real time.

Q2 was messy because many tickets had been sold before the shock. Copa still lifted RASM 7.9%, raised fares sharply and kept ex-fuel unit costs essentially flat. Management estimated that higher revenue recovered about 40% of the incremental fuel cost during the quarter.

Full-year guidance tells a cleaner story. In February, Copa guided to $2.50 fuel, 11.2-cent RASM and a 22-24% operating margin. By August, fuel guidance had risen to $3.60 and RASM to 12.0 cents, while the operating-margin range fell to 17-19%.

Using recent fuel burn per ASM, the $1.10 increase in guided fuel price adds about 1.25 cents of fuel cost per ASM while RASM guidance rose 0.80 cents. That puts implied full-year pass-through at about 64%. Fleet mix and stage length move the exact figure, but most of the shock is already showing up in higher unit revenue.

Booking timing gives pricing more room from here. Management said only about 20% of Q3 had been sold before the fare increases and almost none of Q4. By the August call, around 75% of Q3 and 25% of Q4 were sold. Management also described competitors as showing “a lot of discipline” after fuel moved.

Then came July: capacity +16.2%, traffic +17.4%, load factor 89.7%. Heilbron said that near-90% load factor was being achieved in a higher-yield environment.

A high load factor alone proves very little. You can fill an airplane by cutting the fare. What matters here is the combination: capacity is up, yield is up, RASM is up, and utilization remains high.

There is upside if fuel falls and some of the pricing sticks. Management argues that yields entered the shock below 2019 levels even before adjusting for inflation. I do not need that outcome in the base case, but it creates an interesting asymmetry: high fuel gives Copa more time to reprice; lower fuel could restore margin faster if fares are slow to give back.

Self-help between the aircraft

Some of Copa’s improvement does not require another airplane. Direct and NDC channels reached 89% of sales by Investor Day, up from about 60% in 2022, while sales-and-distribution expense per ASM fell about 30%. That is self-help, not a fuel-cycle gift.

Ancillary revenue grew at a 34% CAGR from 2019 to 2025 versus about 4% capacity growth between those endpoints. The base was small, so I would not turn Copa into a loyalty-program story. Still, more revenue per passenger is exactly what a capital-intensive business needs.

Copa is also densifying the 737-800: more seats per aircraft while keeping a conventional business-class cabin and Economy Extra. Lower unit cost without abandoning the premium product.

A bigger operational change arrives in March 2027, when Tocumen moves from six daily connecting banks to eight. Management calls the change a RASM improver and a CASM detractor. The denser schedule should create more connection options and improve aircraft utilization, but it adds schedule complexity and cost. The economic test is whether the revenue gain beats the CASM drag.

Tocumen also has physical headroom. The airport is adding gates and apron/runway capacity alongside the bank redesign. A growth thesis built around one hub needs the airport to keep up.

Reliability is part of the product too. Copa reported 90.75% on-time performance in 2025. At a connecting carrier, a late inbound is not merely irritating; it can break two legs of the itinerary. The hub only works if the banks connect.

This is what the cash is for

Copa ended 2Q26 with $1.54 billion of cash and investments and adjusted net debt/EBITDA of 0.9x. An unhedged airline in a fuel spike should be holding cash. This is what the cash is for.

Aircraft commitments are still real. At year-end 2025, contractual aircraft and engine commitments net of discounts and PDPs were $0.2 billion for 2026, $0.5 billion for 2027, $0.9 billion for 2028 and $0.4 billion for 2029, with more thereafter. Copa finances aircraft through a mix of operating cash flow and long-term aircraft financing.

Copa’s dividend policy targets up to 40% of the previous year’s underlying consolidated net income, subject to Board approval. The current annualized payment is $6.84 a share. That exact payment should not be assumed for 2027 because the 2026 fuel hit flows into the next dividend decision.

Earnings and operating cash flow cover the payout, but free cash flow after today’s heavy aircraft investment does not. That is acceptable while the growth capital earns good returns. It becomes a problem if the new fleet is merely keeping reported growth alive.

Copa repurchased $45 million of stock in Q1 and then stopped as the fuel environment deteriorated. I read that as prudence. Fuel had just become the dominant uncertainty, the airline remained unhedged, and dozens of aircraft were still being funded. Preserving liquidity in that moment is consistent with the balance-sheet discipline that makes Copa unusual in the first place.

Two variables, not one

My first valuation pass made Copa look cleaner than it really is. I used a 22% 2027 operating margin, which produced about $21 of EPS. The arithmetic worked. The assumption was hiding in the margin: with 12-cent RASM, a 22% margin effectively needs fuel around $3.20 a gallon.

It was a fuel forecast wearing a margin assumption.

For an unhedged airline, the honest way to underwrite the earnings is to put fuel and RASM on the page together.

For the table below I assume 39.7 billion 2027 ASMs: 2025 actual capacity, 14.5% growth in 2026 at the midpoint of guidance, then 7% in 2027. Ex-fuel CASM stays at 5.7 cents, fuel burn is 1.14 gallons per 100 ASMs, net financial expense is $50 million, the tax rate is 13.5%, and the share count is 41 million. Those operating assumptions are mine.

Six times earnings is one cell, not the thesis.

I start the RASM columns at 11.6 cents rather than 2025’s 11.2. Copa already guides to 12.0 cents for 2026 at $3.60 fuel. If fuel remains $4-plus into 2027, continued repricing is more plausible than an immediate return to pre-shock fares.

That 6x outcome now sits where it belongs: visible and conditional. At 12.0-cent RASM and $3.20 fuel, I get about $21.2 of EPS. At Copa’s current $3.60 fuel assumption and the same RASM, the model gives $17.3 and 7.3x earnings. If repricing pushes RASM to 12.4 cents at $3.60 fuel, EPS rises to $20.7 and the multiple falls to 6.1x.

This is why the equity can still work without a heroic multiple. Depending on the RASM/fuel combination, plausible normalized cases sit in a wide band. The market is not offering a free lunch. It is offering a cheap claim on the spread between unit revenue and fuel.

I do not know which cell we will land in. I do not think I need to. Enough of the middle of the table works for me.

Every 0.1 cent increase in ex-fuel CASM costs about $0.84 of EPS in this model, all else equal. The eight-bank redesign therefore has to earn its keep. So do the aircraft.

If Copa can grow earnings at a healthy rate and return a meaningful share of profits, a static 7-8x multiple can still produce attractive shareholder returns. A rerating is optionality.

What breaks it

One ugly quarter does not break the thesis. A sustained change in the operating evidence would.

Boeing concentration sits on top of all of them. All 131 aircraft are 737s. Fleet simplicity is a cost advantage until a technical or regulatory problem affects the family again.

That is why I bought the stock. The operating evidence still points the right way: Copa is adding a lot of capacity, charging more, and filling the airplanes. Management has also shown that it will pull back when a market gets crowded. The balance sheet gives it time to be patient.

I do not need Copa to become a wonderful business. I need it to remain a good one.

If the Panama network can keep repricing fuel, absorb 6-8% long-term capacity growth and earn something close to mid-teens incremental returns, the valuation does not need much help from multiple expansion. If it cannot, RASM and load factor should tell us before reported ROIC does.

The next aircraft matters more than last year’s ROIC.

Independent research on overlooked businesses and mispriced securities.

Sources and notes

Source integrity: Reported figures are traced to company or regulatory sources where available. The route-level schedule statistics are third-party calculations built from OAG data. Forward estimates are my own.

Results and latest traffic evidence: Copa Holdings - 2Q26 Financial Results; July 2026 Traffic Statistics.

Audited financials, tax, cash flow, debt, aircraft commitments and fuel policy: Copa Holdings - 2025 Form 20-F.

Network density, fleet flexibility, distribution, ancillary growth and hub infrastructure: Copa Holdings - 2025 Investor Day.

OAG-based 9M26 route and schedule analysis - third-party calculations using approximately 120,000 Copa flight frequencies.

Original earnings calls, Q3 2024 through Q2 2026: Copa Holdings - Earnings-call archive.

Fleet composition: 2025 Form 20-F; 2Q26 Financial Results.

April 2026 order for 40 firm 737 MAX aircraft plus 20 options: Copa Holdings - 1Q26 Financial Results.

February versus August 2026 guidance: 4Q25 and FY2025 Financial Results; 2Q26 Financial Results.

2025 on-time performance: Copa Holdings - 4Q25 and FY2025 Financial Results.

FY2025 airline operating-margin comparison - primary sources: Copa | LATAM | IAG | Ryanair | Delta | United | American | Southwest | IATA

Three-year ROIC comparison - my calculations from the companies’ annual reports and filings. Methodology: reported operating income after tax divided by average equity plus interest-bearing debt plus lease liabilities; cash is not deducted.

$45 million share repurchase: Copa Holdings - 1Q26 Financial Results.

Valuation model: assumptions are my own; company operating inputs are calibrated to reported 2025/2026 capacity, fuel burn, guidance and share count. Primary calibration source: Copa Holdings - 2Q26 Financial Results.

This article is for research and discussion, not investment advice. Airline earnings are highly sensitive to fuel, demand, capacity, foreign exchange, regulation and aircraft availability. My estimates may be materially wrong.

Subscribe now

Helios Towers is not a telecom operator. No phone plans. No spectrum. No smartphones. It does not even own the radio equipment that sends the signal.

Instead, Helios owns the boring part: the towers.

And sometimes the boring part is where the money is made.

A mobile operator needs towers to provide coverage. But owning towers is capital-intensive, operationally messy, and inefficient if every operator builds its own mast in the same location. The better model is shared infrastructure. One company owns and operates the tower, and several mobile network operators rent space on it.

That is the towerco model.

In developed markets, tower companies such as American Tower and SBA Communications became extraordinary compounders for a simple reason. The first tenant pays for most of the economics. The second and third tenants can be added at very high incremental margins: the mast is standing, the land is leased and the power system already exists. Adding another customer is not free, but it is wonderfully profitable.

Helios applies this model in Africa and the Middle East, where mobile infrastructure is still far from mature. It owns roughly 15,000 towers, is listed in London and, at 192 pence per share, has a market value of about £2 billion.

One detail caught my attention. In May, the company raised its full-year guidance across the board. Since then, the shares have fallen about 17% from their spring high. Fundamentals up, price down. That gap is what this article is about.

A real estate business disguised as telecom infrastructure

I find Helios easier to understand as a landlord than as a telecom company.

Its tenants are mobile operators. Its properties are towers. Its rents are long-term contracts, typically 10 to 15 years in these markets, non-cancellable, with inflation and power-price escalators. Its growth comes from two sources: building or acquiring more towers, and adding more tenants to the towers it already owns.

This is why the REIT analogy is useful.

Helios is not a REIT, and the contracts are not pure triple-net leases. But economically, it has several REIT-like characteristics: long-term contracted revenue, inflation-linked rent escalators, mission-critical assets, and high visibility. The order book alone is $5.3 billion of signed future revenue, with around 70% of it owed by investment-grade counterparties and more than 70% of EBITDA earned in hard currency.

The difference is that a good tower becomes more profitable as more tenants are added.

A building with one tenant can be risky. A tower with one tenant is just the starting point. Add a second tenant, and the economics improve dramatically. Add a third, and the returns can become excellent.

This is the key to the entire investment case.

Helios’s tenancy ratio is 2.2x, with a company target of 2.5x by 2030. That means the average tower already hosts more than two tenants, but still has room to improve. Each additional tenant does not require building a new tower from scratch. It uses infrastructure that is already in place.

The company’s own investor presentations spell out the operating leverage. A new site earns around 12% ROIC with just its anchor tenant. Add a second tenant and returns jump to roughly 25%. With three or more, they exceed 34%. Same mast, same land, same power.

Slow at first. Then powerful.

Why Africa changes the equation

The growth case is not hard to understand. The harder part is believing the market can ignore it for this long.

Africa has very low fixed-line broadband penetration. In the US and Europe, most internet data travels through fixed-line networks. In many African markets, there is no equivalent fixed-line infrastructure. The mobile network is not a convenience. It is the network.

That means more people, more smartphones, more data usage, more 4G, more 5G, and more mobile broadband all point in the same direction: more demand for towers.

Tower density across Helios’s markets remains far below global averages, and subscriber penetration sits near half of European levels. The market data the company itself cites points to 94 million new mobile connections and a quadrupling of data traffic across its footprint over the next five years, requiring more than 28,000 new points of service. On top of that, roughly 19,500 towers in its markets are still owned by the operators themselves: a ready-made pipeline of future sale-and-leaseback deals. Data traffic across Africa and the Middle East is forecast to grow faster than in any other region for the rest of the decade, and sub-Saharan Africa’s population is growing at several times the pace of the developed world.

I am not buying this for the next quarter. The bet is that Africa’s mobile network buildout still has many years to run.

And Helios is not just promising growth. It is delivering it. In the first quarter of 2026, revenue grew 12%, EBITDA grew 14%, and management raised full-year guidance, including a record 3,000 to 3,500 new tenancies for the year.

What went wrong the first time

The early bulls on this stock were not wrong. They were early.

The stock struggled for years because Helios had too much debt, too much perceived emerging-market risk, and too little visible free cash flow.

Investors looked at the story and saw the growth. But they also saw leverage above 5x EBITDA, rising rates, political risk, currency risk, and heavy capex. That is not a combination the market was willing to reward. From its 2019 IPO at 115p, the stock touched nearly 190p, then collapsed below 70p by late 2023.

The company has moved from “interesting growth asset with a balance sheet problem” to a business that funds itself and returns cash.

Net debt/EBITDA has fallen from a peak of 5.1x at the end of 2022 to 3.5x today, and the company is guiding it toward the bottom of its 2.5x to 3.5x target range.

At 5x leverage, investors worry about survival, refinancing, and rates.

At 3x, they start thinking about buybacks.

Helios has crossed that bridge.

The free cash flow inflection

The most important recent development is free cash flow.

For years, Helios looked optically expensive on traditional free cash flow because it was still investing heavily. But that headline number hid the quality of the underlying tower estate.

Helios itself reports a metric called recurring free cash flow, or RFCF. This is not the same as normal free cash flow. It is closer to “cash generated by the existing estate after leases, maintenance capex, cash taxes and interest, but before discretionary growth capex.”

That distinction matters.

If Helios builds new towers, reported FCF can look modest because the cash is being reinvested. But if the new towers earn attractive returns, that is not a weakness. That is exactly what I want them to do with the cash.

The cash machine is finally working. Helios burned through nearly $80 million as recently as 2023. This year it is guiding for $215 to 230 million of recurring free cash flow, and that guidance was raised in May. The company’s own five-year plan points to more than $1.3 billion of cumulative recurring free cash flow between 2026 and 2030. That is roughly half of today’s entire market capitalisation.

The market used to ask: “When will Helios generate cash?”

Now the question is: “What will Helios do with the cash?”

The answer comes from the company’s own November 2025 Capital Markets Day targets: new sites, dividends and buybacks. EBITDA growth above 9% a year to 2030. A maiden dividend of about $25 million this year, targeted to grow to more than $150 million annually by 2030. And over $250 million of buybacks, which have already started and are already shrinking the share count.

That is the virtuous circle.

Falling leverage enables capital returns. Capital returns improve investor confidence. Investor confidence can reduce the discount rate. A lower discount rate can re-rate the equity. Meanwhile, the underlying business continues to grow.

Why the valuation still looks strange

Helios has already performed well. Over twelve months the shares are up more than 60%. This is no longer the forgotten 2023 opportunity.

But the valuation still looks odd.

At around 192p, the enterprise value is roughly $4.5 billion. That is under 10 times trailing EBITDA, and under 9 times this year’s raised guidance. Extend the company’s own 9%-plus growth target two more years, and it is roughly 7 times 2028. Listed tower peers trade between 13 and 20 times trailing EBITDA. Helios trades below 10 times, while growing faster than any of them.

This is where I struggle with the current price.

This is an infrastructure business with long-term contracts, inflation-linked revenues, structural growth, improving margins, falling leverage, and buybacks. Yet it still trades at a material discount to listed tower peers.

Some discount is fair. Helios operates in more complex markets. Political risk, FX risk and customer concentration are real. It should not trade at the same multiple as American Tower.

But should it trade at half the multiple while growing faster?

That is the question.

More importantly, I do not need a heroic re-rating for the investment to work.

Suppose the multiple never moves. EBITDA compounds at the company’s targeted 9% or better, so at a constant multiple the enterprise value grows at the same pace. Net debt is flat to falling, which means the equity grows faster than the enterprise value. Add a dividend yield above 1% and rising, plus a share count shrinking from buybacks, and you get a low-teens annual return with the multiple going nowhere. A partial closing of the peer discount would be welcome, but it is dessert, not dinner.

There is also a new scarcity angle. In February, MTN, Africa’s largest mobile operator, agreed to take IHS Towers private at a $6.2 billion enterprise value. The deal has not yet closed, but once it does, Helios becomes the last listed pure-play on African towers. Anyone who wants this exposure through public markets will have exactly one option. And in a consolidating sector, the only remaining public company is also the only remaining public target.

My view is that the market is still anchored to the old Helios: high leverage, negative FCF, emerging-market risk, no shareholder returns.

The new Helios is different: lower leverage, positive FCF, accelerating EBITDA, organic growth, and capital returns.

The rerating may have started, but it does not look finished. And the recent pullback has widened the gap again: the shares are down about 17% from the May high, in the same quarter that guidance was raised.

The moat

The moat is not technology. There is nothing magical about a metal tower.

The moat is location, contracts, economics and operational capability.

Once a tower exists in the right location, building another one nearby usually makes little economic sense. The second tower would split the same pool of demand while requiring new land, permitting, power, security and maintenance. The incumbent tower has the advantage because it can add another tenant at high incremental margins.

In African markets, power is also part of the moat. Running towers reliably in difficult environments is not trivial. Backup power, batteries, diesel, solar, logistics and security all matter. Telecom operators would rather focus on customers, spectrum and networks than manage thousands of distributed power sites.

That is why tower outsourcing exists.

And that is why scale matters.

Helios was not built in a vacuum, either. Its founding backers in 2010 included the IFC, the World Bank’s private-sector arm, alongside George Soros’s Quantum fund, Madeleine Albright’s investment firm and RIT Capital Partners. Development finance and sophisticated private capital showed up early, precisely because mobile infrastructure was one of the few African infrastructure categories where the economics were tangible. That history matters. It partly explains why the company has been able to refinance, issue debt, and attract global shareholders despite operating in markets many investors still avoid.

What could go wrong

The risks are real.

First, Helios operates in countries with higher political, regulatory and currency risk. The Democratic Republic of Congo, its largest market, has knocked the stock before. More than 70% of EBITDA in hard currency softens the blow, but the local-currency tail is real, and a cheap multiple is partly justified.

Second, the customer base is concentrated. No single customer exceeds 28% of revenue, but a handful of large operator groups still pay most of the rent. If a major one weakens, delays payment, or reduces capex, Helios’s growth could slow.

Third, capital allocation matters. If management chases M&A too aggressively, the market could again worry about leverage.

Fourth, rates matter. Tower companies are long-duration infrastructure assets. Higher discount rates hurt valuations. That is what crushed this stock in 2022, and it would hurt again.

Fifth, satellites. Starlink is live in a growing number of African markets, and the obvious question is whether low-orbit satellites make towers obsolete. I do not think so. Satellites solve rural coverage, not urban capacity. The cost per gigabyte is not competitive for mass-market mobile data. And operators increasingly use satellite links as backhaul for remote towers, which runs through this infrastructure rather than around it. But it is a technology risk worth watching, and I would rather name it than pretend it does not exist.

Finally, the market may keep applying an emerging-market discount for longer than bulls expect.

This is not a risk-free compounder.

But it may be a misunderstood one.

The thesis in one paragraph

I own Helios because it controls the towers carrying African mobile data in markets where the mobile network is often the only network, under contracts that behave much like high-quality infrastructure leases. The bull case was early for years because leverage was too high and free cash flow was not yet visible. Today, leverage is falling, cash flow has inflected, dividends and buybacks have started, and the company still has years of new sites, new tenants and rising margins ahead of it. The stock has rerated, but at under 9 times this year’s EBITDA it still looks too cheap for a business expected to compound EBITDA at double-digit rates while returning capital to shareholders. And if the multiple never moves, the growth, the dividend and the buyback still do the work.

The old thesis was “one day this could become a great infrastructure compounder.”

The updated thesis is simpler:

That day may have arrived.

Half-year results land on 30 July. The numbers to watch: tenancy additions against the raised guidance, and leverage continuing down toward 3x. I will be watching. You should too.

À bientôt.

Ticker: HTWS (London Stock Exchange). ISIN: GB00BJVQC708. Share price at time of writing: about 192p (delayed quote, 9 July 2026); market cap about £2.0bn / US$2.7bn. Sources: company results, investor presentations and Capital Markets Day materials at heliostowers.com, plus public market data.

Disclosure: I own shares in Helios Towers. This article is for information only and is not investment advice. Emerging-market infrastructure carries significant currency, political and liquidity risk. Do your own research.