IBM: How Much Longer Do the Good Times Last?

 Dinosaur World by mcdlttx, on Flickr
Creative Commons Attribution 2.0 Generic License  by  mcdlttx 

I have not written about IBM for a while (or really anything ;-), but one of my recurring themes is IBM’s wholesale transition from what was once the world’s premier technology company (admittedly this was decades ago, but their relative and inflation-adjusted historical dominance is clear) into a financial engineering company. For students of the technology company lifecycle, they offer a fascinating career option for once-dominant companies.

This recent quote from a “former IBM exec who spoke on the condition of anonymity” prompted revisiting the company and again asking some questions about their business model and future prospects:

“Realizing this, IBM’s become a holding company of sorts, buying assets, integrating them and reselling them as an IBM brand.”

“IBM does not want to fund its own development because past history shows we’re not very good at it.”

Stock Performance

IBM has been a remarkably successful stock in recent years (and is about 12% of the gain propelling the DJIA to record highs since the 2009 low) and Wall Street loves the company as a result. Here are returns over the last five years:

IBM Chart

IBM data by YCharts

Top line growth however looks more like the surf report for a birdbath on a windless day, even as the bottom line is up over 50% in the same period:

IBM Revenue TTM Chart

IBM Revenue TTM data by YCharts

IBM’s strong bottom line growth has come from cost cutting and financial engineering, not growth in the business. The strategy they laid out in 2007 has worked quite well in terms of profit growth. Like most technology companies, IBM has a “roadmap” of what they’re going to do, but uniquely, it is a financial roadmap. Amid the roadmap’s blizzard of numbers, revenue growth only merits some hand-waving platitudes. They skillfully focus on profit growth for each segment, obscuring the lack of overall revenue growth. And the billions of dollars in euphemistically described “enterprise productivity” savings have come from squeezing employees, curbing retirement benefits (e.g. shifting from monthly to annual 401k contributions) and moving offshore aggressively (IBM arguably now stands for Indian Business Model). Cringley has chronicled how IBM has become the poster child for cutting your way to prosperity at the expense of customers and employees.

Now Wall Street views technology companies through a different lens, as seen in this comparison of IBM and Microsoft when IBM eclipsed Microsoft in value as well as a recent explanation of the market premiums by a Wall Street analyst. While not quite as absurd as David Einhorn telling Apple how to run its business, catering to Wall Street has taken IBM to a different place than they’d go as a technology company left to its own devices. It isn’t a coincidence each successive generation of powerhouse technology company is more open in its disdain of Wall Street. So while IBM’s financial engineering strategy has worked brilliantly to date, there are questions about how much longer this strategy will work, much less provide superior returns.

Where is the Innovation?

“IBM does not want to fund its own development because past history shows we’re not very good at it.”

IBM is the perennial leader in terms of patents issued, garnering them a wave of beneficent press every year. But if we define innovation as requiring both invention and market impact, it is hard to point to a single, material innovation IBM has delivered in the 21st century. Things like Watson or the latest nanotechnology transistor advance are duly milked for PR, but they don’t ever seem to turn into world-beater products that move the needle. When specific IBM patents are examined, they tend to be egregious examples of all that is wrong with the patent system and the company often finds itself magnanimously “contributing” what they had touted as great inventions to the public domain in the wake of backlash.

But the biggest innovation challenge for IBM, as we’ll elaborate on below, is their also-ran status in cloud computing. My comments from 2008 still hold today:

Meanwhile, on a more serious note, timesharing is back with a vengeance yet there is no sign IBM has ambitions to be a major player in the cloud computing era. Instead they’re fiddling with avatars while the on-premise business starts a long, slow burn. Where is the “one billion dollar” data center capex announcement that signals their ambition to play with the Amazons, Googles and Microsofts? Perhaps it is harder to make a “billion dollar” commitment when it requires real dollars as opposed to “soft” and/or exaggerated dollars? Or has IBM committed all its free cash flow to financial engineering, forcing them to watch the next generation of computing from the sidelines? Selling servers and consultants by the hour is a far cry from offering (anything)-as-a-service. In the cloud world, if you build it as a vendor, you also have to be willing to operate it at scale. And enterprise-scale is dwarfed by Internet-scale, so enterprise chops are not enough. Outsourcing “your mess for less” isn’t a service. One can only speculate that virtual conference rooms in Second Life are inhibiting IBM’s ability to define their strategy. I am still hoping they do OnDemand 2.0 and kill two birds with one slogan.

What about Cloud?

IBM’s value proposition since Lou Gerstner brought the company back from the brink has basically been “we’ll bring you up to average in IT” with IBM (human) services providing the glue to hold it all together (also known as “your mess for less” though it is intentionally really hard to tell whether the cost savings promised in the IBM sales spreadsheet actually panned out as promised).

We can argue about whether IBM offers IT capabilities at the 50th percentile (or 30th or 60th), but for roughly half of enterprises, IBM offered an improvement on what they could do themselves (and IT is arguably the opposite of Lake Wobegon – a land where everyone is below average ;-). The challenge for IBM is cloud providers are more like 80th or 90th percentile (or higher) in their capabilities and have far better cost structures through software as opposed to busloads of consultants. The fraction of the market IBM can compete for will shrink due to cloud computing as complexity gets engineered out.

At the same time, the pendulum has swung away from outsourcing and in-sourcing now has cachet (perhaps in no small part to the quality of the outsourcing experience), which reverses the tailwind that has propelled IBM since Gerstner.

IBM has half-heartedly rolled out SmartCloud, which is at best difficult to understand (e.g. IBM’s definition of self-service is very different than the rest of the industry and requires human involvement to process orders and/or conceal their offerings from critical eyes). Tellingly, their top level message around cloud is a vague plea for open standards which is typically the last refuge of scoundrels and the uncompetitive.

The chip guys used to talk about a “poker chip” which was the cost of a single fab. The cost of a poker chip went up with every generation and now runs in the billions. You need cash to play the game. For cloud, the poker chip is the cost of a cloud-scale data center, which can run into the hundreds of millions of dollars. At minimum, you need a few of them scattered around the globe. To cater to IBM’s preferred kind of enterprise customer, you probably need more than a few to offer granular jurisdictional compliance. We see the pretty pictures of massive datacenters belonging to Facebook, Google and Microsoft, but where are IBM’s poker chips? If you buy chips to play the cloud poker game, that money is not available for dividends or buybacks (and vice versa). You’d think IBM would be anxious to show the world vast structures stuffed to the gills with mainframes, demonstrating once and for all their claim that mainframes really are the best place to run the most modern workloads. Or maybe they just don’t have them.

Some will argue there will always be room for integration between and on top of software services, but cloud leaders are ruthlessly removing humans from the nuts and bolts of IT operations. IBM lacks meaningful or competitive cloud technology assets and per the IBM executive above, they don’t seem particularly confident in their ability to develop new technology. The other option is to “move up the stack” and offer business consulting, which is certainly the way IBM positions the company. But contrary to the story, IBM remains extremely exposed to menial IT functions that are disappearing in the cloud era. And their intense focus on cost reduction through off-shoring doesn’t add to their business consulting skills. The rise of the cloud brings IBM’s inability to innovate to the fore and poses a real threat to their current business model.

Where is the Thought Leadership?

One thing IBM has historically done pretty well is thought leadership. Particularly for a company of their size, they had remarkable discipline in aligning a sprawling empire (at least from the outside) top-down around a common story supported by strong advertising, marketing and top-down sales relationships. Their embrace of Java and then Linux/OSS felt like technology leadership at the time but in fact it  represented IBM’s outsourcing of technology leadership. They had moved from innovating to endorsing technology and felt comfortable in their ability to remain on top as an integrator of the piece parts. In fact, the more piece parts coming from different centers of gravity, the better for the consultant.

I believe OnDemand was a milestone in IBM’s thought leadership history, It was obviously well before its time and presaged a lot of what we take for granted with cloud computing. My theory is IBM quickly discovered they were ahead of the market, and way, way, way ahead of their typical, lower percentile customers who just weren’t going to be early adopters. So OnDemand quietly disappeared. Nor are there any obvious technology or product artifacts from the effort. I suspect they still have deep scars from the experience that are impacting their ability to wholeheartedly embrace cloud, even if they had the means to do so.

IBM’s most recent storyline has been Smarter Planet and it is a different beast than former campaigns. It unambiguously puts a business value message first, and aims foremost at governments and those selling to governments. This was partly a reflection of the timing (it rolled out in November 2008, in the depths of the financial crisis) and IBM wanted to dine at the trough of government stimulus spending. But it is also also a reflection of the sophistication (or lack thereof) of the typical big IBM customer. Governments are rarely among the most able and discerning of IT customers (yes, there are lots of recent efforts to put a positive spin on their need to do better, but when you look at the lowest percentiles of IT organizations, lets just stipulate that governments are well represented).

Meanwhile, cloud computing has become the high order bit for IT today and IBM is hard to find. They’re not only not shaping the discussion to their advantage, they’re not really in the debate. Cloud leaders don’t use IBM and if you want to seriously play in cloud, IBM is hardly going to make your list of strategic partners or suppliers.

Where is the Customer Success?

IBM does a great job with their advertising and customer case studies, which tends to mask the bigger question: where are the customers with a big IBM technology reliance who are leading their industries through application of technology? What disrupter relies on IBM for their technology and/or consulting services? Who is successfully holding off the disrupters and/or revolutionizing their industry using IBM technology or business consulting? Which of today’s startups undergoing explosive growth have done it on an IBM platform?  [Pause here and wait for the crickets].

With no offense to (Sri) Lanka Bell (“a leading telecommunications service provider” and SmartCloud adopter) or the Memphis Police Department (who did some statistical analysis in SPSS), these are not world-beater customers doing breakthrough things with cutting edge technologies. Those IBM TV ads about Smarter Analytics are really about SPSS, a 45-year old product that IBM had to acquire? How many IBM “big data” case studies are running on good old IMS (which shares a birth year with SPSS)? Granted IBM does a nice job making its late majority adopters look good, but it is hard to find IBM customer examples that live up to the rhetoric.

Broadly, I am seeing much greater enterprise customer awareness of the deficiencies in their in-house and traditional outsourced IT capabilities. As technology moves into broad deployment across all industries, almost every company is now a technology company to a greater degree and has to compete against technology-based startups using 21st century technology. Om Malik makes this point really well by pointing out who is buying technology companies now; it isn’t just other technology companies any more. And these are the companies who are IBM’s traditional bread and butter customers.

As a result, there is a business imperative to “do better” when it comes to IT. Some companies are even asking how to walk away from all their IT investments, not because “IT doesn’t matter”, but because it matters so much and the current platforms are such a drag they see starting from scratch as the only option. It will be interesting to see if anyone pulls this off, or if it remains a material advantage for new entrants.

Critically for IBM, this mindset is taking hold amongst companies who historically lived the idea you couldn’t get fired for buying IBM. They’re actually concluding their IBM dependency might not just get them fired, but it could kill them. I remember visiting one household name a few years ago where they proudly showed me the building full of IBM consultants and the CIO responded to my “you have to get off the mainframe – you’re stuck in the land that Moore’s Law forgot” pitch with a retort that began with a patronizing “Son, there’s something you need to understand…” That company has since seen the light and has decided IBM must play a far less significant role in their future if they want to stay competitive.

And when it comes to creating success for their customers, IBM’s pure enterprise footprint looks like an ever bigger liability. They can talk about selling to Chief Marketing Officers all they want, but the reality is IBM can’t bring hundreds of millions of customers to the discussion the way an Amazon, Apple, Google or Microsoft can who actually touch customers at scale and can bring that understanding and technology to bear for customers. Watson makes for a nice demo and game show contestant, but how does it really compete with Google Now or Siri in terms of reach (make your own WebSphere joke here) or even scale? The consumerization of IT as much as anything is the world moving away from IBM.

The Case Against IBM

That got really long. To summarize, the good times rarely last forever and IBM is unlikely to continue to outperform:

  1. Wall Street may love IBM (which is an indictment unto itself…) but mean reversion looks inexorable. IBM has put itself in thrall to Wall Street and achieved superior earnings growth through non-sustainable cost cutting and financial measures. What is good for Wall Street in this case isn’t good for customers or the long term success of the company. Historically low interest rates have also been part of the IBM story (borrowing at a rate below your dividend rate to buy back stock is cash flow accretive) and that too will mean revert one of these days.
  2. IBM is a technology company that doesn’t really innovate. Cost cutting only takes you so far; eventually it cuts muscle as opposed to fat. Real technology companies do actual (non-financial) engineering, but IBM has committed its cash to growing EPS. IBM doesn’t even bother trying to tell a growth story (they should at least be able to grow with  global GDP). Meanwhile, they have lots of legacy business at risk. In particular, IBM makes the bulk of its revenue from busloads of consultants and the plurality of its profits from the mainframe. And a rats’ nest of a legacy software rollup doesn’t make you a software company, investor relations slides to the contrary (are there any IBM software acquisitions that aren’t in harvest mode and have seen material investment?).
  3. The company is missing the boat on cloud computing. IBM is largely an observer in the biggest trend in IT, which directly undermines the core of IBM’s business model. And the legacy software acquisition model of the last decade will not work as well applied to SaaS applications – software services require on-going innovation, don’t drag consulting to the same degree and the economics are not as accretive given the capex needs.
  4. The big industry trends are against IBM. Not just cloud, but also outsourcing swinging back to in-sourcing and the consumerization of IT. IBM is fighting these trends, not setting an alternative agenda.
  5. IBM’s customers are in crisis. Historical affinities of even the most complacent customers will be put to the test in the face of existential, technology-based business threats. You can only paint laggards as technology leaders for so long and if the IT center of gravity is shifting from the CIO to the CMO, IBM’s lack of consumer assets or DNA hurts them relative to the competition.

Disclosure: ironically, I am long IBM though not by any action on my part, and have been too lazy to do anything about it. This post hopefully will motivate me to rectify that.

Book Review: The Big Switch

Nick Carr made his name with the provocative Harvard Business Review article “IT Doesn’t Matter” (free version here), its expansion into a less definitively titled book Does IT Matter? and his generally erudite blog.  The charge of irrelevance hit the industry hard and elicited mostly incoherent and ineffective rebuttals (e.g. “hogwash”), which hampered real discussion of Carr’s argument.

I have gently mocked his thesis previously but found it a mix of the obvious (yes, things get commoditized over time, so you focus on the top of the stack and of course further commoditize the rest of the stack) and the ridiculous (IT had apparently previously been a source of everlasting strategic differentiation, but with the democratization of computing making technology widely available, we should write off the industry in its entirety).  It is like arguing that since everyone has a brain, don’t bother thinking…

Carr has a new book, The Big Switch: Rewiring the World, From Edison to Google, in which he contemplates the future of computing and speculates on the broader societal impact of that future.  The book is lucid, well-written and uses lots of historical examples to make the narrative and arguments come alive.  The first half of the book looks back at the evolution of the electrical industry and argues the computing industry will follow the same path.  The later half offers up social, economic and cultural consequences of the shift, again using electrification as an example of how new technologies have secondary and unforeseen effects.  Carr is less than excited about the consequences of the technology path he believes is inevitable– no one will mistake him for an Internet optimist.

Back in the 19th century, companies generated their own power locally, whether through water, steam or early electrical generation.  The advent of alternating current meant power could be generated remotely and transmitted afar, allowing companies to get out of the power business and buy electricity from the new electrical utilities. 

Carr tells the story of Thomas Edison and his former clerk Samuel Insull.  Edison, with his bet on direct current which didn’t lend itself to long distance transmission, focused on small-scale generators that ran “on-premise”.  His model was to sell every business equipment to generate their own electricity.  Insull predicted the rise of the electrical utility, foresaw it would eclipse the equipment business and left Edison to join what became Commonwealth Edison.  (Empires of Light is a great account of the battle between Edison and direct current versus Tesla and Westinghouse who championed alternating current).

By offering electricity to multiple customers, utilities could balance demand and reap economies of scale that drove a virtuous cycle, allowing them to drive down the cost of power and thereby attract even more customers.  Their strategy was predicated on maximizing generator utilization and the standardization of electrical current.  Companies that outsourced their power generation to utilities no longer had to worry about generating their own electricity, reducing cost, staff, technology risk and management distraction.

Turning towards computing, Carr reprises his “IT Doesn’t Matter” death knell: IT is an infrastructural commodity that every company has access to, so there is no differentiation available, which means it is a dead cost.  He recounts the history of computing, showing a particular fondness for the punch card, and excoriates the industry for cost, complexity and waste.  Siebel is the chief punching bag (while deservedly so, it is an easy target).

His future trajectory for the industry has the Internet playing the role of alternating current, allowing computing to be performed remotely which in turn enables a new breed of computing utilities (with Amazon Web Services, Google and Salesforce as early poster children).  The end result is companies no longer have to run their own complex computing operations.  He calls this new era of computing the “utility age” and states “the future of computing belongs to the new utilitarians”.

Enterprise computing vendors who sell “on-premise” solutions will be marginalized like Edison, unless they can reinvent themselves (as Edison’s company ultimately did, shifting both technology and customer allegience – they’re still around today, a little outfit called General Electric).  Carr dwells on Microsoft’s recent embrace of cloud computing, but questions whether the company can navigate the difficult transition of embracing a new model while continuing to harvest profits from the old model.

Is the Big Switch Big or Not?

I have two critiques of the first half of the book.  The first is mild schizophrenia.  The Big Switch is — wait for it — as follows:

“In the years ahead, more and more of the information processing tasks we rely on, at home and at work, will be handled by big data centers located out on the Internet.”

Wow.  Gather now at the knee of the S-curve to learn what the future holds.  Perhaps he is aiming the book at a more general audience, but with over a billion people regularly accessing the Internet, there are an awful lot of people who have already made the “big switch”.  He does some hand-waving about broadband penetration to explain why the book isn’t over a decade late, with no mention of the failure of the late 20th century’s application service providers.

Carr can’t quite decide whether the big switch to his utility age is a revolution or not.  He equivocates about whether a wave of creative destruction is crashing down today or if it will take decades to play out.  He also qualifies the move to the cloud and how far it will go with suggestions that the future may actually consist of cloud-based services working in conjunction with local computers in corporate data centers and/or local PCs.  This qualification I think stems from his general tendency to paint everything with a very broad brush.  In practice, there are many segments and technologies, each with their own dynamics.  He also plays fast and loose with topology, enlisting highly distributed examples to support a centralized thesis.

The Fallacy of the Perfect Analogy

My second critique is that the book turns on the idea that computing is basically similar enough to electricity that it will inexorably follow the same path.  While there are similarities, it is a mistake to assume they are alike in every aspect.  There are enough differences that blind adherence to an analogy is dangerous:

  • Electrons are fungible, CPU cycles arguably are, but information is not fungible.  While the flow of electrons could be standardized, the flow of information can’t.  His tendency to blur, conflate or confuse hardware and software, clients and
    servers and individuals and IT doesn’t help. 

  • Even when you do computing remotely, you still compute locally as well.  A search engine query, for example, gets run in a giant data center somewhere off in the cloud, but there is still processing that happens locally to submit, display and act upon the results.  The browser is hardly a dumb terminal and the trend is to exploit even more processing locally for cloud-based applications (with AJAX and RIA techniques).  Further, there are strong business incentives to use local code to differentiate the user experience and allow eyeball businesses to push interaction rather than just relying on user pull.  Computing is likely to be much more distributed than electricity production, especially when you consider… 
  • I am not aware of a dynamic like Moore’s Law (and similar rapid improvements in storage capacity and bandwith) for electrical generation, which both projects significant performance improvements over time and introduces the concept of relative scarcity and abundance.  This dynamic undermines the parallel of CPU utilization and generator utilization.  Those who best exploit relative abundance and put the processing closest to the data will prevail (I remain a Jim Gray disciple).
  • Distance still matters at scale. For the same reason that the aluminum industry located near cheap sources of electricity, the algorithm for siting new new half-billion-dollar data centers looks at proximity to both cheap power and end customers.  Likewise, Akamai offers proximity with its edge caching network, which in turn means lower cost and more responsive services.
  • Just as the computing industry is looking longingly at the electrical utilities, the electrical utilities are envious of the more distributed nature of the computing industry.  A less centralized and more intelligent electrical grid promises greater efficiency and resiliency.  The availability of realtime pricing information can increase conservation and reduce peak-loads.  Distributed generation allows locally produced power to be sold back to the grid, making alternative energy sources more compelling.  Digital power opens up new, differentiated offerings for utilities based on the quality of power.  And a more decentralized grid means that a single point of failure doesn’t take down power for 50 million people.  All these trends suggest the electricity industry will look more distributed and information-rich in the future.
  • There are probably other relevant differences as well.

So while the book gets the broad trend to more computing in the cloud right, Carr’s extended analogy obscures a lot of the differences and subtleties that will make or break cloud computing endeavors.  Between the caveats and the broad definitions, there is a lot of leeway in his technical vision (admittedly the mark of a savvy forecaster).  Victory will go to those who best exploit both the cloud and the edge of the network.  Carr’s own examples — Napster, Second Life and the CERN Grid — make this case, even if he either misses their distributed nature or chooses to ignore it.

Utility, Not Utopia

The second half of the book focuses on the broader social and economic consequences of the move to utility computing.  It is the bolder and more thought provoking part of the book. 

Carr again begins by looking back through the lens of electrification.  He succinctly credits electrification with ushering in the modern corporation, unleashing a wave of industrial creative destruction, improving working conditions by displacing craftsmanship for the modern assembly line and the gospel of Frederick W. Taylor, improving productivity which begat a broad middle class and white collar jobs to coordinate more complex organizations, the broadening of public education, expanding demand for entertainment, and enabling the suburbs (cheap cars relied on cheap electrical power to power the assembly line).

He also notes that the early years of electrification were accompanied by great optimism and even utopianism about what the future would hold.  Carr, however, leaves his rose-colored glasses at home as he ponders his utility future:

“Although as we saw with electrification, optimism is a natural response to the arrival of a powerful and mysterious new technology, it can blind us to more troubling portents…. As we will see, there is reason to believe that our cybernetic meadow may be something less than a new Eden.”

Carr basically finds his utility future dystopian.  He spends the remainder of the book worrying about:

The Hollowing Out of the Workforce – the utility future has little need for workers, which reverses the positive virtuous cycle of employment driven by electrification.  He points to increasing returns businesses like YouTube, Skype, craigslist, PlentyofFish and giant data centers with small staffs leading the way “from the many to the few”.  They are free riders on a fiber backbone paid for by others and are ushering in a world where “people aren’t necessary”.  “Social production” (aka “user-generated content”) is simply digital sharecropping and reduces the need for workers further.  Unlike electrification which “played a decisive role” in building an egalitarian society, the utility age “may concentrate wealth in the hands of a small number of individuals, eroding the middle class and widening the divide between haves and have-nots”.

The Decline of Mainstream Media – while electrification “hastened the expansion of America’s mass culture” and gave rise to mass media, the Internet is undermining the media with its explosion of voices and “some of the most cherished creative works may not survive the transition to the Web’s teaming bazaar”.  Newspapers are of course the foremost example.  The shift from scarcity to abundance of content is not a good thing to Carr and “the economic efficiency that would be welcomed in most markets may have less salutary effects when applied to the building blocks of culture.”  The result is a decline of media and shared culture, the polarization of virtual communities (exacerbated by personalization engines) , “social impoverishment and social fragmentation”.

Bad Guys – the Internet in the utility age promises to be a magnet for bad guys, including  criminals, terrorists, botnet operators, spammers, perpetuators of denial of service attacks and fiber optic cable-snapping earthquakes.  The underlying infrastructure is fragile and vulnerable yet critical to the global economy.  This was the least forward-looking of his pessimistic projections.  He mostly reiterates issues.  About the only new claim about the future was that pressure to protect the Internet from “misuse and abuse” will stress the sovereignty of nations  as utility functions migrate to countries with the lowest operating costs.  He is surprisingly silent on whether we should expect the heavily regulated nature of electrical utilities to also apply to computing in the future.

Privacy and the Control Revolution – don’t even think about having any privacy in the utility age:

of us are aware of the extent to which we’ve disclosed details about our identities and lives or the way those details can be mined from search logs or other databases and linked back to us.”

Carr believes computing always has and always will be fundamentally a tool of oppression for the Man, the computing revolution is really just part of a broader “Control Revolution” and the empowerment of the personal computer will be “short-lived” as the Man inevitably reasserts control:

“The sense of the Web as personally “empowering”…is almost universal. …  It’s a stirring thought, but like most myths its at best a half-truth and at worst a fantasy.  Computer systems in general and the Internet in particular put enormous power into the hands of individuals, but they put even greater power into the hands of companies, governments, and other institutions whose business it is to control individuals.  Computer systems are not at their core technologies of emancipation.  They are technologies of control.  They were designed as tools for monitoring and influencing human behavior, for controlling what people do and how they do it.  As we spend more time online, filling databases with the details of our lives and desires, software programs will grow ever more capable of discovering and exploiting subtle patterns in our behavior.  The people or organizations using the programs will be able to discern what we want, what motivates us, and how we’re likely to react to various stimuli.  They will, to use a cliche that happens in this case to be true, know more about us than we know about ourselves.”

Carr is particularly full of disdain for the PC as a device but is conflicted about personal computing.  He readily acknowledges the empowering impact of personal computing, yet simultaneously promotes a dumb terminal future while lamenting the inevitable reassertion of control by the Man (somehow those seem related…).

He concludes on the cheery note that the utility future is no less than another front on “humanity’s struggle for survival”.  Actually, I took that quote from the Gears of War 2 announcement, but it would not be out of place in Carr’s conclusion.  He fears the utility age may devalue quintessential human attributes, making us (even) more superficial, undermining the coherence of the family and relegating us to mere “hyperefficient data processors, as cogs in an intellectual machine whose workings and ends are beyond us”.  Bummer, dude. 

The second half of The Big Switch is kind of a dour read and the utility future is boldly painted with a Luddite, elitist and generally defeatist brush:

“…we may question the technological imperative and even withstand it, but such acts will always be lonely and in the end futile.”

In a book full of references to big thinkers, from Jean-Jacques Rousseau to Alexander Solzhenitsyn, Ned Ludd does not merit a mention, even though the Luddite fear of automation hollowing out the workforce is repeated almost verbatim.  He doesn’t acknowledge the parallel or make a case for why the Luddite fears are more warranted now, despite failing to come to pass in the Industrial Revolution.

And while he bemoans the rise of “a new digital elite”, the shifts in media, and survival of our “most cherished” work, he manages to come across as an elitist himself (not that there is anything wrong with being an elitist of course…).  I’m just not sure the Brahmins get to decide what is and isn’t worthy media.

It is hard to argue with his position on privacy (read No Place to Hide to shatter any lingering techno-optimism on this front — large-scale databases go awry, period), but he doesn’t make the case that the black helicopters of the Control Revolution are just over the horizon.  Individual freedom is pretty much at an all-time high in world history and information technology gets at least some credit for that.  Carr does admit technology is “dual use”, but you won’t find much on the positive uses in the book.

The Big Switch is well worth reading if you’re thinking about the evolution to cloud computing.  It provokes and stimulates as this long-winded review shows.  Carr’s technical foundation is shaky, but he is a good social critic and forecaster, and a great polemicist (and that is a compliment).  My view is Carr’s dystopian future is not inevitable, but averting it will take a conscious and proactive effort.  If nothing else, the later part of the book is a call to arms for what must be avoided.  If the Control Revolution is indeed a revolution, it is time for a counter-revolution.

The Much Misunderstood Larry Ellison

samurai1 It is not often I rally to Larry Ellison’s defense.  In fact, it has never happened, unless you count that incident involving two underage interns, the failed MiG fighter acquisition and ten thousand cubic yards of Jello, but the legal settlement thereof bars further elaboration.

Larry recently made a statement which people are assuming is just a typical, cynical, self-serving, Machiavellian exercise in spin and depositioning as befits a disciple of Sun Tzu and Miyamoto Musashi (bonus points for the commenter who can connect the advice of either of those strategic gurus with dumpster diving).

But what if he actually spoke the truth?  Past performance admittedly might lead you to overlook such a possibility, but Larry’s comment came during Oracle’s quarterly earnings call with financial analysts last month.  It slipped out in the midst of the ritual competitive bashing, in this case of SAP’s new midmarket offering, that Oracle uses to distinguish its conference calls from those of other publicly traded companies:

“So while we think it’s an interesting market — the small market — because it’s large, we just haven’t figured out a way to make a substantial profit in that market. We think it’s hard to make money. Our strategy: add more value, go upstream, sell industry-specific software to our existing customers, and we’ll watch and see how SAP does going after small companies. Especially with in Software as a Service which we think is very interesting, but so far no one has figured out how to make any money at it.

Focus on the last line about the bottom line where Larry questions whether people can make money off of SaaS in the SMB market.  The SaaS euphoria has been driven largely as a technical imperative by enthusiastic new market entrants.  The SMB market has been held up as a green field that lets providers sidestep the myriad complexities of the enterprise.  And there is no doubt this is a vast and underserved market that would love to eliminate the hassles of deployment and operations.

But what about the profit opportunity?  Now Oracle is a company run by investment bankers these days, but it is a company that banks serious profits – over $4 billion last year with net margins approaching 25%.  When they look at business opportunities, they are looking for big profit pools.

Larry Dignan at ZDNet jumped in and channeled his namesake:

What Ellison could have said [is that] no one has found a way to make gobs of money in SaaS. is profitable but you could find the $481,000 the company made in its latest fiscal year in Ellison’s couch. Indeed, without maintenance fees SaaS may be less profitable in the long run.

However, Ellison can make these comments precisely because Oracle has its SaaS plan ready–it’s called NetSuite.

I think he was a lot closer with the first paragraph than the speculation in the second paragraph.  Larry does have a personal investment in NetSuite that is raising interesting questions as NetSuite tries to go public, but I believe that is irrelevant.  No matter how you look at it, a subscription-based SMB-focused SaaS model at scale looks like a tough business model.  NetSuite certainly has no solution to the financial challenges as we’ll see below.

Nick Carr also noticed Larry’s comment in a good piece on the shifting sands of enterprise software although he displays less than his usual level of certainty:

How will the competition play out? I wish I knew. There are at least two unknown variables: the speed with which the SaaS model penetrates larger companies, and the ultimate profitability of the SaaS model. There are some indications that the adoption of SaaS is advancing more quickly than expected, but there are also indications that the hype may at the moment be getting out ahead of the technology. As for profitability, about all we can say is that Ellison is probably right: SaaS is unlikely to be as lucrative as the old licensing model that it’s replacing – which happens to be very good news for customers.

Profitless prosperity may be great for customers, but at some point companies and their investors will have to find some profits or the activity will come to an end (unless you’re in the airline business).


That was a really verbose motivation to take a look at’s financials and make a prediction about their stock price direction.  As the SaaS poster child, they offer some interesting lessons in the search for potential profits from SMB SaaS.  My conclusion is that the poster child’s economics are not only not the shining beacon for the industry you’d expect but actually a disappointment.  Lets go to the numbers:

  • Revenue: $613 million over the last four quarters.  Up from $497 million for the last full fiscal year (we’re half way through the latest fiscal year), which in turn was up 60% on the prior year.  No doubt Salesforce has been a great revenue growth story.
  • Earnings: $8.17 million over the last four quarters, up from Dignan’s $481,000 in couch money for the last full fiscal year (but below 2006’s $28 million).  They’re not doing a great job turning that revenue growth into profits, and this is a company that has been around since 1999.
  • Interest income: $22.43 million in the last four quarters, which is nearly three times the company’s overall net profit in the same period.  Hmm…
  • Valuation: Salesforce has a market capitalization of over $6 billion as I write this and trades at a forward P/E of over 500.  That valuation is supported by less than a million dollars per month of profit and no profit if you take out interest income.  Wall Street has some mighty big expectations for future profits.

So where will those profits come from?  The problem is there is a giant sucking sound on the income statement that exerts massive drag on profits: sales and marketing which is a proxy for what it costs to acquire customers.  Bruce Richardson at AMR computed the numbers:

Over the last six quarters, has spent between 49.7% and 51.1% of revenue on sales and marketing.

NetSuite is no better  and could be worse, again from Bruce Richardson:

As NetSuite’s Zach Nelson pointed out in The Wall Street Journal (September 19, 2007), even with free trials, it tak
es two months and three to five product demonstrations to close a sale. He was quoted as saying, “It isn’t easy to figure out how to acquire customers and keep them happy at a low enough cost that you still earn healthy margins.”

A cursory review of NetSuite’s financials in their recent S-1 filing to go public shows they lost $23 million on $67 million in calendar 2006 and while they have a nice revenue hockey stick over the last three years, the profit line is borderline horizontal and well below the x-axis (i.e. in the red).  They also call out explicitly the challenges of SMB customer acquisition:

Our customers are small and medium-sized businesses, which can be challenging to cost-effectively reach, acquire and retain.
We market and sell our application suite to SMBs. To grow our revenue quickly, we must add new customers, sell additional services to existing customers and encourage existing customers to renew their subscriptions. However, selling to and retaining SMBs can be more difficult than selling to and retaining large enterprises because SMB customers:

• are more price sensitive;

• are more difficult to reach with broad marketing campaigns;

• have high churn rates in part because of the nature of their businesses;

• often lack the staffing to benefit fully from our application suite’s rich feature set; and

• often require higher sales, marketing and support expenditures by vendors that sell to them per revenue dollar generated for those vendors.

If we are unable to cost-effectively market and sell our service to our target customers, our ability to grow our revenue quickly and become profitable will be harmed.

The subscription model Salesforce, NetSuite and many other SaaS vendors use is extremely sensitive to the interplay between the cost to acquire a customer (which is an up front cost before you see any revenue), the average revenue per user (which dictates how long it takes to pay back the up front cost, never mind COGS and overhead) and the rate of churn (which tells you how many customers will stick around long enough to pay back the up front cost).  Small deltas in these numbers can make a big, big difference in terms of profitability.  Vonage is a great example of what happens when those numbers don’t add up harmoniously.  And these parameters are interdependent.  Raising prices likely increases churn, for example.

One argument to support Salesforce’s valuation is they are a young company that is investing for growth and just need to get to scale for their business model.  The difference however is they are not like the traditional software company trying to grow into a relatively high, relatively fixed R&D budget where above some threshold additional revenue falls largely to the bottom line.  Salesforce spends relatively little on R&D and their capex on infrastructure is surprisingly low.  The issue is their sales and marketing cost which seems to scale with revenue.

Looking at Salesforce’s future, they face a couple challenges.  Revenue growth will continue to slow.  It has been slowing for five years.  It is always harder to sustain your percentage growth on a bigger base.  And with minimal profit growth, revenue growth seems to be the proxy for the sky-high multiple, and assumes a big profit payoff is coming some time in the future.

They also will face intensifying competition.  Microsoft will soon offer a hosted version of Dynamics CRM, which has only been available as an on-premise offering to date.  The product offers native Office and Outlook integration, deep customization  and broad partner support.  SAP and Oracle are also peripherally in this market, although both have pricing above Salesforce and haven’t really been serious to date about the SMB CRM market.  There also are other startup competitors like NetSuite.  Competition can impact the key variables in a couple ways, all negatively:

  • Acquisition cost – it is likely to cost more, not less for Salesforce to win a customer in the face of more competition.  Today Salesforce spends over $700 to acquire a new user.  Competitors like a Microsoft or an Oracle may have a structural advantage in terms of acquisition costs because they are global players with an instantly recognizable brand who can spread their sales and marketing investments across both on-premise and SaaS deployment options for their software.
  • Average Revenue Per User – competition also tends to reduce prices.  Microsoft’s pricing of $39-59/user/month for the SaaS version out of the gate will undercut Salesforce’s approximately $71/month in average revenue per user. 
  • Churn – when customers have more choices, they’re more likely to go somewhere else.  Today Salesforce needs about ten months to pay off the initial acquisition cost of a subscriber.  When you add in COGS, R&D and corporate overhead, it is more like 18 months.  And unilateral competitive spite gestures on par with cutting off the nose to spite the face probably don’t help on the churn front either (see The Fat Guy‘s blog).

Small perturbations in these parameters can have a huge impact on the attractiveness of the model, and it is easy to see downward pressure on all of them.  It is hard to see how Salesforce can reduce their marketing spend as competition heats up unless they want to stop growing.  So what happens?

  • Salesforce will continue to pursue more “enterprise” customers and try to move up market.  Their average seat size per deal has almost doubled in the last five years and a very small number of customers (<100) may account for upwards of 40% of their seats.  The enterprise may prove a more cost effective marketing environment, but it also brings with it many more requirements and competition (in their favor, one of those competitors is Siebel, now owned by Oracle, who have the least satisfied customers I have ever seen).
  • Salesforce will continue to push their platform strategy.  It is a nice story, but it hasn’t contributed any discernable network effects to date that reduce acquisition costs.  But they seem to be doubling down that it might, suggesting a lack of other promising strategies.
  • You will see serious multiple compression for Salesforce (meaning their stock price goes down).  Baring some kind of miraculous breakthrough in terms of a more efficient marketing model for the SMB market, slowing revenue growth without any concurrent growth in profits is going to squeeze the stock.  Going from an insane forward P/E of 500 to an almost-as-insane 250 means the stock halves.  But the reality is when the sentiment flips, the compression is likely to be more extreme.  I think the stock hits the wall in the next 9 months (and the company likely will continue to see good revenue growth in that time).  The frenzy of insider selling suggest this possibility may not lost on management.
  • Salesforce will try to sell out and the challenge for Marc Benioff is the timing.  They simply can’t get to the scale needed to be a viable, standalone, long-term company on their current path with their current scope.  Most of the buyers probably understand that even with sustained revenue growth, in the absence of commensurate profits emerging, the company will likely get cheaper over time.  So who might be a buyer?  Some say Google but Salesforce’s inefficient, unautomated mark
    eting model is anathema to Google’s economic model.  My best guess, to bring this now way-too-long post for the three readers still with us full circle, is none other than Larry Ellison, who was perhaps being both truthful and crafty in his comments far above.  He once said of the old Computer Associates: “every ecosystem needs a scavenger”.  That idea evidently grew on him and he’s made it the strategy at Oracle, spending tens of billions to roll up business applications companies, where they can jack up and harvest maintenance revenues from relatively price-inelastic customers and cross-sell across the installed base.  It is a strategy fundamentally based on leveraging their sales and marketing model.  Buying Salesforce would give Oracle a more powerful CRM franchise than the legacy lineup they have today.  It would give them a stronger SaaS offering in addition to their on-premise offerings, although they would not have the advantages of a single code base for both deployment models unless they packaged up Salesforce to license to others to run.  They could continue to push Salesforce up into the enterprise market or use their existing marketing machine to at least sell into the mid-market.  And if the platform play starts to kick in, Salesforce does run on the Oracle database (unlike, ironically, some of Oracle’s other SaaS offerings…).  Obviously Benioff goes way back with Ellison, who had an early investment in Salesforce.  If Tom Siebel can be welcomed back to the Oracle mothership, so too can Marc Benioff.  Just a matter of the right price.

Disclaimer: I am not a financial analyst, don’t play one on TV but do think the above prediction is far more probable than some of the things that have surprised the best and brightest of the financial analysts of late.  I have no positions in CRM or ORCL.

(Image copyright The Cartoon Network)