We sat down with our Server analyst, Ms. Sameh Boujelbene, to discuss the Server market during the third quarter of 2016 and wanted to share some interesting findings:

In the third quarter of 2016, server shipments were down one percent year-over-year (marking the first decline in 15 quarters) and revenues were down four percent year-over-year (marking the steepest decline in 10 quarters).

In the worldwide market, Asia Pacific was the only region that did not experience a decline, helped by sales in China.

As we have previously predicted, 2015 was the last year of Enterprise server shipment growth.

Enterprise server weakness was even more pronounced during the quarter as end users were able to use virtualization to leverage existing investments without having to buy new servers. Additionally, macro-economic uncertainties prompted users to adopt Cloud services because accounting of operating expenditures is more favorable than capital expenditures (CAPEX).

Also in line with our expectations, CAPEX spending by the four leading U.S-based Cloud Service Providers (SP) (Amazon, Facebook, Google, and Microsoft) was solid for the third consecutive quarter. Year-over-year growth was broad-based across these Cloud SPs and mainly benefitted white box vendors. It appears the growth was driven by investment in the U.S. data centers where white box servers continue to dominate.

White box shipments were up 22 percent year-over-year in the quarter as their servers are still being consumed by tier 1 Cloud SPs and are starting to gain traction in tier 2.

Shipments from White box vendors also exceeded both Hewlett Packard Enterprise and Dell shipments in the quarter as both companies experienced turbulence in the Enterprise market.

The week before last, I attended Broadband World Forum (BBWF) in beautiful London to learn more about how telecom operators are getting broadband services to customers. At this year’s conference, I had the pleasure of talking to numerous telecom operators, system and component manufacturers to hear what their thoughts were on recent developments in broadband access technologies. My primary objective was to determine in which cases did it really make sense to deploy fiber all the way to the premises or only part of the way.

It was apparent that many at the conference were pushing the latter option, as the focus of this year’s conference was clearly on G.fast. Advances in the technology have made it a viable alternative to VDSL and even to fiber-based technologies like PON. The crux of the interest is that G.fast enables fast speeds over twisted pair copper or coaxial cable.

Some of the advances that are helping G.fast achieve higher capacities include 212 MHz profiles (versus the current 106 MHz) in future amendments and dynamic time allocation (DTA).

There was less of a focus on achieving gigabit speeds versus last year’s conference, and more of a focus on networks that are simply much faster. It looks like some of the allure of gigabit networks has faded due to the cost-prohibitive nature of delivering fiber all the way to premises. For example, BT has decided to deploy Fiber-to-the-Cabinet (FTTC) and then G.fast to the premises. FTTC generally involves a copper loop length of 300-350 meters. At these distances, G.fast will likely deliver speeds of 300-500 Mbps. Clearly not gigabit networks, but hardly something to scoff at.

In most broadband deployments, an operator’s existing infrastructure greatly affects what specific technologies the operator chooses to deploy next, and we expect the same between G.fast, VDSL, PON, and even Cable architectures going forward. Regardless, the next several years will be interesting for the broadband access market as operators look to deploy newer technologies into their networks.

System manufacturers with G.fast capable products include (in alphabetical order) ADTRAN, Calix, Huawei, Nokia, and ZTE. Much of the silicon powering these system manufacturers’ G.fast chips are either built in-house or powered by Broadcom or Sckipio.

Cheerio until next year BBWF, when I hope to see you in your new home of Berlin.

With IoT (Internet of Things), the cloud, smart cities, virtual/augmented reality, and self-driving cars grabbing all the media attention, it is easy to forget that voice and data services using either a mobile phone or a computer/tablet controlled by a human being accounted for 98% of worldwide carrier revenues in 2015. And with currency adjusted worldwide mobile operator revenues growing at a CAGR of 0% to 1% between 2012 and 2015 despite smartphone shipments growing at a double-digit rate over the same period (Dell’Oro Group 2H15 Carrier Economics Report), carriers are well aware what impact a maturing smartphone market will have on their topline. So it is perhaps no surprise that with the smartphone approaching its 10-year anniversary, carriers and vendors feel a tremendous sense of urgency to identify the next growth engine.

Fortunately the first phase of the 5G standard is scheduled to be complete by 2019 and there is no shortage of suggestions for the potential use cases or applications that will define the next phase in this on-going digital transformation – which ideally will be about so much more than an improved smartphone experience for today’s use cases.

While there are some high level definitions about potential 5G drivers beyond the smartphone – it is extremely difficult today to identify the winners and losers of tomorrow. This is why we disagree with the notion that use cases/applications/services need to be defined first in order for 5G to be successful. Even if the use cases are identified at this point, there is no guarantee they will turn out to be a success for the masses, as illustrated by a host of botched smartwatch launches. The urge to introduce new products before they are ready for primetime or the customers are ready could end up impacting the launch and delay any mass-market appeal. Just as it was extremely challenging in 1996 to forecast that a successful smartphone for the masses would not be launched until 2007, when the vision for a smartphone was already a reality and there were multiple smartphone products on the market, it is equally challenging today to separate the next winner from the next smartwatch.

As a result, we don’t believe it is realistic or necessary to determine the final destination at this point. But if we are headed in the right direction and regulators provide an environment that stimulates competition and innovation, then the final destination will eventually be identified.

A few weeks ago, my colleague Tam Dell’Oro and I attended industry analyst conferences held by Nokia and Ericsson at their respective headquarters in Helsinki and Stockholm.  The two companies are undergoing significant changes and used these forums to communicate their corporate strategies and visions.

Since completing its acquisition of Alcatel-Lucent in January, Nokia has been busy integrating the two companies.  Nokia executives indicated that the transition is proceeding well, but that there is still plenty of work to be done.  Executives also touted their broad network product portfolio and leading market positions.

Ericsson provided an update on their partnership with Cisco that was announced last November, and offered insight into their business transformation that includes the corporate reorganization communicated in April.  From a technology perspective, Ericsson spoke of their expansion beyond their strength in mobile technologies and infrastructure.

Interestingly, both companies highlighted similar technology and business trends that will drive their future investments and businesses.  At the top of the lists were the opportunities in the areas of IoT, 5G, Cloud, and Enterprise verticals.  Both companies emphasized that software and services are important elements of their technology and business competencies.

While some of the components of their visions are similar, the Nokia and Ericsson approaches to fulfillment are quite different.  At Dell’Oro Group, we cover and have great interest in the technology markets in which these two networking giants participate.  Over the coming years, we will be watching closely how these two companies progress against their visions.

The following is an excerpt from the OFC 2016 First News that Dell’Oro Group contributed to.

Finally, the wait is over. While it felt like an eternity, OFC week is now just around the corner. More importantly, all the companies that have been holding back their product announcements for this one week, dedicated to all things optical, can now share what they have been working on up to this point. In this early batch of first news releases, we reviewed press releases from 23 different companies.

A look at the first wave of press releases point us to a belief that the theme of OFC 2016 will be about speed and efficiency.

Regarding speed, the press releases were geared around the development of higher speed components and better testing equipment as well as 100 Gbps WDM system deployments. The companies in this category of announcements included:

ADVA Macom O-Net Communication ProLabs
Gowanda Electronics Maxim Phoenix Photonics Teledyne
Keysight Technologies Methode Polatis VI Systems

 

Regarding efficiency, the press releases announced better manufacturing and test equipment with additional functions to reduce time and waste. Furthermore, this year efficiency announcements went beyond equipment and includes software, services, and organizational changes. SDN is here and companies are positioning themselves to capture the opportunity and extract value. The companies in this category of announcements included:

ADVA Connected Fiber Luna Innovation VPI Photonics
Arden EXFO OptoTest
Aria Fiber Optic Center Polatis
Calient inTest Thermal Solution TDK

 

Further detail on each of the company announcements mentioned are included below. Please take the time to read them in preparation for OFC week, a week dedicated to all things optical.

Similar to the theme of these early press releases, here at Dell’Oro Group, we believe one of the top topics at OFC 2016 will be about higher speeds, about going beyond 100 Gbps WDM to 200 Gbps and 400 Gbps. More specifically, we believe this is the year for 200 Gbps WDM and that discussions at OFC 2016 will levitate towards higher capacity product platforms, higher speed components, and faster test equipment to support the demand for 200 Gbps WDM line cards.

In Dell’Oro Group’s latest Optical Transport research report, we find that demand for 200 Gbps WDM is outstripping earlier projections and we continue to raise our forecast with each report. While the total number of 200 Gbps wavelength shipments is still low, the associated revenue that system equipment manufacturers garner has reached a material level. We estimate that 200 Gbps WDM line card shipments resulted in approximately $230 million of revenue in 2015 and project them to drive at least $720 million in 2016.

There are three drivers for 200 Gbps wavelength demand:

  • Supply: Up until mid-2015, Alcatel-Lucent (now Nokia) was the sole vendor supplying a high volume of 200 Gbps WDM line cards. We anticipate that by mid-2016, at least seven vendors will begin volume shipments of these line cards, reducing a service provider’s dependence on a limited number of suppliers.  By 2017, all WDM vendors will likely offer a 200 Gbps line card.
  • Price: For metro applications, a 200 Gbps line card is priced favorably against a 100 Gbps line card. We estimate that in 2015, the average price of a 200 Gbps line card on a dollar-per-bit basis was approximately 15% to 20% below that of a 100 Gbps line card.
  • Efficiency: Besides doubling the capacity of a single fiber, use of 200 Gbps line cards can optimize network utilization. Since all 200 Gbps wavelengths are delivered on flexible modulation line cards that can operate at speeds ranging from, say 50 Gbps to 250 Gbps, a service provider can choose the appropriate line speed for the current demand.  Therefore, a service provider can choose to operate a link at a higher utilization by running at 100 Gbps until the need arises to shift the line speed to 200 Gbps in the future.

Needless to say, we believe the future is bright for 200 Gbps WDM, and we think it’ll be a focal point at OFC 2016.

I had the opportunity to meet and interact with key stakeholders in the mobile infrastructure industry last week during the Mobile World Congress show in Barcelona.

Key Takeaways from the show:

  • Doing more with less
  • Opportunities drive high level of innovation
  • In-building/densification puzzle not solved
  • Carriers don’t want to wait five more years for 5G
  • The future is promising

The main themes of this show reflected the state of the mobile industry namely that it is in the midst of an on-going digital transformation, meaning MWC 2016continuously changing and more demanding end-user requirements. At the same time, vendors and carriers are struggling to convert these more demanding end-user trends into increased revenue and vendors have become extremely good at doing more with less. Incremental mobile data consumption has grown nearly 10-fold over the past five years and during the show, the leading radio vendors announced new BBUs that will provide enough processing capacity for the foreseeable future for even the most optimistic data traffic consumption scenarios. At the same time, I did not talk to one service provider that expects infrastructure prices to go up.

Despite flat-ish infrastructure trends, interest is high to capitalize on the on-going shift from outdoor coverage to indoor and urban densification. After all, this is a $30 B-plus market that was initially designed to support voice services and the ever changing end-user requirements are opening up opportunities for new entrants while the incumbents want to defend their turf. The end result is that the level of innovation that we have witnessed in the mobile infrastructure industry has been mind-boggling. More BTS architectures have been introduced over the past couple of years than the entire industry introduced in the first 25 or 30 years rolling out 2G and 3G. And during the show, vendors announced several new architectures and solutions which ultimately will improve the efficiency, flexibility, and scalability of RAN resources. Of particular interest during the show was how far some vendors have come when it comes to moving radio functions towards standard IT hardware.

While small cell deployments have clearly moved from the hype phase to reality (the Dell’Oro Group estimates the non-residential small cell market more than doubled in 2015 approaching $1 B), it is still very early days for small cells and there is yet plenty of on-going innovation both from a technology and business perspective targeting a more compelling TCO.

Of particular interest during the show was the continued commitment to basically make small cells more like WiFi – the combination of neutral host and lower-cost equipment/installation/maintenance are more and more seen as a good framework for accelerating small cell adoption particularly in the enterprise. All of the leading small cell players have now announced plans for deploying LTE (LTE-U/LAA, or MuLTEFire) in the unlicensed 5 GHz bands. Increasingly, vendors find the non-existent or negligible interference from outdoor macro and WiFi to be a lucrative proposition – spurring interest in the 3.5 GHz band, where applicable. It is also becoming increasingly clear that the initial vision of having these integrated small cells deployed with dedicated operator specific radios will limit the pace of adoption – and the inherent benefits of the unlicensed bands will open up opportunities to transfer network ownership from the carrier to the enterprise, building owner or third party. As a result, you could stop by Qualcomm’s booth and see a plethora of 3.5 GHz solutions including prototypes with interest from non-traditional players, including Google.

Although initial small cell vendor share estimates suggest the traditional macro RAN players, including Huawei, Ericsson, and Nokia, are capturing the largest revenue shares in the non-residential small cell segment, new entrants see the reduced technical barriers of entry with a spectrum band that is not interfering with the outdoor macro as an opportunity. At the same time, go-to-market strategies and enterprise channel experience will likely play an important role for the indoor enterprise segment. Ruckus – a WiFi provider – announced it plans to bring 3.5 GHz LTE products to market in 2017.

An important piece to make small cells more like WiFi is cost. I remember having a conversation a couple of years back with a CTO with one of the Tier1s in the US and he told me the small cell radio can be free of charge as he would still rather deploy a macro. This is a reflection of the business case challenges with small cells if one uses similar models and processes as one did with macros; as a result the radio equipment will account for a single-digit share of total TCO, resulting in a rotten business case when comparing to a macro (if 90% of the costs remain constant). Much of the emphasis over the past couple of years has naturally been on working on bringing down the cost of the 90% as opposed to the 10%. But with some progress on the 90% side (though it should be pointed out that there is much work left), carriers and vendors believe the time is ripe to address the cost of the 10% portion (which ideally will be higher than 10% assuming progress on the 90%).

For example, Parallel Wireless introduced an enterprise “white box” small cell solution that they hope will make LTE as easy to deploy as WiFi, partly by bringing down the equipment cost to the <$500 range. Similarly, numbers were thrown around by other players suggesting target prices in the <$100 range for 3.5 GHz and 5 GHz LTE APs.

There were also a wide range of new RF mapping solutions launched at the show aimed at making LTE as easy to deploy as WiFi. In short, lots of progress when it comes to the “WiFi-ification” of small cells (I can’t take credit of the term – saw it in a Ruckus article). The irony is that the 802.11 roadmap is also evolving and who is to say that for some of these use cases, particularly in the not overly crowded enterprise scenario, that the performance gap between WiFi and LTE will not be smaller by the time the LTE or 5G-fication (new term, my invention) will be a reality.

Finally, when it comes to 5G, my main conclusion from the show was that 2020 is still four to five years out – yet carriers and vendors need new revenue streams today. At the heart of this matter is the fact that we are at a similar phase today as we were ten years ago, namely four or five years to the next big thing – LTE then, 5G today. The difference being that ten years ago, 3G deployments were sparse while 4G coverage today is available nationwide in most of the advanced economies and well on the way in many of the developing economies. The challenge here is that carriers are deploying new technologies at a faster pace with each technology generation, yet the standards are not accelerating at the same pace. As a result, pre-5G commitments are growing at a rapid pace, even if the LTE-A roadmap will drive strong upgrade investments and deliver adequate performance for some of the potential revenue streams carriers envision with 5G.

As with any technology shift, it is not always clear initially what will ultimately be the killer use case and there is no difference with 5G. I was impressed by the variety of demonstrations and the imagination used for these potential use cases. Ericsson had brought in folks from the industry to demonstrate potential applications with trucks and automation among other things.

AT&T recently suggested that 5G will likely be driven by IoT, 4K video, and virtual reality. I will probably not be one of the early adopters when it comes to virtual reality. However, after trying several times unsuccessfully to get on to the virtual reality roller coaster ride during the show because of the lines, I was simply amazed by the interest in this “ride”, and this was a bit of a wake-up call – this is not just a toy for kids, this is real.

I spent a good portion of the flight home thinking about my kids and how different their future will be with virtual reality and self-driving cars. What would all of this mean for the mobile infrastructure market? Will my models support this type of data consumption? The only conclusion I came up with at the time was that surely there must still be a lot of work left to ensure a more consistent user experience throughout the network. Fresh in mind was my recent visit to IKEA in Palo Alto over the holidays for a delicious meatball lunch and the difficulties I had communicating with my wife because T-Mobile’s network did not provide enough signal strength to send a good old fashioned text message.

In summary, behind those flattish infrastructure revenue trends, there is an awful lot of opportunities as well. But one step at a time…

I was fortunate enough to attend Mobile World Congress (MWC), a conference held in Barcelona, Spain the week of February 22nd. Like previous years, it was by far the busiest show I have ever attended. In three days, I met with over 20 companies, mainly in the area of Microwave Transmission, which is one of the technology areas I cover.

When it comes to microwave technology, the themes were very similar among the microwave manufacturers (with a few exceptions): increasing microwave transmission capacity, utilizing different spectrum, and reducing time to install.

The technologies to support these efforts included:

  • 4096 QAM
  • V-band (60 GHz), E-band (70/80 GHz), W-band (90 to 110 GHz), and D-band (140 to 175 GHz)
  • Multi-band radio frequency bonding
  • Signal beam-forming
  • Antenna auto-alignment

However, with nearly all the manufacturers moving towards the same grouping of technology, how will these vendors differentiate themselves? How will they break away from an overcrowded pool of 28+ microwave transmission vendors and stand out?

The bad news is that it’s not easy. For many of the vendors, future sales will come down to channels of sales, pricing, and luck.

The good news is that a number of vendors have plans to remain above the herd and differentiate based on either being first to market with leading edge technology, targeting the requirements of a select set of customers by application, developing tools to reduce the time to install a microwave link, or enhancing a customer’s experience with the aid of software applications.

I was fortunate to attend MWC this year: Where else could I of had such rich conversations with over 20 companies in only three days while walking a total of fifteen miles between the booths?