I manage the telecom infrastructure procurement budget for a regional network operator—about $2.1 million a year when you add site leases, power, backhaul, and the occasional emergency generator. I have tracked every invoice for the last six years. That makes me the annoying person who asks 'why is there a line item for a line item.'

One headline I keep seeing: American Tower Corporation (AMT) capex up. It is easy to read that and assume lease rates are about to rise. It is not that simple. This article compares two infrastructure options for mobile operators and data-center tenants: leasing space on AMT's shared assets versus building and owning your own site.

This is not a 'rent vs buy your house' argument. It is more like comparing a flip phone and a clear phone. Both are phones, but one has a manual and one has a hidden glass back. You need to know which one you are paying for.

What I Am Actually Comparing

Let me define the comparison before the debate gets noisy. I use a total cost of ownership (TCO) view over a 10-year period. Four dimensions:

  1. Up-front cost and long-term cost
  2. Time to bring a site live
  3. Control and flexibility when technology changes
  4. Hidden costs buried in contract language

My experience is based on roughly 130 lease agreements and 12 build-to-suit projects. If you are a national carrier with different bargaining power, your numbers will look different. I cannot speak to that level.

Dimension 1: Up-Front Cost and TCO

Leasing obviously wins on up-front cost. No land purchase, no foundation, no tower steel, no generator. For one site, that is a huge advantage. But the comparison gets less obvious when you roll in 10 years of lease escalators, annual rent bumps, and the cost of every amendment for added equipment weight or space. I have seen leases where a 'small' change order doubles the effective rent by year five.

When AMT says capex is up, I read it as 'we are investing in assets that will keep leases sticky.' That is their business model. It does not automatically mean building is cheaper. But it does mean you should stress-test your renewal assumptions.

Verdict: For a single site, lease. For a cluster of sites in the same geographic area, build after the first couple of years if you can standardize the design.

Concrete example: we compared a 3-site build against leasing for 10 years. The build had an ugly capex spike—about $480,000 total if you count backhaul and fencing. Leasing those three sites was $72,000 per year before escalators. At first glance, leasing wins. But after in-lease pass-through fees and 3.5% annual escalators, the 10-year cost crossed over in year seven. I know because I track this in a spreadsheet no one else enjoys.

Dimension 2: Speed to Deployment

Leasing wins. No question. If your spectrum license and coverage gap demand a site in 12 months, you do not build. Zoning, permits, power utility lead time, steel procurement—any one of those can kill the timeline.

With AMT or another tower operator, the site already exists. The structural analysis might take a few weeks. The lease negotiation takes a while if procurement is involved, but it is still faster than a greenfield build.

But there is a nuance. Last week, my junior analyst asked me 'how do you turn on a flip phone.' I laughed, then explained that some technology needs a manual even when it looks obvious. The same applies to site builds. We also have an old Platinum BP5450 in the equipment closet—no one remembers why—but it is a good reminder that 'simple' does not mean 'intuitive.'

Verdict: Lease when time-to-market is the top priority. Build only if you have an in-house construction manager and a tested vendor list.

Dimension 3: Control and Flexibility

This is the surprising one. Most people assume leasing is more flexible because there is no ownership. A 10-year lease with a termination option is lower commitment than a 30-year depreciable asset. That is true on paper.

In practice, tower leases come with operational restrictions. You cannot install new antennas without a structural review. You cannot reconfigure for a new technology without landlord approval. You cannot add a competing tenant without a subtenant clause. If your network strategy shifts from macro coverage to small cells or edge data centers, your lease still carries costs for a site you no longer need.

With an owned site, you can decommission, repurpose, or sell it. You control the power enclosure, the fiber entry, the spacing. That control has a price—but it is a predictable price.

I have mixed feelings about this dimension. On one hand, owning an obsolete macro site is worse than leasing one because you cannot walk away. On the other hand, long-term leases can feel like a flip phone with no off switch.

Verdict: If your demand profile is stable and you are confident the site will stay relevant, lease. If you expect to shift from 4G to 5G to something else before the lease expires, owned infrastructure gives you more room.

Dimension 4: Hidden Costs and Contract Language

This is where my job gets interesting. I once compared two lease proposals. One had a base rent of $1,500 per month. The second was $1,250 per month. I almost signed the second until I calculated the full TCO. The cheaper one charged a 'site management fee' that increased annually, a power pass-through calculated on a floor rather than actual usage, and an administrative fee on every amendment. The $1,500 proposal included everything except electricity. The difference was 23% hidden in fine print.

That 'clear phone' idea—wanting transparent pricing—is rarer than you think. I have learned to ask for a line-item breakdown of total annual cost, not just base rent. 'We are at a 4% escalator' is not enough. I need to know the escalator's footing: is it on base rent only, or on full rent plus pass-throughs? That one detail changes the math by thousands.

Last week, I opened a renewed lease and spent 10 minutes looking for the termination clause. It was there, but only if we gave 18 months notice and paid a 'decommissioning fee' equal to six months of rent. That is not a termination option. That is a reminder that the contract is not your friend.

Build projects have hidden costs too. Property taxes, insurance, NOC monitoring, generator maintenance, and the cost of a site lease for land you do not own. I do not want to romanticize ownership. But at least those costs are visible in a capital budgeting process, not buried in a monthly invoice.

Verdict: Leases win on transparency only if the lease is truly all-in. Most are not. Build wins when you have procurement discipline and an operating manual you are willing to follow.

What This Means When AMT Capex Is Up

I do not know every reason behind AMT's capex spending. My best guess is that it is tied to edge data centers and fiber upgrades around existing sites, not just ground-up towers. That could make AMT leases more valuable if you want access to connected infrastructure. It could also mean higher future rents as they recover that capital. AMT's 2024 investor materials, available through their IR page, pointed in that direction.

So the strategic question for a cost controller is not 'Is AMT good or bad?' It is 'Does my network plan need the things AMT is building?' If yes, lease with strong terms. If no, consider whether self-funding a smaller, specific deployment gives you a cheaper long-term position.

So Which Should You Choose?

Here is my honest recommendation, with all the limits of my experience:

  • Lease from AMT or another operator if you need speed, you want to avoid structural risk, and your site count is under 10.
  • Build if you have predictable 10-year demand, a good vendor list, and enough volume to standardize.
  • Do a hybrid if you are not sure. I am a big fan of a primary plus backup strategy. Lease the anchor sites, build the high-traffic edge sites, and keep every contract in a single TCO dashboard.

And if you ever get stuck on a question as simple as 'how do you turn on a flip phone' or a contract as complicated as a tower lease, ask the person who has the manual. In my office, that is me.

If you have only worked with one segment of infrastructure, your experience might differ. I have only managed regional builds and mid-sized tower leases. National or residential-segment dynamics? Not my world. I say that so you can apply the right amount of skepticism to my conclusions.

Technical planning note: validate insertion loss dB, PIM dBc, grounding resistance, and relevant 3GPP TS 38.xxx requirements before final RAN acceptance.