Microsoft Investment Thesis

Published on 04/28/21 | Saurav Sen | 5,452 Words

The BuyGist:

  • This thesis was originally posted on May 29th, 2018. 
  • We go through the company's competitive advantage, durability of competitive advantage, and its management quality. 
  • Since we wrote this, a lot has changed. We regularly revisit our thesis to see what we got right and what we got wrong.

Thesis Summary

Reorienting the organization to become “the productivity company” to play the right sport: Cloud/AI/IoT/Edge. It has a competitive advantage in most of that “stack”. Relative lack of expertise in AI, however, could be the thesis buster.

Competitive Advantage: The Castle

  • Core Competency? Bona Fide software developer – specializes in Productivity software.
  • Better or Cheaper? Veering towards Better. Seamlessness in Cloud/AI/IoT/Edge.
  • Evidence: Historical Profitability? Revenue & EBITDA growing fast. Harnessing Cloud tailwind.

Durability of Competitive Advantage: The Moat

  • Competition? Heavy. Amazon and Google in Cloud. Also, IBM in AI. And PlayStation.
  • Protection? Familiarity & lock-ins with Windows users. And a new Services Architecture.
  • Resiliency of Cash Flows? More resilient now – moving towards subscription models.

Management Quality: The Generals

  • Strategy and Action? Clear, decisive move towards becoming a Cloud/AI/IoT/Edge powerhouse.
  • Alignment of Incentives? Good. Significant portion of incentive-comp. is stock-based.
  • Financial Productivity? High Returns on Capital – higher with new Cloud/AI business.
  • Sustainable Free Cash Flow? About $49 billion. Translates to roughly $128/share.

Why Microsoft?

My worldview on Retail made it obligatory to dig into the Amazon story. And then, as I dug into it, it turned out that Amazon’s base business (in terms of cash flows) is now their Cloud business. And then it turns out that in the Cloud Infrastructure space, their biggest competitor is Microsoft. But they had turned up as the primary competitor in a closely related realm: Artificial Intelligence for Enterprises. As I had started digging into AI, I had concluded that in the Enterprise space – in dealing with companies as opposed to consumers – IBM and Microsoft had a leg up. This was partly because of their focus on that part of the market and partly because of entrenched relationships with big companies. I had focused on IBM at that point because it was trading at a much more reasonable valuation than Microsoft. Now that it’s clear that Microsoft takes up prime real estate at the junction of these two intertwined technologies – Cloud and AI – it’s hard for me to ignore them anymore.

In the analysis below, I will look at Microsoft and its competitive landscape through the lens of “the new information revolution”, which is the confluence of Cloud-AI-IoT-5G-Edge. There is little doubt that it’s where the world is going. And there is no doubt, that it is the sport Microsoft has chosen to play.

Competitive Advantage: The Castle

Core Competency? Bona Fide software developer – specializes in Productivity software.

I’ll start on a negative note (I used to specialize in bonds). Let’s be real. Most people use Microsoft Windows because of Microsoft Office – Word, Excel, and PowerPoint. In recent years, it appears that Microsoft beefed up SQL for databases (I remember it being not-so-robust about 10 years ago). But for many years, we all tolerated Windows in our offices and in our homes because of familiarity with Windows and the inherent MS Office lock-in. The frequent crashes, the virus attacks, and the somewhat boring interface tested our patience. As I write this on a Microsoft Surface with Windows 10, the incessant, uninvited forced-reboots (caused by some obscure “updates”) test my patience every couple of weeks. Despite that, I believe there is hope.

Microsoft has made a conscious decision to move away from being “the Windows company” to now marketing itself as the “Productivity” company. It may sound like BS but there is some merit to it. To be fair, MS Office – for all its faults – offers the best word processing and spreadsheet software. That’s a significant chunk of work that gets done in any office around the world. Microsoft has some other tools, like SharePoint and OneNote (which I love), which presumably enhance productivity. But it’s clear that with Word, Excel and PowerPoint leading the charge, Microsoft has the right to claim that they have an unmatched legacy in “productivity software”. All this productivity stuff has been delivered via Windows – a product that most people use but few actually like. With the Cloud movement, Windows probably won’t be the primary delivery package for productivity. That’s a good thing.

In “the cloud”, there are three primary types of services: Infrastructure, Platform, and Software. In infrastructure, Microsoft was a fast follower after Amazon (AWS). But it’s been gaining ground, in no small part due to the overall secular growth in Cloud Computing. In recent quarters, Microsoft seems to be catching up with AWS in terms of new Cloud adoption. Why has that happened? I think it’s because Microsoft’s core competency is software. It’s one thing to put all your stuff on a Cloud (the infrastructure bit). It’s quite another thing to do useful stuff in the Cloud – that’s the software and platform part of it. That’s Microsoft’s jam.

Microsoft’s domination in “productivity software”, which has lasted several decades, has created inherent lock-ins amongst the primary users of Cloud Computing – Enterprises. Many (if not most) work processes are built around World, Excel, PowerPoint, SQL, SharePoint and other (usually Microsoft) applications. For a company to transfer mission-critical work processes from it’s internal servers to a Cloud is a painful process as it is. Life becomes a lot easier for a company’s IT staff and the end-users if that process gets transferred undisturbed and uncorrupted. To do that transfer, Microsoft with its fingerprints all over these work processes already, has a tangible advantage. It is probably because of this that Microsoft is the clear leader in “Hybrid Cloud”, which is a set up that many companies prefer. Hybrid Cloud refers to a set up where a company uses both a Private Cloud and a Public Cloud to house it’s work and data.

Speaking of Data, it is the new Oil. Actually, that’s IBM’s terminology, but Microsoft makes a similar point. CEO Nadella says something to the effect of “data is the new currency”. Microsoft has a tangible legacy in data analysis via Excel and SQL. I can hear the rumblings of data scientists – Excel and SQL aren’t that robust for serious data analysis. True, but for most firms, it’s good enough. At a minimum, it’s a means to “deliver” analysis and insights to the end-user. No CEO or line-manager is going to open up a table or chart using R or Matlab or Oracle DB. They want to see insights in software they know. But the bigger reason is that with Excel and SQL, Microsoft already has a much more tangible Data Analysis legacy than its competitors in the New Information Revolution realm. Oracle is the premier database company, and it is trying to compete in the Cloud space, but it doesn’t have the usability and interface edge that Microsoft has. Amazon AWS has launched a product called Aurora, which happens to be a SQL-based database application. SQL is a Microsoft sport. Microsoft’s answer to Aurora is called Cosmos DB. I know nothing about both these DBs, but I would guess that Microsoft has an advantage here just based on legacy.

While Oracle’s big advantage in databases is customizability and robustness, I’m not sure how much it has invested in AI. Data and databases are all well and good, but most businesses will want insights and a pretty interface. Excel with SQL and AI is where I see Microsoft providing some advantages over the likes of Oracle. In data analysis – which is becoming the central work process at most businesses – Microsoft just has more tools to offer. Whether they are robust enough or good enough is debatable. But I believe that the combination of tools at Microsoft – born from it’s “productivity software” legacy – will be greater than the sum of its parts.

Better or Cheaper? Veering towards Better. Seamlessness in Cloud/AI/IoT/Edge.

Phasing out of being a “Windows” company to becoming a “productivity” company is a good move – both from a substance and a marketing standpoint.

On Marketing – I’ve touched on this – Microsoft is better off without the “Windows” baggage. I won’t go on a Windows-bashing diatribe because, objectively speaking, it is incredible technology. But with its periodic crashes and forced-reboots, in my mind it conjures up a joke I heard from the now-infamous comedian Louis CK. The joke was something like this: “So, I was at CVS the other day…because, you know, sometimes you have to…in fact, that should be their slogan, ‘CVS – because sometimes you have to’”. That’s what, I think, Windows conjures up in most peoples’ minds: “Windows – because sometimes you have to”.

Microsoft now sells something called “Microsoft 365”, which is an all-encompassing software package that houses most of the Microsoft applications we use today. This is a clear signal of its intention to move beyond Windows. Microsoft 365 is a moniker borrowed from the hugely successful Office 365, which I use. I like the subscription model of pulling up software and disk space from the Cloud, so I never lose my work. And the software packages are infused with new features every now and then, with no intervention. It appears that with Microsoft 365, we’ll soon be able to run our operating system as a Platform-as-a-Service (PaaS) on Azure, which is Microsoft’s Cloud. We may see a day in the not-too-distant-future when we don’t need Windows. I hope, however, that Microsoft 365 is more crash-resistant than Windows. And that it doesn’t force a restart on us willy-nilly (wow – it’s hard to write an objective thesis on Microsoft while using Windows). This brings me to the “substance” argument.

Based on the past couple of years of quarterly earnings calls, Microsoft is all-in on what I call the New Information Revolution. That’s Cloud/AI/IoT/5G/Edge. That’s the full “stack” of this new paradigm shift in computing. Microsoft calls it “Productivity”. And they plan on offering a seamless productivity experience that spans the entire “stack”.

When I say “stack” (which is an industry term) in the context of “productivity” or “the new information revolution”, I’m referring to the broad architectural layout that enables this new revolution. I did my best to illustrate it here:

I apologize if the stack looks too dry. The salient point is this: Microsoft seems to be the only company in the world that is competent in every component of that stack. In some layers, they’re better than their competition. I’ll discuss this competitive landscape in more detail in the Competition section but here is the gist:

  1. In Hardware/Processing Power: Microsoft has been a frontrunner in investing in new Chip technologies specifically designed for AI.
  2. In Databases: I mentioned this earlier, but Microsoft has a couple of decades of experience with SQL, which happens to be the basis of the database software offered by Amazon on AWS.
  3. In Software: Microsoft is a bona fide software developer. As you may have surmised by now, I don’t think all their software is great. But in the productivity arena, they’ve clearly dominated.
  4. In the Edge space: this is an interesting edge (pun intended) they have. One of Microsoft’s greatest successes in the past couple of years has been the Surface device. This was a rare moment in the company’s history when they led Apple in launching a product – a device that seamlessly moves between desktop mode and tablet mode with no hiccups. Traditionally, products have never been Microsoft’s jam. Remember Zune? But this time with the Surface, they got it right within a couple of iterations. I believe they now have the expertise and the design sense to develop more Edge devices (like the cheap tablet they will launch soon, or the Surface Hub, which is insanely cool) to round out their “end-to-end” offering in the new information revolution.

There is another Microsoft product that I think will give them an edge (I couldn’t resist the pun again). Check out the HoloLens. At the moment, they’re marketing it as an Augmented Reality tool for the factory floor or the design studio. But I think the implications will be far greater. Imagine gaming with it or using it to see patterns in a crime scene. The possibilities are endless.

The point is that Microsoft has its fingers in each part of the “productivity” stack. The challenge for them is to make it all seamless and offer a cohesive productivity product that binds together each component of the stack to deliver an experience that’s greater than the sum of its parts – without the baggage of Windows. Microsoft is half way there. And they’re in a better position to deliver “the complete stack” than any other company in the world.

Evidence: Historical Profitability? Revenue & EBITDA growing fast. Harnessing Cloud tailwind.

In the last reported quarter, Microsoft said that revenue attributable to Azure grew about 93% year-on-year. That’s incredible. But it seems their overall “Intelligent Cloud” segment grew just 17%. When I say “just”, I mean it pales in comparison to the reported growth in Azure. That’s because Intelligent Cloud also includes other services such as the SQL server business.

Apart from Intelligent Cloud, the other 2 reported segments are “Productivity and Business Processes” and “More Personal Computing”. The Productivity segment includes Office 365 and Dynamics 365 (their CRM system competing with Salesforce). The Personal Computing segment includes Windows licenses, the Surface device, and the Xbox segment. All segments have grown impressively over the last couple of years. The difference, I believe, is the tactical shift to move everything to the Cloud, and an increasingly subscription-based revenue model. These moves have especially improved margins.

By my quick math, Microsoft’s EBITDA margin has improved from 31.3% in 2016 to 38.7% over the last 12 months ending March 31, 2018. That’s a big jump. As usage of Azure – Microsoft’s Cloud – moves more “up the stack” from simply using memory to using complicated AI-infused software programs on an Edge device, revenues and margins should improve. In fact, I think that margins on Infrastructure-as-a-Service will decrease, as it gets more commoditized. The money is in Software that will be AI-infused.

What’s amazing is that we’re still in the early innings of Cloud migration. As of January 2018, McKinsey said that just 40% of companies they surveyed had more than 10% of their workloads on the Cloud. Even if we assume that about 10% of all workloads of all companies have migrated to the Cloud, there is significant room for growth. Of course, Microsoft must fight Amazon and Google for a significant share of that remaining 90%. With it’s expertise and experience in platforms, software and productivity tools, it should be able to grab its fair share.

Durability of Competitive Advantage: The Moat

Competition? Heavy. Amazon and Google in Cloud. Also, IBM in AI. And PlayStation.

The big boys in the Cloud space are Amazon, Microsoft and Google. No surprises there. But as I mentioned before, that refers mostly to the Public Infrastructure-as-a-Service part of the stack. It’s too early to choose winners in Software at this time. Being the only bona fide software developer in this trifecta, Microsoft should have a tangible advantage. But it’s not so simple.

In my view, the main differentiator in software on the Cloud will be AI. And I believe the main differentiator in AI will be the quality of code AND the quality of data. This is where Microsoft is at a disadvantage – in AI. I have no doubt that Microsoft has some kick-ass programmers (maybe not in Windows). I can buy the argument that that their Machine Learning code is right up there with the best. But how are they training their AI? What kind of data do they have? Are they acquiring any? I don’t have a definitive answer to these questions.

Google has made a giant company out of making sense of unstructured data. From a quality of code standpoint, I would bet on them above anyone else. But they also have a tremendous amount of data on various facets of our lives. That’s creepy but it’s the truth. Many of us use Google Maps, for example. That amount of data collected over years is extremely powerful fodder for an AI system that aids some sort of navigation.

Amazon – I have argued – will likely have a massive competitive advantage in anything to do with Retail. They already have 20 years-worth of data to train Lex – the AI engine that powers Alexa (their AI assistant) among other things. Even Google can’t match that amount of Retail data.

One of Microsoft’s partners from the 1980s – IBM – is also a player in the AI/Cloud realm. It hasn’t been as effective in securing clients for its public IaaS Cloud offering, but it is undoubtedly a big player in AI. Watson, after all, was the first mover. And it’s been collecting data since. I have argued that Watson may have much better training data than Amazon or Google in certain fields like Health and Finance. IBM has been acquiring data for a long time.

My argument is that it’s easy for me to hold on to a concept when I try to visualize Amazon, Google or IBM’s AI capabilities. It’s hard for me to do that with Microsoft. Is it enough for Microsoft to offer some generic machine learning code?

In gaming, Microsoft has never been able to beat PlayStation at their own game. To be fair, Microsoft has carved out a niche for themselves. Still, they’re the second fiddle. And that brings me to another point.

Microsoft has a history of playing second fiddle. In Productivity Software, however, they were fast followers and clearly made the better product. Excel beat Lotus 123 convincingly. But in other products, they followed their competition with a product that wasn’t convincingly better than the original. These days they’re promoting a product called Dynamics 365, which seems to be a “me too” version of Salesforce. I’m not sure how much they’ll be able to crack that market. And that’s really my biggest concern with Microsoft – there are too many “me too” products in their stable.

With Surface, however, they led. And with Azure, the market seems to think that they offer something as good as AWS.

Protection? Familiarity & lock-ins with Windows users. And a new Services Architecture.

Microsoft’s most potent defense against competition is a combination of 3 factors:

  1. Windows lock-in
  2. “Full-Stack” architecture
  3. Large Developer Network

The combination should result in a Cloud platform that’s app-rich and has an air of familiarity. Only Google comes close to providing this combination – they now have experience in developing Operating Systems (Android). They’ve got significant lock-ins as well with the Google ecosystem – Gmail, G-suite, Search etc. But at the end of the day, they are still primarily an Advertising company. Microsoft has no such distractions. They are now exclusively in the “productivity” business.

About going all-in across the Cloud stack – from Hardware to Databases to AI-infused software to Edge Devices – Microsoft’s commitment is probably most evident in its bet on FPGA chips which are designed specifically for neural-network-programming. It’s clear that they are constantly thinking about the entire stack of Cloud Services. CEO Nadella makes that point in every earnings call off-late. He calls it Microsoft’s “architectural advantage”. But I’ll settle for “walking the talk” at this point.

If Microsoft is able to execute that seamlessness throughout the stack, as I mentioned in the Competitive Advantage section, it’ll snowball in terms of users and volume of apps. Imagine a situation in which a firm can use complicated AI data analysis tools and export the insights to Excel to email it to a client, all within a few seconds. Some of this is already happening, but you can see that an end-to-end architecture that’s compatible and seamless will make this job much easier. That seems to be Microsoft’s vision – use its worst product (and ironically its most ubiquitous) as a captive audience to promote its new end-to-end solution for the data explosion era in which we live. A captive Windows audience provides a critical mass of Microsoft Azure adopters, which can tap into an already large network of developers that exist because of Windows. That large group of developers can create tools that Windows-dependent companies can use to analyze their data. More developers would do that if there is a sizeable market for it. That’s the snowball effect that will act as a wide Moat that almost nobody can cross. All this, of course, hinges on Microsoft’s ability to program an effective AI system. That’s where I have some reservations. Maybe they could start by calling it something?

Resiliency of Cash Flows? More resilient now – moving towards subscription models.

Moving away from Windows may turn out to be one of Microsoft’s best decisions. With it’s end-to-end Cloud stack, Microsoft can now deploy productivity tools without being Windows-bound. I mentioned that they are now moving slowly to something called Microsoft 365, which will run on Azure. This will eventually replace the concept of an Operating System as we know it today.

This new architecture allows Microsoft to sell everything on the stack on a subscription basis. This portends an unusually fortuitous time for Microsoft. Not only are they capturing their fair share of the secular growth in cloud computing, but they’re also moving rapidly towards a subscription-based model. Given its expertise and focus on productivity software, margins should remain high. EBITDA margins are already high at almost 40%. Revenues are growing, they’re stickier now, and margins are improving – all at the same time. I can’t see this momentum stopping for the next few years unless their AI logic is just terrible.

Management Quality: The Generals

Strategy and Action? Clear, decisive move towards becoming a Cloud/AI/IoT/Edge powerhouse.

I like companies that are decisive and take a stand on something. Companies that play it safe by being everything to everyone usually end up being nothing to no one. Microsoft, under Satya Nadella, has taken a stand. They’ve put all their chips into participating in and creating a world in which Cloud Computing along with AI will do its thing on multiple devices, seamlessly. Yes, Amazon, Google and IBM have made similar bets. But Microsoft is the only bona fide software maker who knows a thing or two about graphical user interfaces. Sure, they don’t’ have Apple’s finesse, but I must give them points for improving over the last few years.

Speaking of the last few years, under Nadella Microsoft has done something they had never really done before: invent a new product and lead the market. There are two such cases: The Surface and the new FPGA chips.

With the Surface, I really think Microsoft has set up a solid platform for “Edge” devices. Nadella has mentioned what he calls “decentralized computing” several times in the past few earnings calls. This is a massive advantage over its Cloud/AI competitors: Amazon, Google and IBM. None of them have demonstrated that they are top notch device companies.

On FPGA chips, Microsoft seems to be the only one of the 4 Cloud/AI companies that has been decisive about this bet. Google manufactured its own chip for AI – the TPU – while the others still use some form of GPUs (Graphical Processing Units). I’m not going to get into the nuts-and-bolts of semiconductors, but its good enough to know that FPGAs have some of the intense parallel computing power of GPUs, while retaining some of the customization strengths of ASICs. Microsoft seems to think that FPGAs offer the best of both worlds. Time will tell. But I like that they are not taking for granted the status quo in the industry (GPUs). They’re taking a stand, and going all-in. This is a very different Microsoft from the fast-follower company we used to know. I think it’s because they have a deeper understanding of what they want to be.

Alignment of Incentives? Good. Significant portion of incentive-comp. is stock-based.

I found Microsoft’s compensation plan to be very granular – longer term compensation for executives were linked to things like “Surface Margins” and “Commercial Cloud Subscribers” among others, which is quite specific. I won’t go into every detail but there is one that I found interesting from what I heard in the earnings calls, which was corroborated somewhat by the Proxy Report. It seems that Microsoft pays its sales staff partially based on “usage” as opposed to just volume of sales. There’s a difference. The company wants customers to not just buy more Microsoft Services, but to also use more of it. This aligns with revenue generation, margin improvement, and free cash flow growth. It also allows Microsoft to experiment more with pay-as-you-go models for its Cloud-AI offerings, as it competes with Amazon and Google going forward.

All in all, I don’t have any complaints with Microsoft’s compensation plans.

Financial Productivity? High Returns on Capital – higher with new Cloud/AI business.

Microsoft is a high ROE/ROIC business, no matter which way you calculate it. By my estimate, it’s easily north of 20% ROE. The Debt/Equity ratio is also lower than 1x. All this tells me that Management is doing a good job of allocating capital, which is their main job.

As I mentioned before (and it’s worth mentioning again) – Microsoft is going through a particularly fortuitous time in which revenue, margins and free cash flow are all growing in tandem. I’m sure luck has a lot to do with it, but it’s really the bet on the “new information revolution” that Management has made that has put the company in this position.

There is an inherent snowball effect in this Cloud/AI business. First of all, it takes scale to be successful. If a company can’t provide enough datacenters, security and tools on the Cloud, why would anyone use it? So, the incumbents, Amazon, Microsoft and Google, have the biggest advantages. Big initial investments are needed. And then scale perpetuates scale. Bigger scale allows for cheap pricing models on Infrastructure-as-as-Service, which creates more users, which motivates more software and AI tools. All that attracts more users, and the virtuous cycle continues. Amazon has experienced this, which is why their Cloud service AWS is their biggest money-maker now. Numbers suggest, Microsoft is experiencing a similar boost. It’s not surprising.

Sustainable Free Cash Flow? About $49 billion. Translates to roughly $128/share.

The $49 billion Free Cash Flow number is a lot higher than Microsoft’s number now. I estimate Free Cash Flow for the last 12 months to be at $30 billion. But the growth rate – for factors discussed above – has been very high. To put it in context, in 2016, Free Cash Flow was about $11 billion. To give it some more color, Revenue grew from about $83 billion to about $103 during the same period. It would seem that all of that revenue trickled down to Free Cash Flow. Not exactly – while revenue increased, so did EBITDA margins, from roughly 31% to 39%. This is that virtuous loop in action.

In estimating the “sustainable” level of Free Cash Flow, I made the following assumptions:

  1. All three revenue segments grow, in the following way:
    1. Productivity Software and Business Processes grows by 20% - I assume more growth in Office and other software products (maybe even Dynamics 365 if current trends continue).
    2. Intelligent Cloud Revenue doubles – for reasons discussed above, I see this space growing exponentially over the next few years. Competition is limited to Amazon, Google and IBM, all of whom will experience high growth in whatever version of Cloud/AI they offer. The risk in this projection is: Microsoft’s AI turns out to be a lemon.
    3. More Personal Computing grows another 20% - I really think that with the Surface, they’ve got something to work with. Edge computing hasn’t even really begun, and Microsoft is already designing its backend systems to meet the needs of “decentralized computing”.
  2. EBITDA margins keep improving and stabilize at 42%. It’s at about 39% now. Most of the costs are fixed; as revenue grows, margins should improve.

Capital Expenditure remains around current levels. It seems to me that the investments they’re making now are enough to sustain their competitive advantage.

We use cookies on this site to ensure the best service possible.