AT&T announced recently that CAPEX for the U-Verse IPTV & Fiber to the Node initiative (known as Project Lightspeed) would increase from $4.6B to $6.5B. They also announced the scope of the project was being reduced from 19M to 18M homes. This is a sizable increase (41%) in capex for a project that was designed to minimize cost.
WSJ article mentions that:

“AT&T blamed the price increase in the cost of adding more servers, which are needed as the company significantly increases the amount of high-definition channels it plans to offer. AT&T is also paying a premium to its equipment vendors to ensure gear is on hand when needed, the company said”

This significantly changes the economics supporting AT&T's decision to avoid laying fiber. AT&T could deliver video services at 1/3 the cost of Verizon. With the slip AT&T's cost per home went up nearly 50% and is almost 1/2 the cost Verizon FiOS FTTH solution.

The Microsoft IPTV architecture (that AT&T uses) has two types of servers: A-server and D-server. A-servers scale with channels and D-servers scale with subscribers. There are many many more D-servers than A-servers. Server cost increase due to adding HD channels will be much closer to $1.8M than to $1.8B (even on a logarithmic scale),so the argument that the cost is attributed to the increase in servers on account of  “ the cost of adding more servers, which are needed as the company significantly increases the amount of high-definition channels it plans to offer.”  is questionable!!

Also the EVs who supply the Equipments are usually under the contractual monetary penalties on late equipment shipments.

The real reason for the revision of the project cost is most likely higher because, as always, projections were too optimistic (which is also the case regarding project schedule).

Contribution: Ashish Chandrashekhar