The TMT Model Portfolio: 15 Well Positioned Stocks
SEE LAST PAGE OF THIS REPORT Paul Sagawa / Artur Pylak
FOR IMPORTANT DISCLOSURES 203.901.1633 /.1634
psagawa@ / firstname.lastname@example.org
April 11, 2017
The TMT Model Portfolio: 15 Well Positioned Stocks
Our TMT model portfolio comprises 15 large cap stocks that reflect the 5 key investment themes supported by our research. 1. AI/Cloud era leaders. We believe that the scarcity of scientific talent, the importance of strategic datasets, and advantages of AI-tuned hyperscale infrastructure will concentrate the value created by the paradigm shift toward a small set of clear winners, led by GOOGL, MSFT, FB, AMZN and new portfolio constituent IBM. 2. AI/Cloud arms dealers. The shift to cloud-based, AI-infused software will drive significant investment growth in hyperscale datacenters, at the expense of traditional private facilities. We include NVDA, XLNX, MLNX, and QCOM in the model portfolio. 3. SaaS consolidation. The enterprise migration to SaaS applications has momentum – SaaS leaders MSFT and CRM plan to lever AI and scale economies to enhance their products and drive synergies in acquisitions and partnerships. Best-of-breed players, like ADBE, WDAY and NOW, have runway to grow and could be M&A targets in consolidation. 4. Digital media dominance. Linear TV and other classic media are losing consumer engagement to digital media providers like NFLX, GOOGL, FB and AMZN, with subscription and ad spending certain to follow. 5. Wireless competition. 5G will open new applications for wireless, even as aggressive competition drives down pricing and removes consumer pain points. We see TMUS and QCOM as clear winners. Updated quarterly, the portfolio has outperformed the S&P500 by 1112 bp since inception in 2011.
- The next era of TMT will be dominated by AI and the cloud. Our research shows dramatic cost and performance advantages for the leading hyperscale datacenter operators. (http://ssrllc.com/publication/37440/) With time, we believe most computing and storage will migrate to the public cloud, concentrating in just a handful of players. These same companies are dominant in AI, a software technology with the potential to make most traditional applications obsolete and open entirely new commercial opportunities. (http://ssrllc.com/publication/ai-as-a-service-deep-learning-is-fundamental/) This cloud and AI leadership is the primary impetus for including GOOGL, MSFT, FB, AMZN and now IBM in our model portfolio. Of course, these companies also have specific strengths recommending their inclusion – a more detailed rationale for each will follow in the body of this note.
- The AI cloud era will see new hardware winners. The top hyperscale datacenter operators do not buy configured systems, leaving most traditional IT leaders behind. While today’s clouds are overwhelmingly built on x86 architecture, investment in alternative processors, particularly to support AI, is accelerating. We have written in depth on this phenomenon (http://ssrllc.com/publication/37440/). We believe estimates for semiconductor suppliers, like NVDA, XLNX, MLNX, and to a lesser extent QCOM (all in our portfolio), significantly underestimate the impact of the growing cloud datacenter market.
- The growing SaaS market will consolidate. Cloud-based SaaS applications have many cost and performance advantages for enterprises, sparking strong growth for the top players. Still, we see substantial economies of scale, particularly in the infrastructure end of the SaaS bundle (http://ssrllc.com/publication/37440/). AI increases the benefits of consolidation, particularly for companies, like MSFT and CRM, with a clear strategy for applying AI analytics and user interface technologies across their portfolio of applications to increase productivity for their enterprise customers. We expect MSFT and CRM to lead an ongoing roll up of best-of-breed SaaS application players, which include portfolio constituents WDAY, ADBE and NOW. Indeed, CRM itself could be an attractive acquisition candidate for cloud behemoths AMZN and GOOGL, both of which lack the reach into enterprise IT organizations needed for their ambitions.
- Digital media driving TV to an inflection point. US linear TV viewership peaked in 2011, yet ad load and CPMs have continued to grow, leaving execs and analysts complacent about the advance of streaming video. Demographic trends are starkly against linear TV, and we expect a downward turn in ad revenue to hit hard and soon (http://ssrllc.com/publication/advertising-the-golden-age-of-tv-enters-its-golden-years/). On the flip side, purveyors of streaming video – both advertising and subscription driven – are in position to prosper. We have included likely winners GOOGL, FB, NFLX and AMZN in our model portfolio.
- 5G is coming. We have written of the long-term threat of 5G wireless for residential broadband operators (http://ssrllc.com/publication/us-wireless-carriers-the-barbarians-are-at-the-gate/). That threat is nearly upon us, as technology suppliers and carriers are testing pre-standard 5G equipment able to deliver gigabit speed wireless connections reliably and cheaply. To that end, we see component suppliers like QCOM as major beneficiaries. We also note that the wireless carrier market has also been rocked by increasingly aggressive competition, playing to the favor of the primary challenger, TMUS.
- Model portfolio still stumbling post-Trump. At our last update, in December, we noted that our model portfolio had reversed its sizeable gains from before the election to close 150bp ahead of the tech components of the S&P500 for the period. That weakness continued into 1Q17, despite generally strong earnings from our 15 stocks, and we missed our benchmark by 140bp. The underperformance can be tied to two specific stocks, TWTR (down 21.2% since December 12) and QCOM (-15.4%), by lackluster performance by GOOGL, NVDA and XLNX, and by our lack of exposure to AAPL (+30%). Several other picks – notably ADBE, CRM and MLNX, all up more than 20% – nearly offset those stragglers.
- Dropping TWTR, adding IBM. We have held TWTR in our model portfolio from shortly after its IPO. While we still believe in the inherent value of its unique franchise, we have lost patience with the current management and are no longer confident in its ability to execute a return to strong sales growth. Now, the primary hope for investors is an acquisition, which may be finally forthcoming, but we are tired of waiting. We are replacing TWTR with IBM, about which we recently published two major pieces (http://ssrllc.com/publication/ibm-are-we-there-yet). We believe that IBM’s AI capabilities, personified as Watson, are strong and that its strategy of focusing its AI at vertical industrial opportunities is sound. We believe Watson can drive ongoing growth in IBM’s cloud businesses, enough to offset the declining legacy portfolio and deliver sustainable growth for 2018 and beyond. We believe this scenario would change the investment narrative and lead to a rerating of the stock.
Exhibit 1: The SSR TMT Large Cap Model Portfolio – BEFORE Reconstitution
Our Major Research Themes
Our work over the past two years has been concentrated in 5 specific areas (Exhibit 2):
Artificial Intelligence/Cloud Era Leaders – TMT will be increasingly dominated by hyperscale cloud computing platforms and AI-infused software. Hyperscale datacenters offer scale economies, superior resource utilization, flexible scaling, and performance advantages over traditional datacenter architectures and sub-scale private facilities. The largest and most experienced operators have tremendous cost and performance advantages that we believe will drive a steady migration of computing onto a short list of operators. AI is a tool that can make nearly any software better – allowing optimized solutions from ambiguous data. Systems will be able to understand natural world inputs (e.g. sound, images, motion, etc.), interpret human intentions, draw accurate predictive inferences, and adapt flexibly to change. We believe that AI-infused systems will address many trillions of dollars in the global economy, yet the necessary expertise, data and computing resources are concentrated in the hands of a relatively small number of organizations.
AI/Cloud Arms Dealers – The shift to hyperscale computing and AI has clear implications for hardware manufacturers. The leading hyperscale operators do not buy configured systems, instead designing their own server blades and switches, buying components directly and contracting with manufacturing firms to build them. The beneficiaries of this trend are component makers offering parts with differentiated performance. For example, the growth in AI-infused applications is driving demand for AI-tuned GPUs and FPGAs, markets dominated by a limited set of suppliers. Similarly, the massive capacity of these connected datacenters places a premium on networking chips at the leading edge of performance. The central computing and storage tasks are still overwhelmingly performed by x86 architecture CPUs and traditional disk storage, but we see long-term opportunity for ARM-based CPUs and solid state storage to become a bigger share of the mix.
Exhibit 2: SSR’s 5 Key Investment Themes for TMT in 2017
SaaS Consolidation – SaaS applications have many important benefits for enterprises vs. packaged or custom applications running on their own datacenters, driving strong and consistent growth for the top players. However, we note that a SaaS application is a bundle of the application software AND the infrastructure on which it runs. Thus, for SaaS companies operating their own datacenters, the same scale economies driving the extreme concentration of the hosting business apply. In addition, in the AI era, we see substantial synergies from offering a broad range of applications on the same platform – e.g. better sales leverage, common intuitive user interfaces, management tools that can assess productivity based on application usage, etc. With this, we expect consolidation, with larger SaaS companies shopping for best-of-breed apps to add to their portfolios, and, perhaps, IaaS platform players looking to jump into the SaaS game to better leverage their operating advantages and reach enterprise customers.
Digital Media Dominance – Linear TV viewership peaked more than 5 years ago, yet TV ad sales have grown on higher ad prices and increasing ad load, while networks have won ever higher fees from PayTV distributers. This trajectory cannot be sustained in the face of the extraordinary growth of digital media. We believe streaming video will see continued growth in viewership, subscription revenue and advertising sales, at the expense of traditional media, particularly, from linear TV. Moreover, we believe that the digital first players that are already well established have sizeable advantages – e.g. user data, sophisticated analytics, superior delivery platforms, extraordinary global reach, etc. – that will make it difficult for recent converts to the streaming world to compete.
Wireless Competition – Pre-standard 5G wireless networks are poised to begin competing for residential broadband service in select US markets before the end of the decade – a scenario that we have been predicting for years. The low cost of deploying small cell 5G, along with its performance characteristics (i.e. significantly expanded system capacity, support for speed increments from very low to extremely high, spectrum flexibility, and dramatically reduced latency) promise to open this, and many other new use cases, for wireless services. At the same time, the eroding spectrum advantages of the leading wireless duopoly and the implications of the shift in usage from voice to data has spurred competitive vigor in the US market. We see sizeable opportunity for the enablers of 5G and for the carriers challenging the market hegemony from the bottom.
Our Model Portfolio Constituents
ADBE (SaaS Consolidation):
We originally added Adobe to the portfolio in our May 2015 update, recognizing the successful transition of its packaged software business to a SaaS model. The stock is up 62.7% since that time and up over 28.2% since the December update. The company, which is known for its creative and marketing software product suites, managed to grow recurring subscription revenues to 82.3% in the most recent quarter ended March 3, 2017 versus 64.3% two years ago when we added the stock. Cloud subscription revenue is still growing in the double digits clocking 29.3% in the last quarter (Exhibit 3).
Exhibit 3: Key Operating and Valuation Metrics – ADBE
Adobe has a well differentiated offering, with few direct rivals in its software niche, focused on Digital Media, Digital Marketing, and Print & Publishing. The Digital Media segment, which includes creative and document cloud revenue, makes up over 68% of revenue. No other company offers a similar creative suite, but Adobe considers itself to be in competition with Apple, Autodesk, Microsoft, and others, in addition to low-end freemium apps and open source alternatives. In Digital Marketing, which makes up 30% of revenue, the company also sees no identical offerings from competitors but sees some risk from the likes of Google, IBM , Marketo, Oracle, and Salesforce, as well as a plethora of smaller ad-tech names. Adobe’s Printing & Publishing product offerings are likely to face the most exposure to competition against some large-scale electronic and web publishing systems as well as low-end desktop publishing apps, but this product segment only represents 2-3% of revenue.
Adobe fits our SaaS consolidation theme, and it also has surprisingly strong AI expertise. While we expect the company to continue growing its recurring revenue base as it adds features like its Sensei AI to its platform, it stands out along with other SaaS names for potential M&A. Still, with a market cap of about $65B, Adobe may have to be content to extend its scale through partnerships. To that end, Adobe announced a partnership with Microsoft in late March that will give customers of its’s AI-enabled digital marketing software an added kick of data from MSFT’s CRM, Dynamics, and Power BI apps. We believe that Adobe can continue to deliver growth in sales and earnings ahead of market expectations.
AMZN (AI/Cloud leader, Digital Media):
Amazon was a model portfolio constituent at inception and was briefly removed in October 2014 after a string of misses on a heavy investment cycle. After showing improving margins and financial disclosures on the state of Amazon Web Services (AWS) the next quarter, we added AMZN back into the portfolio February 2015. The stock is up 138.6% since then, and 18.6% since our December 2016 update. It is an AI/Cloud era leader and a beneficiary of the shift to digital media.
While Amazon continues to make fruitful investments, the company’s core retail business is inching toward domination. We believe Gross Merchandise Volumes (GMV), including sales fulfilled by Amazon on behalf of third parties, total at least $360B (Exhibit 4). With GMVs growing at better than 27%, Amazon will pass WalMart to become America’s largest retailer within months. With the potential for significant further growth in groceries, wholesale B2B, and other product categories, we are confident for the long term position of this franchise.
Amazon’s next biggest segment, Amazon Web Services (AWS), is the largest hyperscale infrastructure as a service (“IaaS”) cloud hosting provider by revenue. Though the AWS operation is responsible for only 9% of Amazon sales, it drives 74.2% of operating income. We believe IaaS hosting has the potential to displace more than $1T of annual enterprise spending on private data operations, with AWS in excellent position to benefit from the migration.
Amazon’s strength in AI positions it in several interesting directions. It is investing to add AI capabilities to AWS, installing AI-specific infrastructure in its data centers and has hired noted AI scientist Alex Smola from Carnegie Mellon to develop a comprehensive AI platform to offer as a service. It has notably loosened its strict privacy rules to lure Dr. Smola and other well cited computer scientists, most of whom had found the previous restrictions unacceptable. Amazon is also employing AI behind the scenes to strengthen the logistics for its retail operation, using it to power its recommendation engines for customers, and embedding it in consumer products, like the Alexa AI assistant in the Echo. Amazon is pushing 3rd parties to enable their products and services to be controlled by Alexa and to license Alexa to operate on 3rd party devices. All of this has substantial potential, not just for driving existing Amazon services and selling devices, but also as an enabler for future businesses.
Exh 4: Amazon GMV Estimates and Revenue, 2010-2016
Amazon is also disrupting traditional media with its Prime Video streaming service, taking a page from the Netflix playbook with its own original content. Like Netflix, AMZN originals have achieved critical acclaim with Emmy and Oscar wins, drawing attention to its service and pulling in new Prime members. However, unlike Netflix, Amazon offers SVOD services and options for subscribing to third party channels, serves ads that are, so far, limited to other Amazon products and services, and just announced a deal to stream Thursday Night NFL games live. Thus far, Amazon’s primary motive for Video is as a loss leader to drive consumers into its ecommerce ecosystem to buy physical and digital items, but the service is well positioned to add monetization via subscription fees and/or advertising.
Amazon investors know that it is among the least Wall Street friendly companies, giving very little detail and often, forgoing profitability in favor of investing for future topline growth. Still, despite the secrecy, Amazon is emerging as an AI leader and continues to address sizeable markets giving it plenty of runway in upending traditional retail, becoming a serious IT services player, and disrupting the distribution of video content.
Exhibit 5: Key Operating and Valuation Metrics – AMZN
FB (Digital Media, AI/Cloud Leader):
We added Facebook to the portfolio in April 2014, almost two years after its IPO and the stock is up over 138% since. Our bull case is straightforward and we’ve written about it in a couple Facebook focused deep dive research pieces (http://ssrllc.com/publication/fb-all-you-need-is-like/). We see Facebook as fitting both the “emerging AI cloud era leader” and “digital eats traditional media” themes.
Facebook CEO Mark Zuckerberg has proven to be a long-term thinker and has developed a ten-year product roadmap shown at last year’s F8 confab that includes a focus on AI and new device platforms in out years. FB is an AI contender and has strong positioning with respect to each of the three ingredients required to execute on AI: talent, data, and infrastructure. In terms of talent, it is only behind GOOGL MSFT and IBM in AI scientists with over 1K citations with 52 (Exhibit 6). FB’s AI efforts are led by notable AI guru Yann LeCun who has over 30K citations and maintains a professorship at NYU. Facebook also has a massive repository of consumer data generated by interactions amongst its 1.8B regular users. This data is unique in its map of the connections between most humans across the developed world, and is useful to track the interests and activities of Facebook users, as well as sentiment around brands and products. Thus far, Facebook has not proven as effective as its archrival Google in gauging specific purchase intentions, but it is working relentlessly to close that gap.
Exh 6: AI Citation Summary – Tech Companies
As for infrastructure, FB has the largest network of data centers exclusively for consumer applications. Unlike other hyperscale data center owners, FB doesn’t offer enterprise cloud products or services beyond advertising, which suggests significant untapped opportunity – if CEO Zuckerberg can be convinced to move beyond his narrow focus on connecting as many people as possible. It is also planning to significantly increase CAPEX this year to the tune of $7-$7.5B, to continue its AI focused data center build out. This is about the same level Alphabet spent in 2013 (Exhibit 7).
Exh 7: Hyperscale Cloud Operator Capex Spending, 2010-2016
While Facebook has warned on recent earnings calls that it faces a tough compare going forward as it is limited in ad load, we don’t expect a deceleration in ad revenue to dramatically reverse the fortunes of the company, given its co-dominance of the mobile digital ad market with Google. Both companies took nearly all the digital ad growth last year per IAB estimates. The caution around this issue is reminiscent of early concerns that Facebook might struggle with a transition to mobile, which now makes up well over 80% of revenue. With YouTube battling through some high-profile advertiser defections related to unwelcome placements of spots with objectionable content, this may be a momentary chance for Facebook to take some streaming video ad share, although we are not convinced that Facebook doesn’t face similar problems.
Though the 100M hours that Facebook streams daily is only a tenth of the video content that YouTube serves, it is a growing presence, serving ads to videos hosted on its app. Like others, Facebook is investing in developing original content and is readying the launch of a TV app. But unlike Netflix and Amazon, which has worked to expand their streaming services globally, Facebook won’t have to put in as much effort to build audience, given its 1.86B global MAUs. Facebook is also moving ahead with work on virtual and augmented reality, though we believe the more lucrative applications and larger addressable markets will be around AR technologies.
Exhibit 8: Key Operating and Valuation Metrics – FB
GOOGL (AI/Cloud Leader, Digital Media, SaaS Consolidation):
Alphabet has been a staple of our model portfolio since its inception in 2011 and has been positively exposed to nearly every TMT growth theme we’ve identified since initiating our coverage of the sector in March of 2010. The erstwhile Google is up over 216% since the start and 8.4% since our December update, underperforming the underlying S&P 500 tech components over the past 4 months due to 4Q earnings that missed because of taxes and one-time items, as well as recent concerns over advertisers leaving YouTube due to ads placed with objectionable content. We believe this headwind is only temporary given the enormous opportunity set that the company is positioned to attack. Alphabet is the dominant leader in both AI and hyperscale datacenter technologies – poised to push into enterprise applications. It is the world’s biggest streaming video platform, with the largest share of digital advertising sales. It is far and away the leader in autonomous driving technology, with many options for monetizing its stake. It has a stake in ecommerce, and could be the only viable alternative to Amazon for many struggling retailers looking to transition to the digital world. It has 7 services with more than a billion users each and untold mountains of data pertaining to almost every aspect of its users lives (Exhibit 9). It is the early technology leader in the nascent world of AI-powered virtual assistants – a paradigm shift that threatens the hegemony of the mobile app interface model. Alphabet is the 800 pound gorilla in the new TMT landscape.
Exhibit 9: Number of Monthly Active Users by Property
While the building blocks of AI systems are simple, the most powerful deep learning models are anything but (Exhibit 10). With thousands of building block algorithms combined by feedback loops into multiple layers designed to learn as they iterate through huge collections of data, outstanding AI solutions must be carefully built and adjusted for optimal results. This takes experienced talent, big data and powerful processing platforms – Alphabet excels at all of these elements. It has a disproportionate share of the AI talent pool, currently employing about 10% of 860+ AI experts with 5,000+ academic citations and about a quarter of those employed by the private sector. Google continues to hire top talent, most recently bringing aboard noted Stanford AI scientist Fei-Fei Li and poaching Snap’s Jia Li. In terms of data, Google has some 85% share of search in the US and an even greater share when it comes to mobile search, given its control of Android. By its very nature, search is the ultimate data point in understanding consumer intentions and interests and Google has been leveraging this to serve interest based advertising and AI services. We consider Alphabet to have the strongest predictive consumer data set in the world. In terms of infrastructure, Google also controls the largest and fastest hyperscale computing platform on the planet, with many millions of servers at its command. Given its efforts to sell into the enterprise with its hosting platform and productivity software, we could envision Google targeting a SaaS acquisition to kick start its sales efforts and bolt on a value-added offering.
Exh 10: The Three Ingredients of High Impact AI Systems
Alphabet’s YouTube is the most viewed digital video service in the world, now streaming some 1 billion hours of content daily, nearly 8 times the 4 billion monthly streaming hours reported by Netflix. Though the stock has been somewhat hit by news that advertisers were shifting spend from YouTube given ads appeared mixed with objectionable content, we believe Google can readily address the issue with its AI tools and quickly move past the controversy.
We continue to maintain Alphabet in the model portfolio as we believe it has plenty of runway in the long term. Beyond advertising, the company addresses more than $10 trillion in new markets that are likely to emerge over the next decade, with initiatives like Waymo autonomous vehicles, Calico/Verily health and life sciences investments, Google for Work (including Google Cloud Platform), and others. While Google often struggled in the past to monetize its new businesses, we believe that the new Alphabet organization structure is well designed to enable the company to develop, grow, and scale new businesses going forward (Exhibit 11).
Exh 11: Sizing the AI Opportunity – Three Tiers of AI Innovation
Exhibit 12: Key Operating and Valuation Metrics – GOOGL
MLNX (Hyperscale Arms Dealer):
We added interconnect chip maker Mellanox to the model portfolio in our May 2016 update as a component play for the rise of software defined networking (SDN) in hyperscale datacenters. The company makes InfiniBand and Ethernet switching and interface chips, which are staples in high performance computing environments. It also offers end to end SDN solutions. The stock is up 24.9% after our last update as the company delivered topline growth in excess of 25% in 4Q 16.
In traditional networking, there are three basic functions of a switch implemented in firmware right on each switching device. The Data Plane handles the actual switching of the data packets, the Control Plane communicates with other switching devices to adjust to network conditions, and the Management Plane enables administration of the network. SDN’s implement the control and management planes entirely in software, thus enabling the network to be managed centrally. Switches, focusing only on data plane tasks, are simpler, shifting the key performance factor from the ability to manage complex tasks to pure speed. This approach is ideal for networked hyperscale datacenters, which typically have large but predictable data flows.
According to IDC, the SDN market comprised of hardware, software, SDN applications, and services is expected to grow at a CAGR of 53.9% through 2020 to $12.5B. Mellanox competes at the leading edge of interface and switching speed, making them ideal for SDN components for hyperscale cloud operators. Though competing SDN suppliers include Intel, Cisco, and Brocade, we favor Mellanox as a pure play in the high growth market with a hot hand in delivering leading edge performance.
Exhibit 13: Key Operating and Valuation Metrics – MLNX
MSFT (AI/Cloud leader, SaaS Consolidator):
Microsoft is the poster child for successful old enterprise IT to new cloud paradigm transitions. We added the stock in September 2011 for our first model portfolio update after initiation and dividend adjusted returns amounted to 199.1% since then. The stock has underperformed the underlying S&P 500 tech components index since the last update only returning 10.0%. MSFT fits our “Emerging AI cloud era leaders” and “SaaS application consolidation” themes and is well positioned across all three AI ingredients: data, infrastructure, and talent. (http://ssrllc.com/publication/msft-an-ai-powered-vision-for-the-future-of-enterprise-computing/)
Five and a half years ago, when we added Microsoft, it suffered widespread investor distain after a lost decade under then CEO Steve Ballmer (Exhibit 14). Missed opportunities, product misfires and botched acquisitions were hallmarks of the era, but there were also serious untapped assets that have provided a foundation for the company’s renewal under current CEO Satya Nadella. In particular, Ballmer’s much maligned bet on the Bing search engine, which has never been more than nuisance to market leader Google, required building a hyperscale datacenter infrastructure and hiring a world-class team of AI scientists. These assets have been critical enablers for Microsoft’s aggressive move to the cloud.
Exh 14: Microsoft Timeline
Today, Microsoft sits in an enviable position. It has a stellar sales operation well positioned with most of the world’s IT professionals (Exhibit 15). It is successfully transitioning the world’s biggest software application franchise to a SaaS model that will give it privileged access to usage data that could enable bold new products offering insights on productivity to its customers. Its Azure cloud hosting service second, only to AWS and doubling its sales year on year, is perfectly positioned for a new wave of IaaS demand expected to come from traditional enterprises (Exhibit 16). It is also an AI powerhouse.
Microsoft’s roster of scientists with more than 1,000 citations in AI publications is second only to Alphabet, well ahead of number three IBM, the only other serious AI player amongst enterprise IT companies. MSFT also has mountains of data from the daily use of Word, Excel, Outlook, PowerPoint, Yammer, LinkedIn, Dynamics, Bing, Skype and other applications that will feed sophisticated deep learning systems to give unique and powerful insights on individual and team productivity, on organizational efficiency and bottlenecks, on the quality of relationships with customers and partners, and on best practices across the organization. Salesforce, which sees the same opportunity, does not have Microsoft’s resources. IBM has decided to focus on a very different set of opportunities.
Exh 15: Head Count and SG&A Spending
Exh 16: Cloud Infrastructure Services Market Share – Q4 2016
Exhibit 17: Key Operating and Valuation Metrics – MSFT
We believe that Nadella’s cloud and AI focused strategy is a winning tack that will allow Microsoft to extend its dominance in productivity software, to extend into a broader set of horizontal applications, to launch a new category of AI-powered management tools, to take share in the fast growing IaaS hosting market, and to enable compelling new user experiences. This makes it an easy choice for our portfolio.
NFLX (Digital Media):
Netflix, one of the Cramer coined FANG stocks that dominated market performance in 2015, was added to our portfolio in April 2013 and has appreciated a whopping 514% since then. The original content strategy, the impetus to our investment decision, has paid off – differentiating the company with consumers and reducing its dependence on licensing expensive content from other owners. Since 2013, NFLX has added over 1,000 hours of original programming and is set to underwrite the production of at least that amount in 2017. Netflix is not just churning out any content, its original series and films have found critical acclaim winning several Emmy and Oscar awards. Subscriber rolls have more than doubled since we added the name and its international subscriber base is set to overtake domestic subs sometime in the next year.
CEO Reed Hastings’ ambitious announcement at CES 2016 that Netflix would expand over 130 countries was made possible by the original content strategy. Unlike traditional TV networks, which typically license content rights from studios on a regional level, Netflix secured distribution rights for most of its series on a global basis. As a result, most content costs are fixed on a global level, and the company can launch services quickly in English and proceed to localize the service in markets it deems important, which it did in Poland and Turkey shortly after the full global roll out. NFLX only offered 3 languages when it launched its first international services five years ago in Latin America and now offers subtitles in over 20 languages on 6 continents.
Aside from its long tailed international strategy, Netflix has several other monetization levers it can pursue in the future should it’s low-price “all you can watch” strategy hit a wall. It can move to offer streaming video on demand (SVOD) services for premium/new release titles – a strategy already employed by rival Amazon Prime Video. It could offer wider pricing tiers and add advertising. It could also add live sports and events. For now, Hastings denies interest in any of these avenues, but Netflix has a history of belying its CEO’s denials as it sees necessary. This flexibility will be an asset as it moves forward.
Exhibit 18: Key Operating and Valuation Metrics – NFLX
NOW (SaaS Consolidation):
We added ServiceNow in March 2016, at the tail of a precipitous market sell off of SaaS application stocks. It is fast grower – sales were up 35% YoY in the most recent quarter – and the shares have appreciated briskly, up 40.9% since we added it a bit over a year ago. Like other SaaS names, we believe ServiceNow could be a viable acquisition target for a larger enterprise software vendor. With a cap of $14B and likely acquisition price just under $20B, it is still small enough to be targeted by a range of possible buyers. ServiceNow notably appeared the Salesforce M&A target scorecard found in leaked emails from board member Colin Powell that had been stolen by hackers.
ServiceNow provides workflow management software sold on a subscription basis, as well as implementation and configuration services. Originally focused on serving IT organizations, the company has since broadened the scope of its applications to customer support, human resources, security and other enterprise departments where patchworks of semi-automated and manual processes are still commonplace. It has some 3,600 enterprise customers including 30% of the world’s 2,000 largest companies and is intent on increasing its footprint adding customers and upselling existing accounts. Renewal rates for subscriptions have generally been strong hovering between 97-99% and the company most recently reported top line growth of 35% and is guiding for growth in the low 30% range for the rest of the year.
We believe that ServiceNow, with plenty of runway ahead of it and few real competitors, can sustain its trajectory. That growth, plus the potential of a takeout for a premium, give us confidence for the ongoing performance of the stock.
Exhibit 18: Key Operating and Valuation Metrics – NOW
CRM (SaaS Consolidation, AI/Cloud):
Salesforce is a two-time model portfolio constituent, having initially been placed in when we initiated model portfolio coverage and was removed in April 2014 on strong performance. We added it back during our latest update in December 2016 and the stock is up 19.3% since. Salesforce is the leading customer relationship management (CRM) platform and the largest cloud pureplay. The company has also been the subject of M&A speculation for years. Although its rich $60B market cap could preclude a deal, it could still be deemed a viable acquisition candidate for a deep pocketed buyer, such as Alphabet, Amazon or Microsoft. Any kind of deal would require a sizeable premium and buy-in from founder and CEO Marc Benioff who wields considerable clout.
Exhibit 19: Key Operating and Valuation Metrics – CRM
While Salesforce obviously fits our SaaS consolidation theme, the company is also aggressively investing in deploying AI. Benioff committed to AI in 2014, when he acquired RelateIQ and incorporated its relationship intelligence technology into Salesforce’s core application. Since then, the company has stepped up its investment, culminating in its April 2016 acquisition of deep learning think tank MetaMind and the elevation of that company’s founder to its own Chief Scientist post. Salesforce now has the talent to be a player. It has 175 scientists working on AI and 4 experts from the 1,000 citations list, good enough to make our list of the top ten companies in AI. It can use the data from millions of daily user interactions on its SaaS software to gain useful insights for its customers. To that end, Benioff formally announced Saleforce’s Einstein AI platform this past September. Also, the company has engaged in an AI partnership with IBM. Salesforce customers will have access to IBM’s suite of Watson AI APIs (e.g. natural language processing, analytics engines, etc.) and IBM will push Salesforce applications through its extensive sales organization. The synergies to this partnership are obvious, with clear value likely to accrue to both companies.
Salesforce does have a weakness. SaaS applications are sold as bundle, tying the application itself with the infrastructure that hosts it. For Salesforce, its extraordinary CRM and marketing applications carry the modest burden of its less than competitive datacenter operations. Compared to behemoths like Microsoft, Amazon and Google, Salesforce’s platform is subscale and high cost. Ultimately, we see a partnership with or even an acquisition by one of these companies as a likely future resolution. Still, we see expectations for 21% topline and 30% bottom line growth in 2017 as easily achievable in a strong market for SaaS, underscoring its merit for inclusion in our portfolio.
NVDA (Hyperscale arms dealer):
Nvidia is also a two-time model portfolio constituent. We first added it in December 2011 after design wins for its mobile device SoC, but removed it in September 2013 when that momentum failed. We returned to Nvidia in August 2016, recognizing its success selling its GPUs to hyperscale data centers for AI applications. Since our addition, the stock has been up 58.2% and has returned some 4.8% since the December update.
Nvidia shares have been under pressure of late on concerns that its gaming GPU business, which accounted for 57% of 2016 revenue and grew 35% YoY, could encounter stiff headwinds in 2017 – e.g. weaker market demand and renewed competition from AMD. While we acknowledge those risks, we also believe consensus expectations for a precipitous drop from 55% growth in the most recent quarter to just 16% for the whole of 2017 are overly pessimistic, particularly given the company’s substantial opportunities in its datacenter and automotive businesses.
GPUs have performance benefits over traditional CPUs for AI applications. Deep learning AI programs are built from thousands of connected algorithms that can run in parallel while iterating many times through massive data bases. GPUs can accelerate 90% of the algorithms in a typical AI, allowing a neural network to learn faster on GPU based servers than on traditional CPUs. As hyperscale datacenter operators like Amazon, Google, Microsoft and IBM aggressively expand AI support, their demand for GPUs follows. While datacenters made up just 11.3% of Nvidia’s 2016 sales, the growth in the category was over 120%. With hyperscalers beginning to offer hosting on Nvidia-brand GPUs as a standard AI offering, and the rapid expansion of AI development we expect demand for this hardware to continue to grow at a triple digit pace in 2017.
These same GPU performance advantages also play for production AI systems deployed in devices, such as automobiles. Nvidia is a leader in supplying computing platforms for autonomous driving. From Advanced Driver Assistance Systems (ADAS) to full autonomy, NVDA has a comprehensive product portfolio and has built strong OEM relationships that have driven impressive design wins. While automotive makes up only 7.6% of current revenue, it is growing at over 58% and can potentially accelerate with the rollout of fully autonomous vehicles (http://ssrllc.com/publication/autonomous-cars-self-driving-ambition/).
We believe that the cautious narrative around Nvidia shares significantly undervalues the likely near term trajectory of its SI datacenter and automotive businesses, leaving its 2016 estimates easily achievable. Meanwhile, its long-term growth prospects position it to be a dominant player in the cloud/AI era.
Exhibit 20: Key Operating and Valuation Metrics – NVDA
QCOM (Wireless, Hyperscale Arms Dealer):
With Google, Qualcomm is one of two companies to have been model portfolio constituents consistently since inception. Over those 6 years, dividend adjusted returns have been only 13.9%. While our loyalty has not been well rewarded, we remain convinced of the long term value of the company’s intellectual property business (QTL) and its opportunities as a leader in chips for a wireless market (QCT) that could see a new phase of growth with 5G.
Qualcomm had a very bad year in 2014. China’s National Development and Reform Commission (NDRC) commenced an investigation into QTL’s licensing practices, which brought royalty negotiations with indigenous Chinese brands to a standstill. During the investigation, many existing licensees were emboldened to deliberately underreport their sales and withhold payments. China was the hottest growth market in wireless, and the homegrown players were leading the charge but not paying Qualcomm. By 1QFY16, royalty revenues were off 12% YoY.
Exh 21: Smartphone App Processor Market Share, 2010 – 2015
Also in 2014, Qualcomm’s R&D group administered a self-inflicted wound – ARMH, the standards bearer for all smartphone processors, was moving to a 64bit processing architecture, but QCT’s internally developed version would not be ready in time for the next generation of chipsets. To that point, its flagship Snapdragon SoC had been extending its performance lead over commercial alternatives, with only AAPL’s own proprietary implementation ahead of it. Now QCT had to hurriedly transition to the same ARM reference design that most of its rivals used, wiping away much of its performance advantage. Samsung dropped the Snapdragon for its Galaxy 6 in favor of its own Exynos SoC. MediaTek ate away at Qualcomm’s market share from below, without a big performance gap to deter it. For CY15, Snapdragon’s share of smartphone SoCs plummeted from 52% to 42% (Exhibit 21).
Resolutely, Qualcomm fought back. An agreement was struck with Chinese authorities in February 2015 establishing relatively favorable terms, but QTL then had to hammer out resolutions with each of the Chinese OEM’s, most of whom continued to withhold payment, pending agreement. The new proprietary 64bit CPU design, dubbed Kyro, was ready for OEMs for their 2016 products, re-establishing QCT’s performance leadership and winning back market share, notably with Samsung’s marquee Galaxy S7. Pressure from activist shareholder Jana Partners induced management to take aggressive actions to create value including laying off 15% of the workforce, significantly scaling back the generous employee stock compensation program, exiting tangential businesses, and cost-cutting.
While the restructuring has been largely successful, January 2017 brought the company another whammy: a series of lawsuits from the FTC and Apple that threaten to unravel QTL’s licensing practices. Apple is not a direct Qualcomm licensee, but indirectly pays the royalties via its contract manufacturers (“CMs”). Apple and Qualcomm maintain a Business Cooperation and Patent Agreement (“BCPA”) that expressly caps the royalties that Apple pays to Qualcomm via its CMs and provides for rebates back to Apple to manage the cap. Apple claims that $1 billion of these royalty rebates were withheld because of its cooperation with investigators from the Korean Fair Trade Commission (KFTC), a serious allegation that Qualcomm denies. At first glance, it seems unlikely that Qualcomm’s experienced legal team – which oversees any such contractual disputes – would have allowed the rebates to be used in this way. We note that the amount of any rebate payment in question would still have been allocated for financial purposes, and thus would not represent an unaccounted obligation for future earnings.
Exhibit 22: Key Operating and Valuation Metrics – QCOM
Investors have taken this conflict very seriously, but we see significant reason for optimism. Qualcomm’s licensing and royalty frameworks have been successfully defended for over 2 decades against challenges across jurisdictions on every continent. Given the political winds and historical precedent, we expect the FTC suit to be dropped, and for Apple to eventually agree to a settlement, naturally with terms undisclosed. Looking further ahead, we believe that Qualcomm is likely to be the strongest intellectual property contributor to the coming 5G wireless standard, positioning it to sustain its royalty streams into the next era. We also note that QCT has excelled in managing technology transitions – it is typically first to market with SoC solutions integrating new standards and usually sets the bar for performance, driving share gains early in the adoption cycle. 5G will be a key enabler for the long-hyped Internet of Things (IoT) market and for autonomous vehicles. These could be important new markets for Qualcomm. Finally, the major hyperscale datacenter operators have been testing Qualcomm’s ARM-based server chips for certain use cases. We believe it is likely that this could result in another major new market for QCT. It is with this perspective that we enthusiastically retain Qualcomm in the portfolio.
TMUS USA (Wireless):
We added T-Mobile to the model portfolio in October 2014 after the company, led by charismatic CEO John Legere, began show serious share gains from the aggressive “Uncarrier” marketing initiatives that it had launched the previous year. “Uncarrier” features included eliminating service contracts, offering free international roaming, contract buyouts from other carriers, and free data for music streaming from select services in order to attract new customers. With wireless in the US largely mature, competition is characterized as a zero-sum game. Network quality is no longer a major differentiator as data speeds, local availability, and network congestion have replaced unbroken coverage maps as the most important consideration for users. T-Mobile has a robust, uncrowded network in most major markets, and a customer acquisition strategy that is working. T-Mobile stock is up 124.8% since being added to the portfolio and up 14.1% since the December update.
The incumbent duopolists, AT&T and Verizon have been reluctant to directly match T-Mobile offers, for fear of eroding cash flows and pressuring service revenues. However, recent months have seen the big two a bit more aggressive with offers and advertising meant to stem the flow of sub defections. Thus far, T-Mobile’s momentum has rolled on. While AT&T and Verizon like to tout subscriber additions, the numbers are bloated by low revenue connected devices, such as tablets, cars and home security systems. Looking at the far more lucrative post-paid phone subscriptions tells a different story – AT&T has lost postpaid phone subs for 9 straight quarters, while Verizon’s postpaid growth has dwindled to a standstill. In contrast, T-Mobile has added 13.6M postpaid phone sups and grown service revenue by 80.8% since the debut of its “Uncarrier” initiatives in 2013 (Exhibit 23).
Despite the reactions of the market leaders, we expect T-Mobile to continue its success. AT&T and Verizon are stuck in a classic game theory position, where it is better to suffer slow bleeding than self-inflict more serious harm. 5G will open new opportunities – residential broadband, IoT connectivity, etc. – while we see room for T-Mobile to take further share with its consumer-friendly service offers.
Exh 23: US Wireless Carrier Service Revenue Share, 1Q13 – 4Q16
Exhibit 24: Key Operating and Valuation Metrics – TMUS
WDAY (SaaS Consolidation):
We added Workday to the portfolio in April 2013, based on an expectation of strong SaaS application growth. Shares are up 41.9% since we added the name and up 11.4% since the last update. Workday offers a suite of cloud applications for enterprise resource planning including modules for financial and human capital management and competes with the likes of ORCL and SAP. It has over 1,500 customers – mostly medium to large sized businesses. Workday’s addressable market is worth over $60B, and at 35% annual growth it is rapidly taking share vs. the 14.5% underlying growth of cloud ERP software forecasted by Gartner. Workday continues to win big deals at the expense of the traditional ERP leaders SAP and Oracle, most recently signing its biggest contract ever at Walmart. Though the implementation is likely to start as a pilot program, WDAY software can potentially touch over 2M Walmart employees and serve as the ultimate test case for cloud scalability. While Walmart is by far its largest customer, WDAY CEO Aneel Bhusri noted the company has signed on 136 of the Fortune 500 for Workday Human Capital Management.
Exhibit 25: Key Operating and Valuation Metrics – WDAY
Signing on new ERP customers is often challenging given previous bad experiences with deployments and implementations. Case examples of public companies missing earnings on botched ERP implementations include Avon, Goodyear, and UPS. But Workday has established an excellent track record on implementation, bringing new customers online at less than half the cost of an on-premise solution and taking customers live increasingly faster. In terms of products, Workday continues to develop new application modules launching new financials and analytics features that can in turn increase the company’s addressable markets.
With its strong growth and approaching profitability, Workday is a strong investment. It could also be an attractive target for a would-be acquirer looking to expand its SaaS ERP offering or a partner for a cloud host like Google or Amazon looking to extend their offerings to enterprise customers.
XLNX (AI/Cloud Arms Dealer):
Xilinx was added to the model portfolio in our most recent update, and follows the theme of components for AI-tuned datacenters. Xilinx is primarily a supplier of programmable logic and invented the field programmable gate array (FPGA) in the 1980s. FPGAs are integrated circuits designed to be configured by a customer or designer after manufacturing, making it more cost effective for an end-user to deploy than a self designed application specific integrated circuit (ASIC). As a result, FPGAs are broadly found in industry-vertical applications ranging from aerospace/defense, automotive, consumer electronics, financial services, industrials, medical devices, and wireless communications. Importantly, FPGA chips also have AI applications, which is the underlying reason for adding XLNX to the portfolio.
Exhibit 26: Key Operating and Valuation Metrics – XLNX
As deep-learning frameworks evolve, designing custom hardware can be difficult. FPGAs are useful for running specialized algorithms without having to invest in customized hardware, like GOOGL’s Tensor Processing Unit ASIC. FPGAs can be tweaked to run an algorithm faster or more efficiently and are generally considered to be more power efficient than GPUs as measured by performance per watt. Though FPGAs aren’t ideal for training algorithms, they are useful for accelerating the performance of applications. MSFT experimented with FPGAs in powering Bing search finding that FPGAs doubled the throughput of an algorithm at a cost of just 10% more power per server. There are only two major players in FPGAs, Xilinx and Intel’s Altera, which was taken out in December 2015. Thus far, hyperscale datacenter owners have shown a strong preference for Xilinx, perhaps strengthened by their already substantial dependence on Intel CPUs.
Xilinx acquisition rumors are a regular occurrence, and the rapid rise in AI related demand could raise the heat. Even without the possibility of a takeout, we believe that consensus expectations underrate the long-term potential of Xilinx’s datacenter business.
IBM (AI/Cloud Leader):
We recently wrote two pieces outlining our case for IBM and the value of its Watson AI business. (http://ssrllc.com/publication/ibm-are-we-there-yet/ and http://ssrllc.com/publication/ibm-watson-steak-or-just-sizzle/) While IBM portrays itself as two companies – a declining annuity from its legacy datacenter hardware, software and IT services businesses, and a set of growth businesses termed the “strategic imperatives” that includes Watson and IaaS services – we break it down a bit more finely. We believe IBM’s transition back to growth will come from its private/hybrid and public cloud businesses, which include Watson. There are three main ingredients for strong AI development – scientific talent, data and infrastructure – and IBM is well positioned for all three.
IBM’s vertical industry focus for Watson is well matched to its institutional strengths – deep customer sales and services relationships, and longstanding industrial IT experience. Vertical AI applications is an extraordinary opportunity, made more so by a dearth of competition. The other top AI players (Alphabet, Microsoft, Facebook, Amazon and others) have focused on consumer and horizontal enterprise application markets, leaving IBM to compete against promising but untested startups. We think that bespoke AI-rich applications to address substantial business problems in industries like health care, financial services, energy, retail, and others have the potential to be $100B+ market within a decade. IBM’s focus on customizable analytics solutions, with services for data formatting, application development, and system maintenance, and a comprehensive AI and hybrid-cloud friendly platform, is a strong offering well attuned to the needs of these customers. In contrast, the other AI leaders offer relatively bare-bones, DIY AI platforms, while IBM’s traditional IT rivals offer nothing. In most of these vertical markets, the primary competitors will be startups hoping that even closer industry focus can make up for IBM’s obvious advantages.
This is more than enough to fuel our long-term growth expectations for the company. We recently assessed IBM’s likely revenue trajectory (http://ssrllc.com/publication/ibm-are-we-there-yet/) in our deep dive on the company. Given IBM data disclosed, we were able to divide sales into four growth categories – legacy, non-cloud strategic, private/hybrid cloud, and true cloud. The legacy business has been deteriorating in the low double digits, while non-cloud strategic have been showing signs of a slow decline. We believe that strong near 50% growth in the true cloud category, led by Watson, and teens growth in private/hybrid cloud will offset shrinking sales in the other two categories within 4-6 quarters (Exhibit 27). We see this inflection point as a significant investing opportunity and are adding IBM to the model portfolio going forward.
Exh 27: IBM Revenue Forecast by SSR Business Category, FY2016-FY2021
Exhibit 28: Key Operating and Valuation Metrics – IBM
Of the stocks that had been in our model portfolio, Twitter has been the worst performer and is down -72.6% since we added and down -21.1% since the last update. While we still believe Twitter has a unique and differentiated product, a series of management missteps and conflicts at the top have ultimately kept the company from achieving its full potential. Three problems have plagued the company since IPO: communication of Twitter’s value to advertisers and Wall Street, internal strife at the top, and product execution failures.
Exhibit 29: Key Operating and Valuation Metrics – TWTR
First, we believe the company erred during its IPO process by promoting Facebook’s favored metrics to gauge adoption and engagement. For Twitter, whose content is often retransmitted across media outlets, figures like MAUs and DAUs are largely irrelevant in calculating actual audiences reached. The company spent the past couple of years making the case for its unregistered user base, but hasn’t been able to convince advertisers or Wall Street of their value. Second, the company has experienced continued management turbulence at the top. Comedian turned CEO Dick Costolo helmed the company for a year and a half after IPO in a tenure characterized by infighting and ineffectual leadership. After Costolo’s departure in June 2015, the board decided to bring back co-founder Jack Dorsey as an interim CEO despite already holding a CEO job at his other company, Square. Dorsey assumed the role of full-time CEO in October 2015. Though Dorsey’s arrival was initially praised by investors, his tenure has also proved difficult and the company saw several key executive leaves with the departures of COO Adam Bain, CTO Adam Messinger, and VP of Product Josh McFarland late last year.
Third, product enhancements and improvements have been slow to roll out and been ineffective. Twitter remains plagued by fake accounts and bots, which a study published by the University of Southern California last month found that at least 15% of Twitter accounts are fake. Features like “Moments,” with curated content were abandoned, and Twitter’s new user onboarding initiative seems to have had little effect bringing in new and engaged users.
We believe Twitter will eventually prove to be an excellent acquisition for a company that can drive adoption by new users and exploit its unique consumer data assets, but we have lost patience. With Twitter continuing to struggle with product and failing to sell itself last year, we see no near-term catalysts to re-accelerate user growth and engagement and are removing the name from the portfolio.
Our large cap model portfolio has outperformed the broad S&P index 1112 basis points since inception. Over the past 12 months, we outperformed the S&P 500 in 3 of 4 portfolio updates and outperformed the S&P tech components twice. Overall, our large cap model portfolio outperformed the S&P 500 by 1370 basis points and the S&P 500 tech components by 230 basis points since last year’s update in March 2016. The portfolio was hit earlier last year with the January / February 2016 sell-offs, but has rebounded since. We note the absence of AAPL from the model portfolio narrowed our performance relative to the underlying tech components of the S&P 500, while names like TWTR and QCOM face idiosyncratic headwinds. We’ve replaced TWTR with IBM going forward, but continue to maintain QCOM in as we expect the company to weather its litigation with AAPL.
Exh 30: SSR TMT Large Cap Portfolio Performance vs. Benchmarks Since Inception
©2017, SSR LLC, 225 High Ridge Road, Stamford, CT 06905. All rights reserved. The information contained in this report has been obtained from sources believed to be reliable, and its accuracy and completeness is not guaranteed. No representation or warranty, express or implied, is made as to the fairness, accuracy, completeness or correctness of the information and opinions contained herein. The views and other information provided are subject to change without notice. This report is issued without regard to the specific investment objectives, financial situation or particular needs of any specific recipient and is not construed as a solicitation or an offer to buy or sell any securities or related financial instruments. Past performance is not necessarily a guide to future results.