ConnectX was earlier this month in New Orleans which I was personally ecstatic about. I love New Orleans and was thrilled that this year’s conference was there, especially right after the last weekend of JazzFest. Also, I enjoyed great conversations about the industry with some smart people. Here are some of my takeaways from those discussions.
1. In the last 4 years, there has been a direct correlation between revenue growth for the tower companies and 5G capex. The public and private tower companies have enjoyed robust growth from new collocations and modifications related to 5G. It is no secret that carriers are reducing capex spending this year. What is not well known though is how much each carrier is reducing wireless capex specifically. Each carrier will be prioritizing different objectives with their capex spend this year. For T-Mobile and Verizon, each appears to be pulling back on 5G wireless capex and AT&T has significantly reduced wireless capex (See note 3 below). DISH is slowing down and may not build out as much of the US as planned. (See note 4 below) The tower execs say that 50% of their towers have 5G on them already, although I suspect the % is higher. While it is safe to say that all towers will eventually have 5G on them- the question that is not being asked/answered effectively is when. Given the tepid consumer reaction to 5G, is the pullback on 5G densification related to the need to start generating revenue from 5G expenditures? (Verizon just announced today that they were charging $15/month extra for access to mid-band 5G).
2. Today’s newest justification for future lease-up on towers? Anything AI (Artificial Intelligence). Quite a few mentions of AI in conference discussions, but few clear examples of how 5G is needed to accommodate it (at least not yet.) I read a tweet the other day where the author said something surprisingly profound- that today’s AI is the dumbest AI we will see in our lifetimes. That stuck with me because I have been amazed at what ChatGPT-3.5 and Midjourney can do already, despite the limitations. ChatGPT’s app is the fastest growing in history in terms of user downloads- so it is no surprise that tower companies would want to extol how AI will beget more need for 5G and therefore for towers. ChatGPT-4 is better. In the skate to where the puck will be category- it seems to me that we all should be taking a hard look at what the future of AI holds and extrapolating how that translates to consumer and enterprise behavior on mobile devices.
The key limitation to me of Natural Language Processing AI like ChatGPT is that I do not trust feeding any proprietary data into them. Who knows what happens to the data after it is submitted? Who sees it and who has access to it? Enterprises will look for AI tools that are secure from end to end. A slew of new startups aim to address the limitations of protection for proprietary data in ChatGPT. For AI applications in the office/building/factory, secure communications are easier to facilitate using fiber or private networking. But for remote mobile or nomadic access to these tools while on the road, the connection will be via cellular/5G. As much of the processing will occur in the cloud as opposed to on the device, the initial demands on 5G networks may be limited. But as more devices are connected or controlled by AI or Artificial General Intelligence (AGI), it is easy to see a huge leap forward in terms of total # of connected devices which begets the need for better 5G.
I can envision a future where instead of having to program specific directions to multiple devices, I would instead be able to tell my AI what the overall result is I am looking for and it will translate that into directions and send them out simultaneously to both fixed and mobile connected devices. I could also tell my AI to adjust based upon external factors or data on the fly without my involvement. But if I can do it, so can nefarious actors absent a secure connection. Again, for mobile connections or truly nomadic- these will be more secure on 4G/5G.
And how about robots? Elon Musk sees a future where Tesla makes more robots than cars. Some of these robots will likely be independent devices, but others may rely upon a connection to the cloud for advanced processing and direction. For robots that are nomadic or mobile, or those that are “piloted” by a remote human, there will be a need for high throughput and possibly low latency communications. Again- another possible use case for AI and 5G.
The exact future use cases for AI/5G may not be apparent yet- but they are coming, and we can guarantee that the carriers are looking into them. I wouldn’t personally invest in tower companies because of AI, I do see a future where the need for 5G is greater than it is now and where tower companies stand to benefit. These are just my preliminary ramblings about AI/5G – in all candor, I suspect they are the dumbest ones I will ever have in my lifetime. 😊
3. We heard several times that AT&T is focused on fiber this year and that wireless will take a back seat. One industry vet suggested that some wireless construction firms that do a significant amount of work for AT&T are considering scaling back and some are considering leaving wireless altogether. For me, it is hard to rectify the claim that companies are considering leaving wireless while the wireless industry and WIA are pushing “lack of workers” claims left and right. Can both be true? Yes- especially if the issue is not really that there are not enough workers, but that carriers are not willing to pay enough for the workers that are trained.
4. DISH appears likely to request a delay in their 2025 buildout obligations. Generally, the industry expects them to meet their June 2023 deadline, but there is increasing skepticism that Charlie truly intends to build out a nationwide network. With a bunch of markets/sites going on-air within the next month or so, we will see whether consumers will switch to DISH service. Not hard to see a scenario though where subscriber growth is slow to take route and DISH’s debt service (13.5%) discourages DISH from wanting to incur more debt to build out areas of the country already covered by their MVNO agreements with AT&T and T-Mobile. If DISH is successful in delaying or doing away with their 2025 buildout obligations, that is 10,000-15,000 leases that will be delayed or will not be signed at all.
5. It is still an exciting time to be a tower developer. Purchase multiples and valuations for tower assets and cell tower leases are still at or close to market peaks. In discussions, I clearly am not the only one that wonders how long it will last. However, higher borrowing costs have not impacted on the market yet, primarily because many buyers already have dedicated tranches of capital at lower rates and can still afford to pay for quality assets. (As a side note, we have observed a pullback though on valuations for more questionable assets – like those that are not anchored by cellular providers or that are in areas with no zoning or with competing structures nearby.) But while some of the larger tower companies bemoaned the high valuations of their towers (while their own valuations were 25-40% less), the general feeling I came away with is that there is a lot more dry powder left.
6. It is also an exciting time to be a company focused on private networks or in-building. Numerous investors (including public tower companies) are looking to invest in this space. Over the last few weeks alone, we have talked to two separate infrastructure funds that are looking to place $25M or $50M in equity into the sector. Mind you, there are few private network or in-building companies out there that need $25M in equity or that have the valuation that supports that type of investment. I suspect there are more investors looking to place large investments than companies that are large enough to warrant them at this time.
7. In the “perhaps too soon” category, there was a decent amount of chatter that small-cell development will pick up in the US. Crown Castle indicated in their recent earnings call that they are still on target for approximately 10,000 new small cells. Verizon and T-Mobile both seem to be more active this year but given the development timeline from search ring to standing small cell of 12-24 months, this looks to be a 2024 story. I would not want to be a small cell pole manufacturer right now- the industry is nowhere near the projected 800,000 to 1,000,000 small cells that it alleged would be needed and I strongly doubt it will get there. These forecasts surely encouraged several manufacturers to design and sell small cell poles. I must wonder though how many of these companies would have decided not to enter that side of the business if they knew what the last 5 years would have looked like then.
Next year’s ConnectX is in Atlanta. Your guess is as good as mine on what will transpire between now and then. In the meantime, if you want to discuss any of these topics, please don’t hesitate to reach out.