jebmo’s posthaven

I’ve seen a lot of failed machine learning models in the course of my work. I’ve worked with a number of organizations to build both models and the teams and culture to support them. And in my experience, the number one reason models fail is because the team failed to create a minimum viable product (MVP).

In fact, skipping the MVP phase of product development is how one legacy corporation ended up dissolving its entire analytics team. The nascent team followed the lead of its manager and chose to use a NoSQL database, despite the fact no one on the team had NoSQL expertise. The team built a model, then attempted to scale the application. However, because it tried to scale its product using technology that was inappropriate for the use case, it never delivered a product to its customers. The company leadership never saw a return on its investment and concluded that investing in a data initiative was too risky and unpredictable.

If that data team had started with an MVP, not only could it have diagnosed the problem with its model but it could also have switched to the cheaper, more appropriate technology alternative and saved money.

In traditional software development, MVPs are a common part of the “lean” development cycle; they’re a way to explore a market and learn about the challenges related to the product. Machine learning product development, by contrast, is struggling to become a lean discipline because it’s hard to learn quickly and reliably from complex systems.

Yet, for ML teams, building an MVP remains an absolute must. If the weakness in the model originates from bad data quality, all further investments to improve the model will be doomed to failure, no matter the amount of money thrown at the project. Similarly, if the model underperforms because it was not deployed or monitored properly, then any money spent on improving data quality will be wasted. Teams can avoid these pitfalls by first developing an MVP and by learning from failed attempts.

Return on investment in machine learning

Machine learning initiatives require tremendous overhead work, such as the design of new data pipelines, data management frameworks, and data monitoring systems. That overhead work causes an ‘S’-shaped return-on-investment curve, which most tech leaders are not accustomed to. Company leaders who don’t understand that this S-shaped ROI is inherent to machine learning projects could abandon projects prematurely, judging them to be failures.

Above: The return-on-investment curve of Machine Learning initiatives shows an S-curve compared to traditional Software Development projects, which have a more linear ROI.

Unfortunately, prematurely terminating a project happens in the “building the foundations” phase of the ROI curve, and many organizations never allow their teams to progress far enough into the next phases.

Failed models offer good lessons

Identifying the weaknesses of any product sooner rather than later can result in hundreds of thousands of dollars in savings. Spotting potential shortcomings ahead of time is even more important with data products, because the root causes for a subpar recommendation system, for instance, could be anything from technology choices to data quality and/or quantity to model performance to integration, and more. To avoid bleeding resources, early diagnosis is key.

For instance, by foregoing the MVP stage of machine learning development, one company deploying a new search algorithm missed the opportunity to identify the poor quality of its data. In the process, it lost customers to the competition and had to not only fix its data collection process but eventually redo every subsequent step, including model development. This resulted in investments in the wrong technologies and six months’ worth of man hours for a team of 10 engineers and data scientists. It also led to the resignation of several key members on that team. Each departed employee cost $70,000 per person to replace.

In another example, a company leaned too heavily on A/B testing to determine the viability of its model. A/B tests are an incredible instrument for probing the market; they are a particularly relevant tool for machine learning products, as those products are often built using theoretical metrics that do not always closely relate to real-life success. However, many companies use A/B tests to identify the weaknesses in their machine learning algorithms. By using A/B tests as a quality assurance (QA) checkpoint, companies miss the opportunity to stop poorly developed models and systems in their tracks before sending a prototype to production. The typical ML prototype takes 12 to 15 engineer-weeks to turn into a real product. Based on that projection, failing to first create an MVP will typically result in a loss of over $50,000 if the final product isn’t successful.

The investment you’re protecting

Personnel costs are just one consideration. Let’s step back and discuss the wider investment in AI that you need to protect by first building an MVP.

Data collection. Data acquisition costs will vary based on the type of product your building and how frequently you’re gathering and updating data. If you are developing an application for an IoT device, you will have to identify which data to keep on the edge vs. which data to store remotely on the cloud where your team can do a lot of R&D work on it. If you are in the eCommerce business, gathering data will mean adding new front-end instrumentation to your website, which will unquestionably slow down the response time and degrade the overall user experience, potentially costing you customers.

Data pipeline building. The creation of pipelines to transfer data is fortunately a one-time initiative, but it is also a costly and time-consuming one.

Data storage. The consensus for a while now has been that data storage is being progressively commoditized. However, there are more and more indications that Moore’s Law just isn’t enough anymore to make up for the growth rate of the volumes of data we collect. If those trends prove true, storage will become increasingly expensive and will require that we stick to the bare minimum: only the data that is truly informational and actionable.

Data cleaning. With volumes always on the rise, the amount of data that is available to data scientists is becoming both an opportunity and a liability. Separating the wheat from the chaff is often difficult and time-consuming. And since these decisions typically need to be made by the data scientist in charge of developing the model, the process is all the more costly.

Data annotation. Using larger amounts of data requires more labels, and using crowds of human annotators isn’t enough anymore. Semi-automated labeling and active learning are becoming increasingly attractive to many companies, especially those with very large volumes of data. However the licenses to those platforms can represent a substantial add to the entire price of your ML initiative, especially when your data shows important seasonal patterns and needs to be relabeled regularly.

Compute power. Just like data storage, computer power is becoming commoditized, and many companies opt for cloud-based solutions such as AWS or GCP. However, with large volumes of data and complex models, the bill can become a considerable part of the entire budget and can sometimes even require a hefty investment in a server solution.

Modeling cost. The model development phase accounts for the most unpredictable cost in your final bill because the amount of time required to build a model depends on many different factors: the skill of your ML team, problem complexity, required accuracy, data quality, time constraints, and even luck. Hyperparameter tuning for deep learning is making things even more hectic, as this phase of development benefits little from experience, and usually only a trial-and-error approach prevails. Typical models will take about six weeks of development for a mid-level data scientist, so that’s about $15K in salary alone.

Deployment cost. Depending on the organization, this phase can either be fast or slow. If the company is mature from an ML-perspective and already has a standardized path to production, deploying a model will likely take about two weeks of time by an ML engineer, so about $5K. However, more often than not, you’ll require custom work, and that can make the deployment phase the most time-consuming and expensive part of creating a live ML MVP.

Above: The pyramid of needs for machine learning.

The cost of diagnosis

Recent years have seen an explosion in the number of ML projects powered by deep learning architectures. But along with the fantastic promise of deep learning comes the most frightening challenge in machine learning: lack of explainability. Deep learning models can have tens, if not hundreds of thousands, of parameters, and this makes it impossible for data scientists to use intuition when trying to diagnose problems with the system. This is likely one of the chief reasons ineffective models are taken offline rather than fixed and improved. If, after weeks waiting for the ML team to diagnose a mistake, they still can’t find the problem, it’s easiest to move on and start over.

And because most data scientists are trained as researchers rather than engineers, their core expertise as well as their interest rarely lies in improving systems but rather in exploring new ideas. Pushing your data science experts to spend most of their time “fixing” things (which could cost you 70 percent of your R&D budget) could considerably increase the churn among them. Ultimately, debugging, or even incremental improvement of an ML MVP can prove much more costly than a similarly-sized “traditional” software engineering MVP.

Yet ML MVPs remain an absolute must, because if the weakness in the model originates in the bad quality of the data, all further investments to improve the model will be doomed to failure, no matter how much money you throw at the project. Similarly, if the model underperforms because it was not deployed or monitored properly, then any money spent on improving data quality will be wasted.

How to succeed with an MVP

But there is hope. It is just a matter of time until the lean methodology that has seen huge success within the software development community proves itself useful for machine learning projects as well. For this to happen, though, we’ll have to see a shift in mindset among data scientists, a group known to value perfectionism over short time-to-market. Business leaders will also need to understand the subtle differences between an engineering and a machine learning MVP:

Data scientists need to evaluate the data and the model separately. The fact that the application is not providing the desired results might be caused by one or the other, or both, and diagnosing can never converge unless data scientists keep this fact in mind. Because data scientists now have the option of improving their data collecting process, they can do justice to those models that would have been otherwise identified as hopeless.

Be patient with ROI. Because the ROI curve of ML is S-shaped, even MVPs require more way work than you could typically anticipate. As we have seen, ML products require many complex steps to reach completion, and this is something that needs to be profusely communicated to stakeholders to limit the risk of frustration and premature abandonment of a project.

Diagnosing is costly but critical. Debugging ML systems is almost always extremely time-consuming, in particular because of the lack of explainability in many modern models (DL). Building from scratch is cheaper but is a worse financial bet because humans have a natural tendency to repeat the same mistakes anyway. Obtaining the right diagnostic will ensure your ML team knows with precision what requires attention (whether it be the data, the model, or the deployment), allowing you to prevent the costs of the project from exploding. Diagnosing problems also gives your team the opportunity to learn valuable lessons from their mistakes, potentially shortening future project cycles. Failed models can be a mine of information; redesigning from scratch is thus often a lost opportunity.

Make sure no single person has the keys to your project. Unfortunately, extremely short tenures are the norm among machine learning employees. When key team members leave a project, its problems are even harder to diagnose, so company leaders must ensure that “tribal” knowledge is not owned by any one single person on the team. Otherwise, even the most promising MVPs will have to be abandoned. Make sure that once your MVP is ready for the market, you start gathering data as fast as possible and that learnings from the project are shared with your entire team.

No shortcuts

No matter how long you have worked in the field, ML models are daunting, especially when the data is highly dimensional and high volume. For the highest chances of success, you need to test your model early with an MVP and invest the necessary time and money in diagnosing and fixing its weaknesses. There are no shortcuts.

Jennifer Prendki is VP of Machine Learning at Figure Eight.

Campbell’s imaginative life often revolved around the desire to maintain emotional order in an emotionally disordered world. And at the heart of his personal metaphysic lay a desire to extol the virtues of a type that he presumed himself to be — the “competent man,” as exemplified by the firm-but-kind father figures in Heinlein­’s best novels and stories. Unlike the qualified, limited-ability academics who spurned him at Duke and MIT, the “competent man” knows a little bit about everything, such as how to “change a diaper, plan an invasion, butcher a hog, conn a ship, design a building, write a sonnet, balance accounts, build a wall, set a bone, comfort the dying, take orders, give orders, cooperate, act alone, solve equations, analyze a new problem, pitch manure, program a computer, cook a tasty meal, fight efficiently, die gallantly. Specialization is for insects.”

User Exodus Didn't Materialize In Q3 ... And It Won't In The Future

In my last article on Facebook (NASDAQ:FB) , I lowered my price target on Facebook from $240 to $220. I was fearful that Facebook would report a Q3 users exodus in not just Europe, but North America as well. Europe and North America are by far Facebook's largest markets for monetization, so any weakness from Facebook in either one of these markets can have an adverse, and material affect on Facebook's revenue growth. But in Q3, Facebook reported results that were in some ways better than Q2 on the user side.

(source: Facebook IR)

In Q2, Facebook lost 3 million DAUs from Europe, a 1% sequential decline. This came mainly at the hands of GDPR's impact on user growth. In Q3 however, Facebook only lost 1 million DAUs in Europe, despite management's Q2 warning about DAUs being hit harder as GDPR takes full effect. Meanwhile, Facebook's North American DAU count stayed flat for the third straight quarter, at 185 million DAUs, showing that Facebook's relevance in the US and Canada hasn't budged.

The question now is if a user exodus will ever materialize. In my opinion, no massive user exodus will come to fruition, which is necessary to trigger a massive slowdown in core Facebook's revenue growth. As of the Q3 data, a user exodus coming from any of Facebook's major geographies seems like a faraway chance. In my opinion, Facebook's user base should continue to grow, the question is at what rate. Wall Street seems to be pricing in a catastrophic hit to core Facebook's user numbers, whereas the Q3 data and confidence displayed by management suggests quite the contrary.

That being said, younger people and millennials as a demographic are leaving Facebook. Ever since older generations have adopted the Facebook platform, Facebook has lost its cool factor, and is nearly irrelevant amongst teens. Even Twitter is more relevant amongst teens.

Facebook is expected to lose 2.2 million 12-17 year olds by 2022. Future purchasing trends are dictated by Generation Z/teens, with millennials dictating current purchasing decisions. And with Facebook losing users in this key segment, advertisers may be more hesitant to use Facebook for their advertising needs. However, Facebook remains strong amongst older users. Facebook's user base is active and while stalling remains lucrative for advertisers and will continue to be in the future.

In addition, Facebook has "sweeteners" for user in place. Products like Facebook Watch and Stories may eventually keep users drawn to the Facebook platform, and prevent the Titanic-esque sinking in Facebook's user numbers that some have predicted.

Core Facebook's Monetization Runway

Facebook's previous astonishing revenue growth rates came from the company's ability to combine robust user growth in key markets (North America and Europe) with increased monetization of the Facebook platform. As Q3 demonstrated, the user growth in Facebook's core markets is beginning to slow if not completely stagnate. The only DAU user growth is coming from Facebook's Asia/Pacific and rest-of-world (ROW) segments. In my opinion however, core Facebook doesn't need high levels of DAU growth to increase its core revenue at strong growth rates. The real long-term path for the core Facebook platform's growth comes from Facebook further ability to monetize the platform. I believe the combination of the oligopolistic trend in the internet advertising space, the growth of the overall space, and Facebook's market leadership and coming market share gains will all act as tailwinds for core Facebook's monetization.

In my opinion, the internet advertising space is controlled by two main players, with a third emerging. Those two players are Facebook and Google, with Amazon (NASDAQ:AMZN) being the emerging player. Ever since Facebook began gaining traction amongst advertisers, the company has developed high-quality ROI that have made them and Google (NASDAQ:GOOG) the two "must haves" in the internet advertising world. The main reason Facebook has been able to generate this massive ROI has been its ad targeting practices. Recently, these practices have been held in a negative light, with regulators and sell-side research analysts becoming increasingly skeptical of Facebook's ability to continue their current targeting. Facebook has been accused of scraping every bit of data on the user, and even tracking that user as he/she operates across the internet and into the offline world. All of these measures allow Facebook to create a digital profile on each of these people.

While this may be deemed unethical, the nature of the internet advertising industry has always been the advertiser's ability to target its users most effectively for advertisements. Before internet platforms like Facebook and Google, marketers relied on legacy media platforms like TV and print for their marketing needs. While these marketing platforms are still used today, the ROI on advertisements from legacy platforms is substantially lower than that from internet based platforms. Why? It is simply because these marketers don't know precisely who they are targeting through a TV screen. The marketers only the basics about who is watching the ads. Think of the user as a puzzle. Marketers have only one piece of the puzzle with legacy platforms. Whereas marketers that utilize the internet have nearly the whole puzzle right in front of them. And best of all, it is actually cheaper on a CPM basis to use Facebook than to run TV ads. So Facebook is less expensive and offers better long-term products to marketers. Here is the current expectation for the 2019 market relative to the 2016 market.

From 2016 to 2019, mobile's share of the pie is expected to grow from just 15% to 27% in three years. Meanwhile, Facebook's CPM is roughly $7. In comparison, TV CPM is at ~$17.

(source: Statista)

All we need is for a negative catalyst to hit legacy media. In my opinion, the increased competition from platforms like Netflix (NASDAQ:NFLX), Roku (NASDAQ:ROKU), and some of the legacy media companies themselves in the OTT streaming market could provide a legitimate threat to TV and print to expand. All it will take is for some of the recurring ad buyers to see a drop-off in usage of TV and print media for these firms to pull their budgets from legacy media and plop it into the next best thing. Again, Facebook has a $7 CPM versus TV's CPM of $17, meaning you can get better targeting, ROI, and are reaching a far larger audience, for ~40% of the price. And with Facebook being the market leader in this huge market, it seems like Facebook would be a likely target for these budgets.

This market is massive, coming in at $178 billion in total spending for 2017.

(source: Recode)

Ironically, the same legacy media companies that are at risk, are actually accelerating the demise of their old selves. Shedding their old skin and reinventing what it means to be say Comcast (NASDAQ:CMCSA) or CBS (NYSE:CBS), by entering the same streaming market that is eating away market share from their traditional over-the-air media businesses.

And finally, a key catalyst for recurring marketing on legacy platforms is exposure to one-time events that will occur regularly. Such events include the Olympics, the NFL, or most recently the midterm elections. People regularly tune into TV stations to watch these events play out. But increasingly, these events are coming to social media, with Twitter (NYSE:TWTR) streaming live NFL games, Snap (NYSE:SNAP) bringing the Olympics to its users, and Facebook itself becoming a key source of news for its users. The internet advertising space is quickly closing in on legacy media, and legacy media appears to be attempting a reinvention of itself to survive.

Regulators Won't Do Anything

Another critical component to the bear argument is that regulators will crack down on Facebook's ability to operate their current business model. There is a saying that I believe rings true here:

"There is either no regulation, or there is either way too much regulation."

In my opinion, the odds of any material regulation infringing on Facebook's ability to do business is a long-shot chance. Why? Let's first look at Facebook's lobbying dollars relative to others.

(source: Recode)

Meanwhile, competitor Google spent $18 million last year on lobbying, with Amazon spending $12.8 million. But Google and Amazon are working on products that could change industry, and need a nod from regulators to allow them to develop these products. Such products include Google's Waymo autonomous driving division. Facebook on the other hand, hasn't created some new product, and has been hit with PR nightmare after PR nightmare. If anything, Facebook's lobbying spend will probably increase again this year, with more and more PR issues mounting against Facebook ever since the Cambridge Analytica scandal.

(source: The Wrap)

Even if we look to the EU and GDPR, there is no adverse affect on Facebook's monetization path as much as there is an effect on Facebook's user count in Europe. Lobbying, as far as I know, is legal in Europe. And while I don't have the numbers for how much Facebook is lobbying in the EU, I believe Facebook could have some form of lobbying efforts. That being said, when you turn to small businesses, you see that small businesses (for the most part) are putting their marketing budgets into internet platforms like Google and Facebook. This is because of the low cost and elevated ROI these platforms deliver to users. If regulators crack down significantly on Facebook's monetization efforts, then small businesses (a key running point) are adversely effected. Because of that, the likelihood of regulators, at least in the US, implementing some form of meaningful regulation on Facebook are unlikely.

Another point I would like to make about European regulation, is the fact that regulators have done very little to hurt Facebook's business. The EU has slapped fines that are minuscule to Facebook's overall revenue stream and that are one time occurrences. And with GDPR, no real changes are made to the way Facebook operates in Europe. Simply, Facebook provides greater clarification to its users about what Facebook is doing with it's users data, and tells them in a down-to-earth and simple way. This is why we get less users from Europe, but the continuation of strong ARPU.

Until Facebook directly and adversely effects the politicians regulating Facebook, do not expect much change coming to Facebook's business model.

Amazon Isn't A Threat, Yet

A key argument that has been promoted by the bears on both Facebook and Google, but particularly with Facebook, is that Amazon is going to take market share from Facebook and Google in the ads space with Amazon's new advertisement offerings. In many surveys, some marketers actually have allocated some of their budgets to Amazon and away from Google/Facebook. However, there are a few reasons that Amazon's ads business will not take share from Facebook/Google, yet.

First of all, Amazon's internet advertising business is built around ads served on the Amazon site, which limits these ads to consumer goods. Amazon's advertisement platform is niche. The real problem is, how will Amazon move from purely consumer goods sold directly through Amazon, to the entire ads space encompassing all different products? It will be difficult, and a roadblock that Amazon will get over. But it is all in due time.

Secondly, Amazon's ad services are new, with no proof thus far that this business can generate higher ROI than Facebook/Google with competitive pricing. Amazon legitimately be a threat to Facebook/Google in ads, but they are yet to demonstrate the ROI that would allow Amazon to take meaningful share from the Facebook/Google duopoly. Until Amazon does that, it is not a real threat.

Core Facebook Targets

For this DCF model, I break up the Facebook business into Instagram and Facebook. I will start by giving my five year targets for core Facebook, and then give my targets for Instagram later. If you are wondering about the lack of WhatsApp and other growth initiatives, it is simply because any target with other businesses is much more arbitrary.

Facebook calculates ARPU based on its MAU number, not its DAU number. In general, I expect a continuation of the current trend, with Asia and ROW to continue MAU growth, and Europe/North America to slowly fade. As you can see however, I expect gains in ARPU/monetization despite user loss. This model assumes a revenue CAGR of 7.8%, an attempt to express how core Facebook's slowing growth days are ahead of it.

Facebook Is A Growth Company

As I showed above, I do not believe the core Facebook platform is itself a rapidly growing revenue generator. As a matter of fact, I believe the current slowing user growth numbers that Facebook has delivered over the last few quarters will be a recurring theme over the years to come, with continued deceleration in user growth ahead. And even then, a revenue CAGR of 8% isn't exactly a growth number. That being said, Facebook is a growth company. The growth from Facebook will come from Facebook's suite of platforms.

(source: Business Insider)

The Facebook business has leapfrogged from a slowly growing and again platform, to having one of the richest online ecosystems on the planet. Facebook will play an integral part in the development of technology for years to come. This is based solely on Facebook's relevancy in the tech space. Let's look at each of these avenues.

Instagram: First up is Instagram. Instagram has become an increasingly relevant platform in society. In my opinion, Instagram is the most important part of Facebook's growth story. Instagram is a very relevant platform, with 400 million DAUs of Instagram's Stories feature, and 1 billion MAUs.

(source: BTIG)

Ever since Instagram copied Snapchat's hit "Stories" feature, it has gained enormous traction, with Instagram even pulling users away from the platform that created the product, Snapchat. Stories have become so relevant amongst Instagram users, that Facebook is in the middle of a slow transition from the core Newsfeed to Stories, a transition that has led to slower than anticipated revenue growth rates. Revenue growth has indeed slowed, but Stories are setting up to be the next Newsfeed in the mobile frontier. While Wall Street sell-side analysts freak out about Facebook's decelerating revenue growth rate, they are failing to gauge the context of this deceleration. It is only inevitable that advertisers switch to the higher ROI and stickier mediums. This new medium is Stories, and Facebook is making the transition into this arena.

The best part about Instagram is that Instagram is yet to make any meaningful moves in monetization, particularly with advertisements. Eventually, Instagram will be so large and impactful on Facebook's overall revenue, that Instagram will break out the user and monetization data. Until that time however, there is a certain randomness to assumptions about Instagram's business. There is plenty of room for Instagram to run, with Instagram taking young users that leave the Facebook and Snapchat platforms.

But the real reason that Instagram has such a strong future is the lack of a competitive threat on the horizon. Instagram's greatest and most relevant competitor has been Snapchat. But Instagram has proven time after time that Instagram just beats Snapchat. Here is how it works: Snapchat will spend hundreds of millions of dollars per year in R&D and build some new product. If this product proves to be a hit, then Instagram comes along and copies it, and gains more traction on the Instagram platform at Snap's expense. And because Snap doesn't have this software legally protected, Instagram can walk away with millions of dollars of software for free.

The key mistake Snap has made that Instagram hasn't made is about listening to the user base. When many users are annoyed with a new feature on Instagram, the team at Instagram fixes it right away. Whereas the Snap team has believe that Snap knows more about what is right for the platform than the user does. If the user doesn't like the experience, there is no rule saying that they have to continue using the platform. Instagram has worked for users, Snapchat has not.

As such, it is not a surprise that Snap's user base has been shrinking.

(source: Snap IR)

And while I am no longer short Snap, my entire short thesis and short position was based around the belief that Instagram would take users from Snap. Sure enough, Snap has been losing users.

There is no denying it. Facebook's Instagram is increasing in relevancy and is becoming one of the most important platforms across all age groups and demographics. Facebook may eventually ARPU near the levels that Facebook is at.

The following ARPU data is based on the number of DAUs instead of MAUs. If I estimated Instagram's MAUs, then ARPU would be much higher.

Again, it is difficult to do an in depth breakdown of Instagram because the data points are scarce, being updated every few quarters instead of every quarter. As Instagram plays a greater role in the overall Facebook business, we should expect Facebook to break out Instagram in its earnings releases. Next you have Facebook's messaging platforms, WhatsApp/Facebook Messenger.

WhatsApp/Messenger: To be clear, I am not pricing in any impact from products outside of Facebook and Instagram in my DCF model. In my opinion, estimating users and ARPU for platforms like WhatsApp, and revenue for Oculus is almost completely arbitrary. As such, I'm only providing my analysis on this platform.

Thus far, Facebook's messaging apps have reached significant scale, but nearly no monetization. This is to be expected however, as monetizing chat based platforms is more difficult. Showing advertisements via chat just doesn't work, with Snapchat acting as living proof. I still believe however that the messaging apps have plenty of monetization runways ahead of it. First of all, WhatsApp is not completely a chat app. It also holds its own Stories-esque feature known as WhatsApp Status. But advertising isn't the only monetization runway that Facebook has with regards to its messaging platforms. Facebook has created a credit tracking program installed into Facebook Messenger, and has an opportunity to expand into the mobile payments business, a business that Facebook has said they are interested in for some time. And while the monetization of these services is set to be far lower then the core advertisements placings in Instagram and Facebook, the sheer scale of the user base should boost drive revenue growth from this segment of Facebook's business.

I don't break out this segment of Facebook's business, solely because of a lack of data on monetization and user growth. There is also potential for this monetization initiative to be a total flop, as Snap has had extreme difficulty monetizing its mostly chat based user base.

Overall, Facebook's growth trajectory will come from these two products, but further growth could come from products like eCommerce, Oculus (NYSE:VR), Facebook's AI tools, original content, smart speakers, and IGTV. Facebook, while projected as a company that is taking society backwards, is actually investing heavily in R&D and future innovative projects, mostly pertaining to software development. To believe Facebook is standing still, or even going backwards is a key flaw that could spell ruin for the bear thesis.

Valuation

For this valuation (and basically every valuation I perform), I will be using a discounted cash flow model. Currently, Facebook is debt free, and has $41.21 billion sitting in cash as of fiscal Q3. Facebook ginormous cash pile is another key reason I own the stock. On an ex-cash basis, Facebook trades at ~17X 2019 expected earnings, with 24.7% revenue growth. If you value Facebook at a PEG of 1 on consensus EPS and growth rates, then Facebook stock would be valued at ~$184. The stock is at ~$144. So using a simple PEG ratio gets me to a target of 28% upside. But the discounted cash flow model spells upside greater than 28%.

Here are my calculations for cost of equity:

This assumes an equity risk premium of 5.32% against the 10 year bond's yield of 3.114%. Facebook's one year unlevered beta is 1.39.

Because Facebook lacks debt and a cost of debt, the cost of equity is equal to the WACC. So Facebook's WACC is 10.51%. This is further verified, as Credit Suisse analyst Stephen Ju uses a WACC of 10.5% in his DCF valuation of Facebook.

Here are the combined estimates for Facebook's business over the next five years:

To value the stock, I'm using a terminal growth rate of 3%. Here is the valuation of the stock:

Conclusion

The current market price of Facebook's stock more than bakes in the risk associated with Facebook. The company's stock trades at a seemingly value multiple, while being a growth business. The headwinds forecasted by bears are irrational. Eventually, when Facebook overcomes these fears, the market will realize Facebook for what it is, a growth giant with a portfolio of some of the most valuable digital assets in existence. Coupled with strong financials, improving user count, and a long-term positive trend in ARPU, the bears are just plain wrong. Facebook remains a buy.

Disclosure: I am/we are long FB, GOOG.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Additional disclosure: I am not a financial adviser. This is not financial advice. Everything said here is my personal opinion. Please do your own due diligence with regards to investments in these securities.

LAGUNA BEACH, Calif.—The head of Alphabet Inc.’s Waymo unit said it plans to launch its first commercial self-driving car service in the next two months and expects businesses to be among its biggest customers.

Speaking at The Wall Street Journal’s WSJ Tech D.Live conference on Tuesday, Waymo’s John Krafcik said the new service will charge individual passengers for rides as well as businesses, such as Walmart Inc., who want to pay to shuttle their customers to stores. The service will initially be available to a small group...