On September 9, Scott McElroy, VP of Technological Realization at AT&T (T), announced HSPA would be available in six markets by the end of the year. HSPA is a software and infrastructure upgrade that doubles capacity for wireless devices from 3.6 Mb/s to 7.2 mb/s. AT&T has been adding cell sights instead of upgrading software and fiber. HSPA software upgrades require copper lines to be upgraded to fiber because the copper backhaul simply cannot carry the amount of traffic that the HSPA software can process.
An upgrade path that included HSPA+ with 21 Mb/s capacity was previously announced, but was conspicuously absent from this speech. On September 15, Kris Rinne, VP of Architecture and Planning explained why: AT&T has decided to "focus on LTE rather [than] HSPA+." In other words, AT&T is going straight to LTE for a 2011 rollout that matches the Verizon's (VZ) LTE upgrade path. This is big news for investors because the inevitable LTE buildout is going to be sooner and faster, thus more expensive than previously thought.
Rinne demonstrated in her presentation that data traffic on the 3G network had increased by almost 5000% in three years. (I know, these numbers keep getting bigger - I just report it, I do not create it.) She noted that mobile broadband, email and M2M connections had increased traffic substantially -- on top of the iPhone.
LTE delivers a megabit of data for three percent of the cost of basic EDGE 3G, while HSPA+ delivers a megabit for fourteen percent. The conclusion is that AT&T believes that there will be enough traffic to justify the extra expense of the LTE capacity upgrade.
I own AT&T long-term. The issues I see: enterprise networking equipment is rapidly becoming less expensive yet more powerful, and that will reduce AT&T's revenues. I am not concerned about wireline, that copper is going to get used for something; I have no insight, I'm just confident that connected wire is not whale oil. Data traffic seems like a winner, reduce cost per Mb/s and increase data wildly; however, the experience of Japan and Europe, which adopted faster devices earlier, is that those telecoms have not in fact made more money. That could have more to do with government regulation, but that analysis is beyond me; thus, I consider AT&T a hold.
Most important new factor is that AT&T and VZ are hurrying, which means they will be competing for services and equipment, thus less able to pressure networking prices as much as they could if they were moving methodically. AT&T's new domain purchasing systems means that AT&T is not dealing directly with most of its suppliers, so profit margins from companies further downstream will be slightly effected. CDMA and other 3G products will continue to sell in Tier 2/3 carriers and third world countries, but CDMA equipment was already under pressure and some companies (Nortel Networks in particular) were losing money on CDMA products.
This article updates earlier articles that you can review for evaluations of telecom networking sectors. Only 20% of Tier 1 backhaul is fiber, so a huge buildout will occur over the next several years. ALU and ERIC are AT&T's biggest network suppliers. With intense international price pressure from Huawei and ZTE (ZTCOF.PK), I am avoiding those companies, but ERIC is the healthier of the two.
CSCO, CIEN, JNPR, TLAB, ADCT, CTV and MOT are good bets for significant involvement. NSN, NT, NEC (NELTY.PK), Fujitsu (FJTSY.PK) and Hitachi (HIT) have been significantly involved in the past. BRCD, ELX, and QLGC are good bets for SAN/LAN equipment (MLNX is growing but small, visibility is limited). Some chip companies will also do well but they are outside my interest (ALTR, XLNX, MRVC etc). NT's 40Gb optical equipment has been dominant in the past, but since NT is BK visibility is limited-- will CIEN buy NT's MEN division, or someone else, or will AT&T go another direction? PWAV, ADTN, SCMR, ZHNE and HSTX are small cap value companies that are AT&T customers as well; I hear Cramer is excited about a couple of these stocks, check his site if interested. I sold HSTX after his recommendation bumped it.
Everyone says they are ready and have bench tests to prove it, but in fact field tests are never the same. Motorola (MOT) and Starent (STAR) are already building out LTE in Asia. STAR has a high PE, high GM, high growth, and are stealing market share from other packet software companies. MOT is also a microwave backhaul leader. Testing becomes more important as network speeds increase. Agilent (A), JDSUniphase (JDSU) are the biggest testing companies and XXIA and EXFO are strong as well; there is little upside on A at 60 PEx, my JDSU price target is $11.
Optical companies have consolidated significantly in the last year, but pricing pressure is still intense; chip, equipment and system builders have generally been losing significant money and more consolidation is expected before the sector becomes healthy. However, most of these companies are priced around 1X revenue so value in the sector is possible. FNSR is the number one component and subsystem supplier. Their active optical cables are being adopted already, and they are reverse-splitting shares one for eight on September 25th. When 40Gb modulators begin selling again, FNSR will be profitable, my short-term price target is $2 pre/ $16 post reverse split. JDSU is number two, and Opnext (OPXT) number three. Oclaro (OCLR) has never made money but the technology is above average, my short-term price target is over $2. INFN has heavy long-haul competition, has little upside at $8 and is being squeezed into an ever-smaller niche long-term, but has the potential to surprise by during the next four quarters.
In order to handle the expected increase in IP traffic, T, VZ, S, CLWR and DT need to lay billions of fiber and optical equipment while upgrading networking equipment as well. Faster networks are less commoditized. The companies with stronger software and higher profit margins are the better bets to take a larger share of the LTE buildout.
Generated with Seeking Alpha transcripts and the above referenced TelephonyOnline article
Disclosure: long OCLR OPXT FNSR JDSU STAR MOT CIEN ADCT CTV PWAV CRNT ELX ZHNE