Ten Investible Things that We Think Will Happen in 2017

sagawa
Print Friendly
Share on LinkedIn0Tweet about this on Twitter0Share on Facebook0

SEE LAST PAGE OF THIS REPORT Paul Sagawa / Artur Pylak

FOR IMPORTANT DISCLOSURES 203.901.1633 /.1634

psagawa@ / apylak@ssrllc.com

twitter.jpg
@PaulSagawaSSR

January 2, 2017

Ten Investible Things that We Think Will Happen in 2017

2016 was an odd year in TMT investing. The first 6-7 weeks were brutal for the cloud-stocks we favor, and we spent the next several months digging out. By Nov. 8, our 15-stock model portfolio was up 290 bp vs. the tech elements of the S&P 500, only to see the Trump rally kill 130 bp of relative performance. Still, of our investible things for 2016, we got 7 right against 3 wrong. Solid marks for negative calls on fixed broadband, security, smartphones, and OTAs; and for thumbs up on self-driving cars, big internet players, and high $ deals. Linear TV didn’t quite break (we’ll double down for 2017), the spectrum sale, still in progress, doesn’t look likely to shake anything up, and our top picks were ruined by TWTR’s botched deal making. This year, we are looking for AI-as-a-Service, SaaS consolidation, cloud partnerships, AI-based security, virtual assistants, heavy competition in smartphones, the linear TV inflection point, the end of net neutrality and the rise of robo-taxis to make the news that moves the market. Our top longs? XLNX, CRM and TMUS. The naughty list? EXPE, ORCL and CBS.

  1. AI as a Service becomes a thing – Deep learning AI technology can improve almost any software and enables entirely new computing use cases that could revolutionize swaths of the economy. Many companies now realize its potential, but very few have the prerequisite skills and infrastructure. We believe 3rd party AI hosting from GOOGL, MSFT, AMZN and IBM will take off this year, driving growth and profit for those hosts, but also for the suppliers of specialty hardware needed to enable the AI services – NVDA and XLNX lead the list. (http://ssrllc.com/publication/ai-as-a-service-deep-learning-is-fundamental/)
  2. The SaaS roll up picks up speed – In 2016, MSFT bought LNKD, CRM bought DWRE and ORCL bought N (amongst other SaaS buys). We think that GOOGL and AMZN could join the SaaS M&A party in 2017 to supplement their growing enterprise IaaS businesses. Look for many SaaS names to be in play (WDAY, NOW, DATA, SPLK, SHOP, etc.) and for a few major transactions to be completed. GOOGL, AMZN or MSFT could even up the ante with a play for CRM. (http://ssrllc.com/publication/saas-applications-cloudy-with-a-chance-of-ma/)
  3. Cloud hosts draw unusual partners – Many web-based businesses (AAPL, Snap, SPOT, NFLX, WDAY, PAYP, and others) have signed deals to move some of their workloads from their private datacenters to AMZN, MSFT or GOOGL. We expect bigger deals ahead, as the cost and sophistication advantages of the hyperscale operators grow ever more compelling. Meanwhile, we expect more traditional IT players to cut deals with the big three to sell hosting to their enterprise customers with value added services atop.
  4. New paradigms for data security – Major data security incidents will continue, as the focus turns more clearly to circumventing human error and complicity. Demand for commoditized and increasingly ineffective security appliances continues to weaken. Cloud hosts will begin to offer AI-based security tools to anticipate, identify and counteract incursion more quickly and accurately. High end intrusion detection market remains strong, but AI-based competition looms. Trump administration push for “backdoors” could raise new security risks. (http://ssrllc.com/publication/cybersecurity-trust-no-one/)
  5. Virtual assistants get momentum – AI assistants will get better quickly as usage builds – improved voice recognition, more conversational interactions, greater anticipation of needs, wider ranges of 3rd party services, and more/different access points. GOOGL Assistant will be recognized as the best and will extend its advantages. AAPL and MSFT will launch AMZN Echo/GOOGL Home competitors. The pressure on the app model for the primary device UI will become apparent. (http://ssrllc.com/publication/ai-assistants-the-next-user-interface-paradigm/)
  6. Smartphone competition heats up – Samsung will be back – with edge-to-edge and flexible displays with fingerprint sensors integrated right into the screen. AAPL will follow with big hype for the 10th anniversary iPhone 8, but sales will be unlikely to live up to huge analyst expectations. Meanwhile, GOOGL Assistant will become a big selling point for Android phones, a boon to a growing wave of much cheaper, “good enough” Chinese smartphones in global markets. ASP declines will accelerate as unit growth continues to flatten. (http://ssrllc.com/publication/smartphones-mobile-maturity/)
  7. The linear TV narrative breaks – TV media has run hard over the past 3 months, but the party will end in 2017. Pay TV sub declines will accelerate. Broadcast and cable network ratings will continue to decline. TV ad spending will decline, even more so than might be expected in a non-election, non-Olympic year. This will be good for digital advertising, and its disproportionate beneficiaries, GOOGL and FB, and bad for the whole TV ecosystem. (http://ssrllc.com/publication/video-media-the-cord-cutting-myth-gets-real)
  8. Goodbye, net neutrality – The Trump FCC may not re-designate broadband outside of Title II, but it need not write or enforce any regulations circumscribing “zero rating”, “fast lanes”, content connectivity fees, or any other manifestation of carrier oligopoly power. We think most carriers will step lightly, at least at first, and focus on practicies that “seem” consumer friendly, like zero rating. We also expect carriers to test the Trump DoJ with M&A – S and TMUS could resume their dance, VZ could get more aggressive shopping for content, and CMCSA could look to roll up more internet subs.
  9. Autonomous taxis on the road – Uber’s self-driving cars have been carrying commercial passengers in Pittsburgh, albeit with a driver on standby. Look for GOOGL’s Waymo to join the party soon. We expect completely driverless trials to begin in Asia by the end of 2017, with clear movement toward similar trials in the US thereafter. With proof of concept in hand, momentum toward robo-taxi fleet operations will be clear, with GOOGL, Uber and BIDU the leaders. (http://ssrllc.com/publication/autonomous-cars-self-driving-ambition/)
  10. Our best picks for the year – Longs: XLNX, CRM, and TMUS. FPGA vendor XLNX will see AI-related datacenter demand, and could be an M&A target. CRM leads a SaaS group that underperformed in 2016, but is poised for upside in 2017. It too could get M&A attention. TMUS benefits from the Trump FCC’s expected laissez faire and Softbank’s likely interest. Shorts: EXPE, ORCL and CBS. OTAs started to get pressure last year, but the rise of virtual assistants will be a further threat to EXPE and others. ORCL is very poorly positioned for the cloud era and has serious customer relationship woes. CBS has been a “Yuge” beneficiary of the Trump rally, but a reality TV president won’t deliver an audience or ad growth.

How did we do Last Year? Out of 10 predictions, 7 were spot on. We said broadband would show cracks – subs were down, with telco results particularly weak. We called out security companies – big names FEYE, PANW and IMPV were off more than 25%, and most others underperformed. We predicted poor device sales – AAPL and Samsung led the disappointments. We were critical of on-line travel – EXPE, TRIP and SABR all badly underperformed, with only PCLN beating the index. We were right about self-driving cars becoming a big deal and our favor for the big guns of the internet proved correct as well. We thought we’d finally see some big M&A – T (TWX), QCOM (NXP), Softbank (ARMH), MSFT (LNKD), and others cooperated. Our 3 duds: We had high hopes that the FCC 600MHz auction would shake up the wireless market, but no dice. We said linear TV would suffer – subs, ratings and ad sales disappointed but the stocks were up more than 25% YTD. Finally, of our 3 best longs, only MSFT beat the benchmark. GOOGL slightly underperformed, while TWTR’s botched deal making killed it for 2016.

Exh 1: The 2016 Predictions Scorecard

Another Auld Lang Syne

It’s prediction time again. 2016 was a pretty good year for our crystal ball, although the surprise election of Donald Trump killed the strong momentum of our 15-stock model portfolio, turning strong pre-November 8 outperformance to a mild full year disappointment. (N.B. TWTR’s bungled deal making didn’t help either) We hope to top our 7 for 10 prognostication performance with this year’s calls.

Not surprisingly, AI plays a key role in most of our 10 predictions for 2017. 1. AI-as-a-Service platforms at GOOGL, MSFT, AMZN and, possibly, IBM will successfully enable less capable 3rd parties to develop deep learning systems. 2. The SaaS market will continue to consolidate, with MSFT, CRM and perhaps GOOGL and AMZN building critical mass and levering their AI chops to transform the enterprise software paradigm. 3. The top IaaS hosts will draw big name partners as customers and as distributors/integrators. 4. Data security will change drastically – the status quo solution of technology fences and intrusion alarms isn’t working. We think the answer is cloud-based AI that can anticipate threats and adapt quickly, along with MUCH better employee policy enforcement. 5. AI virtual assistants will prove to be a very big deal, getting much better on most dimensions and threatening the app UI model. We think GOOGL has the edge over AMZN, AAPL and dark horse MSFT. 6. The smartphone market will get very competitive – A reinvigorated Samsung and low cost Chinese interlopers will give the iPhone 8 a tough time. 7. Linear TV won’t have a wild election season to save them this year – pay TV subs, broadcast/cable ratings and TV ad sales will all disappoint. 8. The Trump FCC will allow zero-rating, fast lanes and other net neutrality no-nos, but the carriers will focus on consumer popular programs and push for M&A. 9. Commercial trials of autonomous robo-taxis will begin in 2017 and the hype will build rather than diminish.

Our 10th prediction is our top picks for 2017. While we are confident in all 15 stocks in our model portfolio, we are picking XLNX, CRM and TMUS as our highest conviction names for the full year. XLNX is a specialist in FPGAs, programmable chips that are becoming popular for AI – both in datacenters and in devices. XLNX already has deals with BIDU and IBM for its AI optimized products. CRM is an obvious leader in SaaS consolidation and a growing presence in AI for enterprise applications. The trends play to its strengths even as its stock has been a bit out of favor. We also believe that GOOGL, AMZN and/or MSFT could make a play to acquire it before year end. Finally, TMUS has been harvesting wireless market share from VZ, T and S. We expect this to continue under Trump laissez faire. Talk about an acquisition by Softbank will also buttress the stock. We also considered doubling down on last year’s picks GOOGL and MSFT, and believe that Citron’s take down of NVDA is excessively short sighted.

We are also picking shorts. ORCL is a repeat selection – we believe it is far behind MSFT and CRM as the enterprise market turns to the cloud, and it is further burdened by an unhappy customer base. EXPE experienced trouble in 2016 – we think the AI assistants will add to the pressure of travel service providers and AirBnB. Last, we see TV pureplay CBS as particularly vulnerable to disappointment amongst falling ratings and weak ad spending. Other stocks that we thought about on the short side – T, HPE and AAPL.

We Didn’t Expect THAT

Our 2016 predictions piece got a lot of things right. We said that the residential broadband market would show cracks – industry reports suggested the first YoY decline in US broadband subs for 2Q16, with telcos taking the brunt and then some. We said the cloud would begin to eat security – overall, security stocks underperformed badly in 2016, with the high-profile DNC breech highlighting the futility of even sophisticated tools to stop user stupidity. We said devices would disappoint – bingo! iPhone unit sales were down -8.3% and Samsung literally went up in flames. The on-line travel reality check came as predicted, at least for EXPE, TRIP, and SABR, all of which melted down on weak results. Self-driving cars were indeed the talk of the industry, with Uber, GOOGL, TSLA, MEYE, NVDA, and others making big moves. We said the big guys would continue their winning streak – AMZN, MSFT, NFLX, GOOGL, and FB all smashed analyst expectations for growth, although GOOGL and FB were not rewarded for it. We also got the M&A we were looking for – T/TWX, QCOM/NXP, Softbank/ARM, MSFT/LNKD, ORCL/N, CRM/DWRE, and numerous other deals made bankers happy.

On the flip side, media stocks soared despite disappointing pay TV subs, linear TV ratings and ad revenues – chalk one up to Mr. Trump. We had also called for a disruptive spectrum auction, but GOOGL and other possible interlopers stayed on the sidelines. Finally, our model portfolio performance had dug itself out of a deep January hole, only to see its gains wiped away by the Trump rally. Of our 3 top longs, only MSFT outperformed, GOOGL shares fell off the pace after election day, and TWTR self-destructed (again) with help from its bungling bankers. Our shorts were better – ORCL, JNPR and IPG all missed the benchmark – but not so much that we would brag about it.

On to 2017 …

1 – AI as a Service Becomes a Thing

In 2016, we wrote extensively on the rapid coming-of-age for deep learning based artificial intelligence (http://ssrllc.com/publication/ai-as-a-service-deep-learning-is-fundamental/). We noted that there were three main ingredients to successful deep learning system development – talent, data and AI-tuned data processing capacity – but that all three were in relatively short supply, with a small number of technology leaders holding powerful assets (Exhibit 2-3). Some of those AI leaders – specifically, Alphabet, Microsoft, Amazon and IBM – have looked to build their AI capabilities into platforms that less capable organizations might use to infuse deep learning technology into their software.

Alphabet’s Google Cloud Platform took the lead here – it has published its Tensorflow AI programming rubric to an enthusiastic open source community, developed a custom ASIC processor optimized for its rubric (Tensor Processing Unit or TPU), deployed the TPU widely into its hyperscale datacenter infrastructure, and launched a relatively comprehensive AI-as-a-Service offering, including support for Tensorflow and other popular rubrics, access to proprietary training datasets, and APIs to pre-trained AI modules, such as language translation and image recognition. Microsoft has a similar platform, with its own AI programming rubric (CNTK), optimized infrastructure based on NVDA GPUs and Altera FPGAs, and pre-trained APIs. IBM supports its rival’s AI development tools on its Watson Cloud. It has partnered with NVDA on AI-optimized processing, and offers the most extensive set of AI APIs in the market. Amazon was a bit later to the AIaaS platform idea, but is investing heavily. It supports Tensorflow, CNTK and other open source rubrics on AWS, has NVDA-based GPU capacity available, and recently hired noted deep learning scientist Alex Smola from Carnegie Mellon University with a mandate to build an industry best AI-optimized hardware and software platform for AWS (Exhibit 4).

Exh 2: Requirements for AI and Neural Networks

Exh 3: AI Citation Summary – Tech Companies

Exh 4: Deep Learning Services Offered by the Big 4

We believe these companies will push AI forward very quickly and that AI-as-a-Service will be a big story in 2017. Initial revenues will be small, but given the huge data sets and processing demands of deep learning training and the market potential for hosted web-based services incorporating AI, we expect the buzz to be considerable. Look for all four – Google Cloud Platform, Microsoft Azure, Amazon Web Services and IBM Watson – to sign flashy AI development deals with notable partners this year.

2 – The SaaS roll up picks up speed

In 2016, Microsoft bought LinkedIn for its rapidly growing HR SaaS application and the data being generated by its user base of 400M professionals. Salesforce bought Demandware to be the backbone of its Marketing Cloud SaaS platform. Oracle agreed to buy SaaS ERP company Netsuite, along with a handful of other small SaaS operators (Exhibit 5). Private equity also stepped to the plate, paying similar multiples as the strategic buyers for several SaaS companies. We wrote about this phenomenon this past summer (http://ssrllc.com/publication/saas-applications-cloudy-with-a-chance-of-ma/)

We believe that consolidation of the SaaS software space is inevitable. Economies of scale and long investments in the technologies behind hyperscale datacenters and their operations have given Amazon, Microsoft and Google huge cost and performance advantages in cloud hosting. Since SaaS products are a bundle of application software and the infrastructure hosting it, independent SaaS companies with their own datacenters will, inevitably, see cost disadvantages in the infrastructure half of their bundle. This will drive them to look at hosting relationships with the big three that could naturally lead to closer combination. Moreover, there are considerable synergies to multi-application platforms – from bundled selling to common user interfaces and support systems.

Exh 5: SaaS Acquisitions by Major Enterprise Software Companies

AI will increase the advantages of scale in SaaS applications. Data captured across a broad suite of apps and interpreted by a common AI can drive better productivity, unearth underutilized assets, improve information flows, and make clear the workings of an organization. Common AI interfaces can give employees new tools to find needed information, to execute complex tasks, and to interact in a more human way. This is a further incentive for AI-capable organizations like Microsoft and Salesforce to aggressively roll up smaller best-of-breed SaaS applications. Companies like Workday, Service Now, Shopify, Tableau, Splunk, and others will be getting calls from bankers.

We also believe that Google and Amazon could enter the pool of possible acquirers. Both would see substantial benefits from a meaningful SaaS application position. Neither company has a strong sales presence beyond the largest tier of potential enterprise accounts for their IaaS hosting platforms. Indeed, these companies have the resources to bid for the biggest fish in the SaaS pond, and we would not be shocked to hear rumors of interest in Salesforce to rise again in 2017. Such a deal would instantly make either Google or Amazon into a major enterprise software player, and it is likely that Microsoft would make its own play to keep the two consumer Internet behemoths out of the application business.

3 – Cloud hosts draw unusual partners

Even without a big splash Salesforce deal, the big cloud hosts will get much more active in the enterprise space. The advantages of the hyperscale datacenter operations at Amazon Web Services, Microsoft Azure and Google Cloud Platform are so substantial that web based businesses of all kinds – even some that are tangential competitors – will be thinking seriously of moving to hosted operations (Exhibit 6). Apple famously relies on all three of the top clouds for its iCloud and iTunes products. Amazon is the platform of choice for many businesses, including Netflix. Microsoft has Expedia, Ebay and others. Google has hosted Snapchat from its start and recently signed Spotify. We think that more cloud application players, including SaaS companies, will announce hosting deals in 2017.

Exh 6: Basic On-Premise versus Cloud Cost Comparison

We also think that traditional IT companies will be forced to cozy up to the big three hosts as well. IBM has its own cloud operations, but others, like Cisco, Hewlett Packard Enterprises, and Dell/EMC, could expand upon the relatively weak reselling relationships they have with cloud partners to create much more integrated offerings. Microsoft, with longstanding existing relationships with the hardware vendors who have bundled its software into servers, has taken the lead in cultivating these partnerships, but Google and Amazon, without their own well-entrenched enterprise salesforces may have greater urgency. Look for the traditional vendors to commit to exclusive sales relationships with specific cloud partners and to emphasize new products that integrate on-premises datacenters with cloud services much more closely.

4 – New paradigms for data security

2016 was another bad year for data security. In February, the US Department of Homeland Security and the FBI were both hacked, with the personal details of 30,000 employees released to the web. In the same month, the IRS revealed that more the financial records of more than 700,000 taxpayers had been stolen in a breach the previous year. In March, millions of patient medical records were stolen in a pair of incidents, and 1.5 million customer records taken from Verizon Enterprise Systems showed up for sale on the dark web. In April, the personal information of every voter in the Philippines was compromised. In May, a US security company found more than 270 million email names with passwords available for free from Russian cybercriminals. June saw the highest profile breach, when the Democratic National Committee servers were hacked and politically damaging emails exposed through Wikileaks. In August, malware in Oracle’s MICROS point of sale system was discovered that may have exposed the details of many millions of retail transactions, including customer credit card numbers. In September, Yahoo announced that access information for more than 500 million email accounts, including full names, passwords, birth dates, phone numbers, and the answers to security questions, had been stolen, likely by a foreign government operation. In December, it announced a separate breach in which more than a BILLION accounts had been compromised. In November, the X-rated site Adult Friend Finder had 412 million customer accounts hacked and the information put up for sale.

These breaches are happening despite years of spending for security technology, with human error the usual point of vulnerability – Hillary Clinton Campaign Chair John Podesta’s unwitting click on a spearphishing e-mail is a vivid example of this sort of incursion (Exhibit 7). We noted in 2015 that enterprise security investment was better spent on improving user training and enforcing policy compliance than on fancy new cyberweaponry, and it would seem, based on the disappointing performance of many security technology providers, that organizations are beginning to follow that blueprint.

However, we do think new help may be on the way in the form of AI-based security agents. These systems are increasingly used by cloud operators to identify suspicious activity early, to anticipate the moves of hackers, and to identify and shut down breaches as quickly as possible. These agents can adapt to changing cybercrime strategies far more deftly than human experts. These solutions could make moving sensitive data to the public cloud a security imperative rather than a security risk, and prompt faster migration by enterprise organizations. This is an obvious boon to Amazon Web Services, Microsoft Azure and Google Cloud Platform. We also expect that we will see high profile startups with independent AI-based security solutions that could displace current solutions at the high end of the cybersecurity market.

Exh 7: Cybersecurity Incident Classification Patterns, 2015

5 – Virtual assistants get momentum

Amazon won’t say exactly how many Echo speakers, featuring the voice-powered Alexa virtual assistant, it sold in 2016, allowing only that holiday sales of the product were up nine-fold vs. 2015 and that it was the company’s biggest electronics product during the season. Given an estimated installed base of about 2 million Echoes at the beginning of the year, it stands to reason that the population is well into the double-digit millions today. Google’s analogous product, the Google Home, began shipping at the beginning of November and quickly sold out. It features the Google Assistant voice-powered AI agent that was also integrated into the company’s popular Pixel smartphones, which also sold out for the holiday season. Apple’s Siri has been with us much longer, shipping on every iPhone sold for the past 4 years, and Microsoft’s Cortana assistant lurks on the 400 million PCs running Windows 10 and the 30-40 million Xbox One gaming consoles in its installed base.

Critical mass is an important step on the way toward much better virtual assistant functionality. These installed base devices are collecting data on queries and commands that will allow ever better understanding, improved execution and more accurate anticipation of user needs. Users will grow more accustomed to the capabilities of these systems and begin to rely upon them more often and for more reasons. 3rd parties will attach their services, and greatly expand the universe of things that the assistants can do. The improving functionality and growing word-of-mouth will help to drive even greater adoption and use, and the self-reinforcing virtuous cycle will gain more momentum. We wrote about this a few months ago (http://ssrllc.com/publication/ai-assistants-the-next-user-interface-paradigm/) and believe the pace of improvement and adoption will surprise investors in 2017.

Exh 8: Four Roles for AI in Virtual Assistants

Exh 9: Roles of Deep Learning Positioning Among Major Virtual Assistants

As for winners, Apple had a 3-year head start with Siri, but its relatively weak AI capabilities and aversion to collecting customer data has kept progress slow (Exhibits 8-9). Perhaps the new scientist-friendly HR policies which have helped it recruit highly regarded new talent to its ranks can reinvigorate its relatively moribund AI virtual assistant efforts. If not, we believe it could be a serious competitive liability for future versions of the iPhone. Amazon’s innovative Echo established the AI assistant powered home speaker category, and the company’s e-commerce dominance gives it a formidable distribution advantage over the similar products that are starting to come to market. The company has the AI chops to make strides, but does lack the ability to expand Alexa’s presence onto a smartphone or into a car. Google’s Assistant can lever the world’s best AI talent roster, 18 years of work on Search, and a potential base of more than a billion Android users. Less than 6 months into its life, Google Assistant is already beating Siri and Alexa on its language fluency and figures to get much better very quickly as it learns from use. Microsoft’s Cortana is a dark horse – the company certainly has the scientific bona fides in AI to be a real player. We believe the company will focus on its agent as a front end for its productivity software as it aims to use AI to revolutionize enterprise computing.

6 – Smartphone competition heats up

As we predicted a year ago, 2016 was a rough year for the device market. The iPhone 6S couldn’t live up to the huge sales of the prior year’s model, which introduced big displays to the Apple installed base and inspired many millions of users to upgrade early (Exhibit 10). The death of the carrier subsidy model in the US market further depressed upgrade activity, as did the mundane hardware improvements introduced by the various smartphone vendors. Consumers seem comfortable with the resolution of their displays, the speed of their processors and the quality of their cameras. The malaise was exacerbated by the high profile overheating problems that led Samsung to kill the just introduced Galaxy Note 7 flagship product entering the all-important holiday season. Overall, the pace of smartphone unit growth slowed, the premium segment declined, and the companies tied to the market suffered.

Many analysts have high hopes for 2017. Apple will introduce its 10th anniversary iPhone, likely named the iPhone 8, in September, and most anticipate a much more fundamental upgrade than the minor tweaks introduced with the 7 this past fall. The logic goes that a splashy new model, perhaps with an edge-to-edge display and a virtual home button, will spur another super-cycle of upgrades akin to 2014’s iPhone 6 bonanza. We are not so confident.

Competition will be much fiercer in 2017. We expect Samsung, stinging from the Note 7 debacle, to beat Apple to the punch with an edge-to-edge display for its Galaxy 8, due in the spring. We expect Chinese brands – Huawei, Levono, Xiaomi, and others – to aggressively target global markets with excellent smartphones at very significant discounts to the currently popular brands. We expect the trend toward longer replacement cycles to continue as customers on financing plans experience the benefits of lower monthly bills and as technical improvements to phone specs see decreasing returns. On a global basis, we expect modest smartphone unit growth but growing price pressures to yield flattish sales (Exhibit 11).

This is bad news for the Apple narrative. Even if the iPhone 8 can spur a “super-cycle” it will just borrow upgrade demand forward and set investors up for deep disappointment in the next years. Meanwhile, price pressures will threaten those rich margins and cash flows inherent to the DCF models that show the stock as cheap. Moreover, if we are right about Google Assistant, it could become a serious competitive advantage for Android vs. Apple’s iOS and Siri, another crack in the brand armor. If Apple is not a short now, we believe it will be once the iPhone 8 has been introduced.

Exh 10: Quarterly iPhone Revenue and Unit Sales, 1FQ14 – 4FQ16

Exh 11: SSR Global Smartphone Forecast, 2015-19

7 – The linear TV narrative breaks

The linear TV business disappointed in 2016. Pay TV subscribers were down in general, with many top networks feeling a further hit from the growth of “skinny bundles” that do not include all the channels typically offered in a cable subscription. TV ratings for both broadcast and cable networks were disappointing, with Olympic and election coverage ratings concealing sharp declines in viewership for regular programming. The NFL, long the tent pole content for linear TV, showed shockingly weak ratings for the 2016 season. Overall, US TV ad sales were up 3.5% for the year, but this is a modest disappointment relative to initial forecasts for the typically strong quadrennial year.

Exh 12: Media Stock Performance, 2016

Despite all of this, media stocks appreciated an average of 13.7% during 2016. A big piece of this played out in the Trump rally, when TV media outperformed the rest of tech by 780 bp (Exhibit 12). The narrative seems to underplay the deterioration of the linear TV model for advertising and subscriber fee revenue, while overstating the opportunity for traditional media players to exploit their content brands in the growing streaming market. Magma, the market analysis arm of the IPG advertising conglomerate, predicts US TV ad sales to decline 0.9% in 2017, a figure that we believe is optimistic given trends in viewership. The amount of video content being produced continues to rise, with the high-quality programming available on streaming-only services like Netflix and Amazon Prime widely recognized by consumers. Traditional media has tended to overprice its content in the streaming market – compare the value of HBO Now at $15/mo or even CBS All Access (no ads) at $10/mo to Netflix at $10/mo or Amazon Prime Video (free with a $99/yr membership). We are not sure that the take rate for the newcomers amongst cord cutters will be as strong as some project. We are concerned that tough news flow and unrealistic expectations for their TV businesses set stocks like CBS, Fox, Comcast, and Disney up for disappointment in 2017.

8 – Goodbye, net neutrality

In 2015, after years of misfires, the FCC reclassified broadband providers as “common carriers” under Title II of the Telecommunications Act of 1996, allow it to more broadly regulate the services that they provide. Effectively, this reclassification allowed the FCC to block practices such as “fast lanes” which provide much better performance for content provided by the carrier or its paid partners relative to that provided by independent parties, or “zero rating” in which carriers set aside certain content as exempt from usage limitations. The FCC held that such practices gave broadband providers the ability to demand tolls from internet based businesses which, given the uncompetitive market for broadband services in many parts of the country, would have no choice but to pay. Critics, considering net neutrality as government interference in free markets, saw the tolls as fair compensation for the investment of building telecommunications networks.

Under a Trump appointed FCC Chairman, expect net neutrality to be gutted. The process for reversing the reclassification is arduous and might not be completed by the end of a first Trump term, but the commission can easily change the rules and thresholds for enforcement such that carriers would be free to establish fast lanes and zero rating programs at their pleasure. Already, AT&T has announced wireline and wireless zero rating for its DirecTV streaming TV service, confident that the neutered FCC will pose no obstacle. The practice will clearly favor strong content/distribution partnerships and will promote more M&A akin to the AT&T/Time Warner and Comcast/NBC-Universal marriages.

Major streaming content providers, like Netflix, Amazon, Google, and Facebook, will cut deals to be included in “fast lanes” and zero rating bundles. We believe that carriers will be careful not to price these deals too aggressively, at least initially, as the telecom and cable industries remain very unpopular with voters who could easily inspire congressional action as a backlash if public sentiment turns especially sour. As such, we believe the benefit for the carriers will be apparent but still modest and that the cost for major internet players will be small.

9 – Autonomous taxis on the road

Consumer awareness of self-driving technology went from zero to sixty in 2016, with Tesla’s autopilot (and the crash of a particularly inattentive Florida man), Uber’s Pittsburgh experiment, and Google’s ongoing trials making headlines outside of geeky tech publications (Exhibit 13). We have written extensively on self-driving cars (http://ssrllc.com/publication/autonomous-cars-self-driving-ambition/) and believe that progress toward fully autonomous fleets of robo-taxis offering transportation-as-a-service subscriptions is accelerating. A big step is the launch of limited commercial trials. Uber has started picking up passengers in Pittsburgh, albeit at no charge and with an engineer in the front seat ready to take control. Alphabet’s Waymo, having taken delivery of 100 Chrysler mini-vans equipped to its specifications, looks ready to offer its own livery service this year. We expect Baidu to have a service in China this year as well, and may have the inside track to approval for testing operations without redundant human drivers on standby.

We think this will raise the profile of autonomous driving even further, perhaps driving new excitement around Alphabet stock as the magnitude of its opportunity becomes more apparent. Self-driving components will also get attention – good for Mobileye, Nvidia, Delphi and others. We are a bit concerned that Tesla’s claims around self-driving technology may be overstated, particularly given its public dispute with Mobileye and its apparent lack of AI scientific experience (Exhibit 14).

Exh 13: Autonomous Car Release Target Dates

Exh 14: Two Roads to Level 5 Autonomy

10 – Our best picks for the year

We maintain a 15-stock model portfolio which we update on a quarterly basis. Of these 15 names, we are choosing Xilinx, Salesforce and T-Mobile as our three highest conviction picks for full year 2017 performance (Exhibit 15-16). Xilinx is the second largest vendor of field programmable gate arrays (FPGAs), processing chips that can be programmed and reprogrammed on the fly. This flexibility, and their surprising power efficiency, have made them very popular for AI applications. Xilinx already has strategic partnerships with Baidu and IBM to supply FPGAs for their data centers, while Apple’s inclusion of an FPGA supplied by Lattice Semiconductor in the iPhone 7 suggests the even broader opportunity possible in AI-enabled devices. The company’s biggest rival, Altera, was bought out by Intel for $16.7B in 2015 leaving Xilinx as an enticing acquisition possibility as well. Estimates for 5% topline growth in FY17 with scant earnings leverage should be easily managed.

Exh 15: SSR Large Cap TMT Model Portfolio Constituents and Performance, 2016

Salesforce is at the center of the consolidation of the SaaS market, dominating in the customer relationship management element of ERP and using its growing strength to branch out. It has recognized the opportunity presented by AI and has invested heavily to raise its game, giving it further advantage over its most traditional rivals SAP and Oracle. While Salesforce will certainly be an acquirer in both SaaS and AI this year, we also see a strong possibility that Alphabet, Amazon or both could make a play to acquire it, possibly bringing Microsoft into the bidding as well. The stock was a drastic underperformer for 2016, but we believe that it could break out in 2017.

Finally, T-Mobile is the right stock at the right time. We profiled its strategy of harvesting market share from the carriers above it – Verizon and AT&T in 2015 (http://ssrllc.com/publication/tmus-wireless-aint-what-it-used-to-be/) – and we expect it to be even more effective under more a laxer Trump FCC. While the duopolists hustle to add content and justify the high prices needed to maintain their 5% dividend yields, T-Mobile will enjoy a more substantial umbrella to steal more price sensitive customers. We also believe that the Trump administration will be far more hospitable to a potential Softbank deal to combine T-Mobile and Sprint.

Exh 15: SSR Large Cap TMT Model Portfolio, January 2017

On the short side, we are emphasizing companies that we believe are on the wrong side of the emerging AI/Cloud era. For many years, on-line travel agencies have had it good – the travel service provider market was a fragmented mess, while expensive traditional travel agents were ripe for disintermediation. We think the travel market will evolve again as super-intelligent AI virtual assistants render the aggregation capabilities of providers like Expedia irrelevant and threaten their outsized margins. Expedia stumbled a bit in 2016 and we think there could be more trouble in 2017.

Oracle is a return visitor to our short list. Its core business is shrinking and its growth businesses are poorly positioned against SaaS rivals, particularly given the potential of AI to revolutionize enterprise software. Perhaps most importantly, its customers are broadly dissatisfied. Oracle is on a long streak of quarterly disappointments and we think expectations for 3% growth and leverage to earnings are extremely optimistic.

Finally, CBS is up 25% in four months despite horrific NFL ratings and on-again, off-again merger rumors about Viacom. Analysts expect 1% topline growth from the company, but we think that may be wishful thinking, given ongoing ratings erosion and the prospect of advertiser blowback at the May Upfronts. We think the recent action is unsustainable and look for a substantial correction in the media sector in general and in CBS particularly.

Exh 16: SSR’s 2017 TMT Predictions

Print Friendly