AT&T has so much overcapacity it gave Dish a great deal on the cost of bandwidth. That allows Charlie to build a minimal network, just meeting his FCC buildout requirement of 70% in all districts.
He can build a lousy network, falling back on AT&T to cover his rear. Very preliminary signs are that’s just what he is doing.
The result: he can easily meet his buildout requirement in his $10B capex estimate. Wall Street speculations he’d have problems raising the money are now superseded. His operating costs will be higher, but they will be related to paying customers.
Dish’s 600 & 800 MHz holdings have excellent reach, especially if he designs the network for coverage, not capacity. I bet he could reach 70% with 20,000 cells, he certainly can with 30,000.
5G high quality radios + antenna are < $25K in Chinese quantities. Those are massive MIMO and more than Dish will need. US prices are much higher, of course.
If we assume 30,000 @ $50K/each, that’s1,500,000,000. At $100K each, $3B.
He’s getting everything cheap. Dish is only the second O-RAN network in the world after Rakuten. Everyone really wanted the contract; Mavenir, Samsung, Dell, etc. need reference customers. I’m sure they bid low to win and Charlie took a low bid.
Dish still needs to buy backbone routers, OSS/BSS, switches, etc, but those prices keep coming down. The construction companies are complaining ing he’s barely willing to cover costs. That’s surprising in a tight market, but they all want to get in knowing there’s more work in the next few years.
A friend with data centers turned him down because of the price demand, but he presumably will find others. Dish doesn’t need that many aggregation points. Latency in today’s switches/routers is remarkably low, allowing aggregation even 4 or 5 hops from a tower. Level 3’s “5 ms” from 90% of the country is suggestive.
AT&T has to build out mid-band for competitive 5G marketing, even though it has plenty of capacity. So they (and VZ) have massive over-capacity the next 5 years, maybe longer. Wireless growth in 2020 was only 18%. That’s a plague year, but the trend has been down.
So they could, if they wanted to, give Charlie a great deal. My number for the marginal cost of a subscriber on an existing network is $4-8; with the midband capacity bump, it is even lower.
Scott Wallsten and I calculated that $10-20 would be the right wholesale price with competition. I’ve no way to know how AT&T priced, but they might have decided to guide Charlie to buy rather than make.