Jeremy Poland/iStock via Getty ImagesThe satellite division of Space Exploration Technologies (SPCX), known as Starlink, announced plans to access at least 80,000 cell sites in the U.S.This has massive implications for American Tower Corporation (AMT), which owns 41,764 cell towers in the U.S. Specifically, we see strategic derisking and direct AFFO accretion as leases commence. AMT is trading at opportunistic pricing at about 16X forward AFFO and we believe the stock will go higher on AFFO accretion of as much as $3.54 per share.Strategic deriskingThe tower REITs have always maintained that satellites were a complementary technology to terrestrial communications equipment, while the market has at times feared that a sufficiently built-out satellite network would obsolete macro towers.Starlink’s intentions to obtain such a robust terrestrial network largely confirm that the tower REITs were right. Satellites are indeed complementary, achieving great results in rural and hard-to-reach areas, while macro towers remain superior for populated and well-covered areas.AFFO accretion for AMT on Starlink’s leasingIn this section we will attempt to calculate how much AFFO/share AMT could gain from leasing to Starlink.Allow me to lead with a disclaimer that these are preliminary estimates with very wide margins for error. I will explain the reasoning behind the estimates as well as reasons the numbers could turn out to be off. We will be using 2 different methodologies for calculating potential leasing revenues:Overall market perspectiveTower ratiosOverall market perspective estimate of accretionVerizon Communications (VZ) has about 70,000 cell sites in the U.S.AT&T (T) has about 75,000 cell sites in the U.S.T-Mobile (TMUS) is a bit harder to track because they still have some Sprint towers that will churn off. TMUS likely has slightly more than T and V.Given these numbers, the minimum 80,000 cell sites in the U.S. proposed by Starlink would put them on par with the other majors.This would make them a true 4th tenant for the tower REITs. In theory, that would increase U.S. tower revenues by 33%. In 2Q26, U.S. and Canadian towers had revenues of $1.274BAMTThe vast majority of the U.S./Canada segment is U.S. as AMT owns 41,764 towers in the U.S. and only 250 in Canada as of 2Q26 (which they have since sold).Let’s estimate that at $1.25B in quarterly revenue, or $5B annually. Moving from 3 tenants to 4 would, in theory, increase revenue by 33%, or $1.65B. Since the existing towers have sufficient space for additional equipment, there is virtually no incremental cost in moving from 3 tenants to 4. Thus, the $1.65B incremental revenue would largely translate to incremental AFFO.Divided by 465.9 million shares outstanding, that would be an additional $3.54 AFFO/share.Think of this as a back-of-the-napkin ballpark kind of estimate. Leases are still in the early stages of being negotiated, and nothing is set in stone. With that in mind, here are a couple of factors that could cause the actual impact to be higher or lower than this estimate.The actual impact could be higherIncumbent telcos own some of their own towers, while Starlink will likely have to rent most or all of their towers. Thus, the 80,000 of Starlink might hit harder on lease revenue than the 75,000 of T.Reasons actual impact could be lowerLess equipment per tower: While Starlink is targeting a similar number of towers to the other majors, they have far fewer customers at this point. Customer data usage is the primary driver of backhaul volume, so Starlink would likely need less equipment per tower until its customer numbers grow considerably. Less equipment would likely mean less revenue per tower leased.Slow build: Elon Musk companies have a history of hitting their estimates, but often significantly later than originally projected. I would anticipate the 80,000 cell site figure will take at least a few years to be realized. So even if the $3.54 AFFO/share is reasonably accurate accretion for AMT, only a small portion of that will be realized in 2027 with most probably 2028 or later.Estimate of accretion based on tower ratiosThere is a bit of fuzziness regarding what 80,000 cell sites means depending on what one means by a “cell site”.If Starlink means 80,000 towers, it is a very high number, as WIA estimates that there are 158,500 towers in the U.S. I suspect cell sites as used by Starlink refers to macrocell sites. In that case, there are 254,850 sites in the U.S.WIAUsing the more conservative number, Starlink would lease roughly 31.4% of towers in the U.S.AMT has a relatively large share of towers, owning 41,764. AMT’s towers are more akin to the stricter purpose-built tower definition, so AMT owns approximately 25% of the entire U.S. tower market.AMTSupplementalDepending on whether Starlink’s 80,000 guidance was referring to strict purpose-built towers or macrocell sites, they would be leasing somewhere between 31.4% and 50% of towers.That would be approximately 13,113 to 20,882 leases on AMT towers. Again, I think it is more likely they meant macrocell sites, so our calculations will use the 13,113 number.According to Digital Infrastructure, the average lease rate for a carrier is between $18,000 and $42,000 per year.Below are sample MLAs between tower REITs and carriers. This set averages $1,900 per month or $22,800 per year.Digital InfrastructureHowever, the MLAs are just the base revenue. As these tenants add more equipment, they pay bonus rent. Overall rental income is substantially higher.AMT’s 41,764 towers in the U.S. are generating annual revenue of just over $5B. That is roughly $120,000 per tower per year. Since there are only 3 main tenants, that implies that each tenant is paying approximately $40K per year, all-inclusive of escalators and bonus rent for additional equipment.If we assume similar numbers, the incremental revenue to AMT from 13,113 Starlink leases would be about $299 million at first if Starlink has minimal equipment.As Starlink gets to an equipment volume more comparable to the other carriers, lease revenue to AMT would move up to closer to $524 million ($40,000 × 13,113).Divided by 465.9 million shares outstanding, that is $1.13 AFFO/share accretion.Reasons this estimate might be lowBetter capture:We assumed AMT would get proportional capture, but it is likely that AMT would get a significantly higher percentage of leases than their raw percentage of towers. There are 2 reasons for this:Tower quality: AMT’s towers, along with those of CCI and SBAC are true institutional-grade macro towers and well above the national average qualityMany competing towers are not available for lease. Towers owned by V or TMUS are unlikely to be leased to Starlink because Telcos see Starlink as a competitor who is trying to put them out of business.Reasons this estimate might be highIt will likely take many years for towers leased by Starlink to get up to an equipment volume that would result in $40K per tower per year.Overall estimateDepending on how we estimate the earnings accretion, we think it will be somewhere between $1.13 and $3.54 AFFO/share.2026 AFFO is estimated at $11.11 so it represents a growth of 10% to 31%.That is a lot, but it is not all that surprising, as the tower REITs lost roughly that much AFFO when they lost Sprint as their 4th tenant.Going from 4 tenants to 3 was quite painful for the sector, and going back up to 4 would be bullish.AMT’s CEO Steven Vondran was clearly bullish on Starlink news on the 3Q26 earnings call:“We are really excited about the prospect of maybe having a new customer at scale in the U.S., and we're really encouraged by their public statements and what they're saying that they plan to do and also by the concrete steps that they're taking to build that layer cake of spectrum that you need to be a viable player in the U.S. market. So those are all really encouraging signs that we're seeing there. It also reinforces what we've been saying all along, which is satellites are a complementary technology to terrestrial networks that you need a terrestrial presence if you're going to try to operate in the wireless space in the U.S. And so this -- again, it's playing out the way we thought it was going to. You're seeing some positive momentum there. In terms of specifics with us with Starlink, we're not going to get into any specific discussions that we're having. But you should assume that anybody who wants to build a network in the U.S. at scale is going to come to American Tower as one of their premier tower vendors, if not the premier tower vendor. We have an unmatched portfolio in the U.S. in terms of where our sites are located, the quality of those sites, and we can get them on air faster than anybody else.”Risks to AMTWith regard to the Starlink network buildout, we see 2 main risks to AMT.It is still entirely possible that it doesn’t happen. Contracts are not yet signed, and Starlink still has relatively few customers.Starlink could pose a threat to incumbent carriers. While the incremental Starlink revenue is a clear positive for AMT, their success could weaken the credit of existing tenants.Summing it upAMT is trading at just 16X AFFO which is about right for their existing growth of maybe 4% annually. However, if Starlink becomes a legitimate 4th tenant, it could add an additional 10-31 percentage points of growth beyond the organic growth AMT already has.With this additional growth, AMT would clearly be undervalued and could trade up significantly.