Why Microsoft Is A Sell

About: Microsoft Corporation (MSFT)
by: Phil Anthropy

New tax law brings heavy expenses for years.

Cloud computing has no moat.

Windows 10 has quality and performance issues.

Microsoft is late to the artificial intelligence (AI) party.

Despite good recent performance, growth may decelerate.


As someone who has used every version of Windows from 3.0 to 10, I write this article with a degree of regret. Microsoft (MSFT) has been a mainstay of user computing for decades, but empires come and go. In my view MSFT's long-term success is questionable.

I had previously written a very different type of article, describing MSFT's history in depth, evaluating the evolution of Windows and critiquing the company, but that was deemed off track for investors. This article concentrates on the business fundamentals.

The company's recent performance has been admirable, with double digit revenue growth. Why, then, a sell recommendation? It's because I don't think this level of growth can continue.

"Software as a service" (SaaS) is the new watchword for business computing, as is "infrastructure as a service" (IaaS). This is simply a reinvention - albeit a vast improvement - of the decades-old concept of timesharing, where users run software on remote servers or mainframes.

As long as communications links remain fast and reliable, remote computing can be more cost-effective than managing local hardware and an operations staff. However, remote computing cannibalizes money from the purchase of hardware/software for local use, so for new MSFT cloud customers, it can represent a partial revenue shift rather than a revenue increment.

The cloud has no moat. There are many competitors, including Alibaba (BABA), Amazon (AMZN), Google (GOOGL) and IBM (NYSE:IBM) among others. Most are companies with deep pockets. MSFT's principal advantages at this stage are single-vendor service and specific expertise. The specific expertise stems from MSFT being in the best position to support its own Windows operating system (OS) software, which dominates the market.

These advantages are substantial for now, but in the long run MSFT's captive audience will erode, in my view. Competitors are likely to have (or to develop) largely equivalent software, a cheaper cloud and better artificial intelligence (AI). Kubernetes, described later, makes the cloud a commodity.

Microsoft Business Structure.

MSFT's Annual Report for 2018 (let's call it AR18) p. 10 sets forth the business strategy:

To achieve our vision, our research and development efforts focus on three interconnected ambitions:

• Reinvent productivity and business processes.

• Build the intelligent cloud platform.

• Create more personal computing.

This sounds reasonable. Essentially it means developing more integrated software packages for business use, selling a lot of cloud computing, proliferating Windows 10 and competing in the tablet space with Surface tablets.

But where is the moat? Other vendors have automated business process software and cloud computing. PCs with Windows may fall out of favor once cloud computing through smart phones takes hold, which I believe will happen in the next few years.

MSFT has three major business segments, corresponding to its three targets of research, mentioned above:

  • Productivity and Business Processes (PBP)
  • Intelligent Cloud (IC)
  • More Personal Computing (MPC)

MSFT's PBP has a rich inventory of software, but much of it is not unique. Increasingly, open source products are competing. OpenOffice is free, and has substantially similar functionality to Microsoft Office. Although a single vendor approach to software is desirable for many businesses to simplify management and security, cost may tip the scales in favor of a free product. As the large company market becomes saturated, MSFT's efforts to target smaller firms may run up against resistance due to cost considerations.

IC is a sound marketing strategy. AR18 doesn't provide granularity on the customer mix, which is important in predicting growth and retention.

MPC is just another way of saying "sell more end-user software." That has been MSFT's path to success for decades, but can the growth continue?

On the surface (so to speak), things look good.

According to AR18 p. 2, "In fiscal 2018, we delivered $110.4 billion in revenue and $35.1 billion in operating income and returned $21.5 billion to shareholders through dividends and share repurchases."

But there are storms ahead.

On AR18 p. 6 net income is shown as $16,572 million, which would have been much higher except for footnote a: "Includes a $13.7 billion net charge related to the Tax Cuts and Jobs Act, which decreased net income and diluted earnings per share ('EPS')"

More on this later.

First the Good News.

MSFT's most recent quarterly conference call was a celebration of success.

Office 365 Commercial now has 180 million users. Our EMS installed base reached 100 million. And the Outlook apps on iOS and Android surpassed more than 100 million users for the first time this quarter.... ...Windows 10 is now active on more than 800 million devices and continues to gain traction in the enterprise as the most secure and productive operating system....

This quarter, revenue was $30.6 billion, up 14% and 16% in constant currency. Gross margin dollars increased 16% and 18% in constant currency. Operating income increased 25% and 27% in constant currency. And earnings per share was $1.14, increasing 20% and 22% in constant currency.

Now the Bad News.

From AR18 p. 22 [Emphasis mine]:

On December 22, 2017, the Tax Cuts and Jobs Act (“TCJA”) was enacted into law, which significantly changes existing U.S. tax law and includes numerous provisions that affect our business. During fiscal year 2018, we recorded a net charge of $13.7 billion related to the TCJA. Refer to Note 13 – Income Taxes of the Notes to Financial Statements for further discussion.

Earnings are likely to remain sub-par for a number of years due to the new tax law. Here's a more detailed description from AR p. 73, Note 13 [Emphasis mine]:

On December 22, 2017, the TCJA was enacted into law, which significantly changes existing U.S. tax law and includes numerous provisions that affect our business, such as imposing a one-time transition tax on deemed repatriation of deferred foreign income, reducing the U.S. federal statutory tax rate, and adopting a territorial tax system. The TCJA required us to incur a one-time transition tax on deferred foreign income not previously subject to U.S. income tax at a rate of 15.5% for foreign cash and certain other net current assets, and 8% on the remaining income. The TCJA also reduced the U.S. federal statutory tax rate from 35% to 21% effective January 1, 2018. For fiscal year 2018, our blended U.S. federal statutory tax rate is 28.1%. This is the result of using the tax rate of 35% for the first and second quarter of fiscal year 2018 and the reduced tax rate of 21% for the third and fourth quarter of fiscal year 2018. The TCJA includes a provision to tax global intangible low-taxed income (“GILTI”) of foreign subsidiaries and a base erosion anti-abuse tax (“BEAT”) measure that taxes certain payments between a U.S. corporation and its foreign subsidiaries. The GILTI and BEAT provisions of the TCJA will be effective for us beginning July 1, 2018.

The taxes on overseas deferred income and assets should cut into MSFT's earnings for several years, following which the lower corporate tax will be a tailwind to greater profitability. However, for the next few years the higher taxes may limit not only MSFT's growth, but also its stock buybacks, which have contributed to the rise in the stock price. The question, then, is whether that rise will continue or whether we're approaching a top. In my view, it's the latter.

We see from Yahoo finance that MSFT's FY 2018 net income plummeted (from $25,539 million to $16,571 million, YoY) as a result of the tax expense; otherwise it would have been another year of substantial earnings increase. Forward P/E ratio was about 25, while PEG ratio was a mediocre 1.84. Without the new tax expense P/E and PEG would have been much lower, of course.

By comparison, Google/Alphabet (GOOGL) had net income nearly double that of MSFT, with much less tax expense as a percentage of profit than MSFT (a surprise). GOOGL's P/E ratio is about 22 and PEG is about 1.60. From afar, GOOGL looks like a slightly better choice than MSFT from a "fundamentals" standpoint, and GOOGL also has more money to grow with over the next few years, when cloud computing should really accelerate. Google is a much larger company than MSFT in terms of revenues, and scale is important when building cloud infrastructure. GOOGLE's net tangible assets (from balance sheet) are nearly four times as large as MSFT's.

Proprietary vs. Open Source.

The major moat for MSFT is its Windows OS, most recently Windows 10. That moat does remain intact, but I believe it will dry up eventually.

Windows has never been lightning fast. Each new version is larger and slower. Each new version has been plagued with bugs and security flaws necessitating frequent updates. In the past, the only alternatives to Windows were Linux/Unix and the Apple (AAPL) Mac OS, with Windows the clear winner, but that may change.

Smart phones and tablets now handle a significant proportion of the world's telecommunications. This doesn't mean that people no longer need laptops, but it may come to that, especially with gigabit Internet, 5G cell phones and cloud computing. Why bother with a laptop when a future smart phone with 64+ gb could attach to a display, keyboard, mouse and other peripherals through a combination of Bluetooth, wi-fi, USB and perhaps fiber? If SaaS IaaS in the cloud can supplement powerful apps on the smart phone or tablet, who cares about local hard drives and a Windows OS?

Moreover, a big chunk of new transaction-oriented software is browser-based and doesn't require a specific OS for that browser, nor a specific OS for the back end servers. Microsoft Server and Linux/Unix are already competitors.

Newly designed supercomputers may someday take over server functions, since massively parallel processing and specialized hardware will be needed to support advanced AI and speech handling, with limited preprocessing by the cell phone. VB.NET, a dialect of MSFT's mature and proven Visual Basic product, can be implemented on non-Windows Oss.

Windows 10 - Not Ready for Prime Time?

In my opinion Windows 10 is a clunky product with poor performance and alleged significant invasion of privacy. The poor performance derives principally from disk thrashing - where disk usage stays at 100% too much of the time - but also likely from dubious coding. The Internet is replete with complaints about Windows 10. Bloatware tasks come back later if they are ended, and some cannot be disabled. The OEM version of Windows 10, with its many invasive background tasks (Microsoft Compatibility Telemetry, Mobile Plans, Microsoft Store, Cortana, etc.) is bad news. It is user-unfriendly for laptops to revive from "sleep" without operator intervention, which wasn't the case before. Sudden changes in communications protocols make certain legacy software fail. Certain updates may fail to install, requiring extensive troubleshooting to remediate. It would take a separate article to enumerate the various issues.

Windows Server is better, without as much bloatware, but it has some of the same shortcomings. The "desktop experience" variant of Server is similar to Windows 10.

You might have noticed that recent update alert messages in Windows 10 say "Windows 10 is a service." You don't own the software. The Windows 10 user license agreement (eula) is restrictive, and could become even more so in the future. For example:

...You may not ... use the software as server software, for commercial hosting, make the software available for simultaneous use by multiple users over a network, install the software on a server and allow users to access it remotely, or install the software on a device for use only by remote users.

The eula goes on to limit remote access to only users with Windows 10 PCs, although in fairness it doesn't limit the number of CPU cores to two, as was done in Windows 7. There are also constraints on the sale or transfer of the license. It pays to read the eula.

In my view, the restrictions on Windows 10 in the eula are increasingly becoming a disincentive for its use. If there are other viable alternatives, many may switch.

Yet MSFT is hanging its future hat on Windows 10. AR18 p. 11 says:

Windows 10 is the cornerstone of our ambition, providing a foundation for the secure, modern workplace, and designed to foster innovation through rich and consistent experiences across the range of existing devices and entirely new device categories.

This is like saying that a Sherman tank will win the Daytona 500. There is no low-end laptop I know of that will run Windows 10 briskly, even for simple browsing with a half dozen tabs open. SSD disks do not solve the disk thrashing problem, because cheaper SSDs are little better than hard drives in terms of throughput, particularly during write operations.

Windows 10 would likely require a complete redesign and rewrite from the ground up in order to mitigate its numerous problems. Unfortunately, it must remain backward compatible to the extent possible, and that is a real performance problem. Its modular design, with dozens of background processes and services, and with dynamic linking, is flexible but inefficient. One solution is for someone to come up with a brand new operating system, possibly even an entirely new PC architecture that runs Windows via an emulator but has a superior native design. AAPL can already run Windows as a separate boot on a Mac, and the virtual machine concept is not new. Users deserve a system with better response time.

In my view, linking MSFT's future to Windows 10 - at least, in its current form - is a mistake. Admittedly, that OS is currently running on 800 million devices, up from 500 million in FY 2017 (see MSFT's 2017 Annual Report). However, one must wonder how much of that expansion was due to purchase of new PCs or laptops because Windows 10 ran too slowly on the old ones. That growth rate may not continue. In comparison, AAPL's IOS is running on well over two billion devices (see here). According to one source, there are 2.5 billion Android users. IOS and Android could be upgraded to greater functionality on future devices, with capabilities rivaling laptops, or else could easily access SaaS in the cloud. Windows 10 may end up as another Vista or Windows 8, neither of which were rousing successes.

Cloudy Future.

Cloud computing in a commercial environment implies the divesting of most of hardware (and operations staff) to save money, and instead "renting" hardware and software from a vendor. Users link in from "workstations" (PCs), and all the work gets done on the servers, just as it did in the old days of timesharing. Cloud computing also facilitates distributed information sharing, including video. The company that will get the lion's share of cloud computing in the future is the company with the lowest prices, the best support and the best artificial intelligence (AI).

Cloud computing is becoming commoditized through an open source concept called kubernetes. A free book is available here (requires registration); there's plenty more free info available elsewhere. Kubernetes controls and helps virtualize computing over a distributed network. Software collections are put into "containers." These containers can be grouped into "pods," which are set to run on the same (or nearby) "nodes" (computers). Replication, proliferation/reduction, and backup can be automated. This is an open source framework, supplemented by different vendors using proprietary add-ons for monitoring and batch controls.

Kubernetes makes it possible to migrate from one cloud vendor to another with far more ease than historically available under separate, proprietary architectures. Large customers will likely opt for long-term contracts, but those contracts will eventually expire. When they do, less competitive vendors will suffer financially. For every customer MSFT obtains for cloud computing, it will need to provide good service and support in order to keep those customers. Does MSFT have a good track record in user support? Look at the multitude of postings on the Internet and draw your own conclusions.

Red Hat, a Linux-based company recently purchased by IBM, encourages use of methodology that permits easy migration from one vendor to another. See here. In the future there will be no loyalty to a particular IaaS company. There are also numerous companies providing management services for IaaS, a good choice when customers want to avoid using the same company for both IaaS and implementation management, to help ensure independence from a specific vendor's platform.

An appealing feature of MSFT's cloud, Azure, is "hybrid," where a computer at the "edge" - hosted at the user site - can be part of the cloud. This is convenient but often unnecessary for customers. Also, that feature could be offered by other vendors.

The cloud is not completely benign. Scammers are hosting their counterfeit websites in the MSFT cloud so they'll get a Microsoft domain name. Then they conduct phishing attacks. Some reports say MSFT does remove such sites when detected. Will MSFT end up with liability for this hosting?

The Competition.

Although cloud computing does not have a moat, due to economies of scale the market for IaaS is a competitive oligopoly. See here for a list and review of over a dozen vendors. The major players are AMZN, BABA, GOOGL, IBM and MSFT. Let's compare them, using figures from yahoo finance and elsewhere for 2018 (dollars in billions; numbers rounded).

Market Cap $899 $402 $790 $117 $961
Net Income $10 $88 $30 $9 $17
Net Assets $25 $356 $158 $23 $30
Forward P/E 48 24 22 9 N/A*
PEG (5 yr. expected) 0.7 72 1.6 3.5 N/A
Debt/Equity Ratio 143 22 7 331 91

*The MSFT trailing P/E was 28.

Note that the BABA PEG seems anomalous, but that's what yahoo reports. The high number is likely due to erratic net income levels, and may end up much lower later on.

The IaaS cloud will require extensive capex. The company in the best position to embark on that capex appears to be GOOGL, not MSFT. The GOOGL D/E ratio is low, with enormous financing likely available if needed. Google also has an advantage as the developer of kubernetes. However, in my view the company best positioned to gain market share and improve growth at this point is IBM. Despite being highly leveraged, IBM has (or can develop) AI intellectual property that I don't think can be matched by competitors. Note that this is my opinion based on IBM's technical history; I have not done in-depth research to confirm it. IBM and Red Hat now have a combination of IaaS and AI that could be a winner, despite IBM's tepid television ads. In this zero sum game market, IBM's gain is MSFT's loss.

Moreover, a short-term win by MSFT or any company is no guarantee of customer retention. The cloud is a utility, like electricity. Compute power will be purchased from the cheapest technically qualified source, after transition costs are factored in.

On the horizon is a continuing expansion of computing capabilities by China, whose economy will soon rival our own. Competition from Alibaba's cloud (see also their white paper) is likely to increase.

The AIs Have It.

Most claims of AI by MSFT and other companies are overly hyped. The mathematical foundations of AI have been available for many decades, but little has come to market. Certain classified AI systems may well be very advanced, but nobody's talking. There's a big difference between data mining, simple question answering, pattern recognition, language translation and genuine AI. Simple question answering systems are well understood, but more complex processing is still a work in progress.

A true AI system is not just deductive; it's also inductive, which means it takes in raw data and draws its own inferences, rather than getting a set of inferences fed in and coming up with conclusions from existing data. True AI will come when humans not only fail to understand how the system solved a problem, but also why. We are a long way from there.

As mentioned previously, if I had to pick one company with advanced AI, it would be IBM. Their research has been going on since the early days of computing. I therefore discount the AI component of MSFT's business plan, because it hasn't been around long enough to be mature, in my view. MSFT was an OS and software applications development and sales company for most of its existence, not an AI research facility.

Kicking the Tires.

During a recent MSFT Azure webinar I asked (all questions were "chat" texts) what Azure's competitive advantages were, as compared to other vendors. There was no reply. A few days later a sales rep contacted me by email. Again I asked the same question, and again there was no reply. A different rep later called me. I asked my question and was told that I would get an email with details. It never happened. If this is MSFT's level of service at the sales level, then the company may have insidious issues that are not apparent on the surface.


MSFT's business process software is not functionally unique; there is competition from both commercial and open source software. MSFT may have more upside as the rising tide of cloud computing lifts all boats, but eventually the cloud will be fully commoditized; it appears to me unlikely that MSFT will end up with the lowest prices and best service. Windows 10's poor performance may eventually irritate enough customers to drive away business, particularly if smart phones evolve into mini-PCs.

MSFT has been profitable and successful as a company so far, but the technology landscape is changing. It is not clear that MSFT can adapt to those changes rapidly enough to maintain growth. My opinion: in the long run, MSFT is a sell.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. 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.