In late 1989 I spent some time thinking about what was then called the “emerging new world order”, and began to rough out a think piece titled World War Four. I scribbled an introduction in January of 1990, but since I'd already discussed the gist of the message verbally with just about everybody who was likely to read such a paper, I put it aside and never completed it.
The essence of World War Four was the argument that amidst all the optimism and triumph engendered by the collapse of communism—the end of World War Three (the Cold War, 1945–1990), we were in all likelihood at the threshold of World War Four (1990–????). World War Three, a relatively bloodless war which nonetheless consumed far more wealth than any of its predecessors, was about which economic and political system was appropriate for large industrialised nations. That's pretty much been settled now. World War Four was, I predicted, going to be centred on the essential definition of a nation and the appropriate size and scale of political entities. I went on to argue that 45 years of bipolar confrontation had simply put a lid on these issues, which have been at the heart of the overwhelming majority of all modern wars, and that we were probably entering into an era where borders were going to be redrawn all around the world.
Well, of course, if I'd finished the piece and managed to get it published anywhere, I'd probably be spending all my time on the talk shows as the Political Prophet of the Nineties, so it's just as well I didn't. But I stand by my 1990 predictions, including the one that there will be a serious secession movement by one or more states of the U.S. before the end of this decade.
So what does this have to do with the title of this paper? Well, nothing really other than the fact that I've put everything aside to finish this paper because I have the feeling that if I don't it's going to suffer the same awful fate as World War Four—namely come true and be considered totally obvious before I get around to predicting it. I've been developing the ideas in this paper over the last 14 months—ever since the idea popped into my head during a conversation with Bill Gates—and as I've explored it and thought further it seems more and more compelling. I know for a fact that Gates is thinking in this direction as well, because I asked him and he said, “Yes”. Until recently I didn't think, however, he had the whole idea put together as cleanly as I did. Recent events make me suspect he's way ahead of everybody.
As with World War Four, I've discussed aspects of this with many of you. Because I've tended to focus on one aspect or another of the whole picture, and because my views have been evolving, in part due to your valuable comments, I'd urge you to read this document anyway to make sure you see how all the pieces fit together.
In a world where Microsoft is worth more than IBM, it's hard to imagine that less than 10 years ago a substantial part of the financial and venture capital community were skeptical of the fundamental viability of the software business—in fact it was fashionable to ask whether software was really a “business” at all, in the conventional sense.
Certainly, in 1983 and 1984, one could point to Microsoft, Lotus, and Ashton-Tate as profitable, rapidly growing companies, but rare was the company which broadened an initial, seemingly random success into a consistently successful product line. Further, a software company seemed threatened from all sides—from piracy, from next-to-zero-price competition, and from an unsettled distribution environment. Rarely was software even mentioned in reviews of trade shows, and people like me who argued that software drove hardware successes—that the Apple 2 was successful because of Visi-Calc and the IBM PC because of Lotus 1-2-3 were considered more than a little daft.
And yet today, in a world where software companies post profit margins unheard of in most legitimate businesses, the old doubt about the viability of the software business remains, but in another guise. In 1983 they asked, “Can you really make money, long term, selling software?”. In 1993, they ask, “Can anybody other than Microsoft make money, long term, selling software?”.
This is not a facetious question; when company after company which were once pillars of the industry: Ashton-Tate, Software Publishing, Borland, one after another collapses, is gobbled up, or is on the ropes against deadly price-cutting competition, one cannot help but ask whether any software company can truly consider its position secure as long as there is a well funded competitor willing to sell at close to the marginal cost of goods in order to gain market share.
I believe we are in the midst of a fundamental shift in the way software is distributed—a transformation in the relationship between software vendors and their customers fully as significant as the emergence of computer retailing, spurred by the nascent desktop computer. Companies which anticipate this transition, who encourage it and prepare to benefit from it, may find themselves in as unassailable a market position as any company in this century—the “natural monopolies” of the next.
How odd it is, that every day we use the word “program” to talk about the products we create and sell in the software business, and also to describe the information and entertainment we watch on the television and hear on the radio. And even though the word “software” has become commonly used to describe video cassettes, compact discs, and other non-computer material, rarely do software executives see a link between the “programs” they sell and the “programs” they watch. Many managers in the computer software business, and most analysts who follow it, came from a hardware or turnkey systems background; thus it is inevitable that they often think of software as a product like a television, rather than an ongoing service like the “programs” the television delivers.
Okay, I'm going to jump way ahead of myself here in a brazen attempt to grab your attention. If you agree with this analysis, I'm sure you'll summon the strength to trudge through the more deliberate development of the argument that follows.
It's 1997. You call up the 800 number to order another computer, and after you've chosen between the Alpha-III and the Octium chip and the 15 and 30 gigabyte hard drive, the salesperson tells you that the machine comes with the “Basic Package” of Windows NT, Word, Excel, Access, Money, and Multimedia Producer, and asks if you'd like to turn on any additional software at the time. You request Project, Designer, and Visual C++, and they're enabled also. In any case, you're told, “it's all on the CD-ROM, so you don't have to decide right now”.
The computer shows up, and you start using it. Late one night you decide you really need the German language spelling checker add-on for Word, so you put in the CD-ROM, call Microsoft's 800 number to order it on your credit card, and get back the code you enter to turn it on in your smart card. Or if your computer has a modem, you can do it just by clicking the mouse.
Basic costs you $10 per month. “Premium” programs range from $.75 to as much as $100 per month for exotic niche applications. Every month you get your “Microsoft Bill” itemising everything you subscribe to and what it costs. Most folks just have it paid automatically from their credit card, generally 6 months at a time. If you don't pay, you don't get the new authorisations for your smart card and the program stops working. So you pay. Every now and then a new CD-ROM shows up with all the latest updates and upgrades and new products, each with its “try me, buy me!” demo you can run right away. As long as you have Basic, the CD-ROMs come automatically in the mail every 3 months.
Now let's look into the other end of the binoculars; from Bill Gates' chair rather than his customers'. Today, there more than 125 million MS-DOS personal computers installed. Given the rapid adoption of Windows and sustained high sales rate of new machines driven by price performance improvements in new chips, I believe it conservative to expect that 100 million Windows NT machines will be installed 4 years from today, most equipped with CD-ROM, multimedia accessories, and contemporary peripherals; some upgraded from current high-end MS-DOS machines, but most new machines of the Pentium/Alpha generation and their successors. Further, let us assume that Microsoft is unsuccessful in selling any software other than the Basic set (I'm sure you'll concede, based on Microsoft's new product success rate, this assumption is conservative). Well, multiply it out. That's 100 million machines times US$10 per month times 12 months per year, and the answer is: US$12 Billion-with-a-B-like-Bill per year of automatic recurring revenue for which the marketing costs are essentially nil and distribution margin is nonexistent since fulfillment is direct.
Now given an utterly reliable, competitively unassailable annual revenue stream of US$12 billion per year, you can invest in fundamental and applied research, technology development, new product development, marketing, and launch at levels no other player can approach. These investments translate directly into additional recurring revenue to the extent the premium products are adopted by the 100 million and growing installed base, and the proceeds fund further development. In this environment, competitors are forced to either cut prices (and thus their margins), or search for a genuine technological edge and rush it to market before the folks who employ more than 50% of the research people in the field stumble onto it.
Well, Walker's gone right off the deep end again, without even bothering to fill the pool this time? Smart cards, CD-ROMs, expiration of programs, blah, blah, blah. He's made up a whole fairy castle industry out of thin air without the slightest proof that it could even be viable.
And yet, there is something about it that seems oddly familiar…. Let's see…. Aha!!!
Programs are programs. So let's start by looking at what's on the television. No, not the pap on the screen, what's on the television—not the bloody penguin but the little black box the penguin's standing on. Today, not far out in the distant 1997 I was talking about, more than a hundred million people in Europe and North America buy programs—television programs—on a monthly basis. They buy them from a cable television operator or, if they live outside an area with cable service or wish a wider selection, by subscription to a satellite broadcasting system. (Satellite broadcasting still seems a little exotic in the U.S., though that is rapidly changing with the Hughes-Thomson-RCA DirecTV system; In Europe it's everywhere—it's hard to find a home in Britain without an Astra dish—satellites work better in Europe and Japan because they aren't as big as the U.S—a 2 foot dish works just fine, and you can buy the whole rig for about US$300). If you use a satellite dish, your receiver has a little slot where you put in a smart card. If you decide, for example, to subscribe to Turner All-Colourised Movies, just pick up the phone, call the toll-free number, give your subscriber ID from the smartcard, and zap-flash in 30 seconds you're watching Bogey in living—well—pasty colour. This technology is off-the-shelf stuff available in every Radio Shack in the U.S. and any T.V. store in Europe.
Most cable television subscribers pay US$5 to US$10 per month for Basic and monthly fees for Premium services like:
| Low rent cable channels (A&E, BRAVO, etc.) | US$0.79 – 1.00/month |
| Network packages (Denver 5 or Primetime 24) | US$4.00 – 5.00/month |
| Premium Channels (Disney, HBO, TMC, etc.) | US$7.00 – 10.00/month |
and the typical satellite user (who receives all the channels included in Basic cable for free) pays between US$150 and US$300 per year for premium services. Of course there are ultra-premium niche services such as real-time stock and commodity quotes, etc., for which one may pay up to US$100 per month.
You do not buy your television programs, you subscribe to them, and the revenue flows back through the chain to those who manufacture them (have you noticed how often you see “An HBO Picture” in the titles in the theatres?). And if folks pay $150 a year or more for television programs, is it absurd to suppose they will pay $120 a year for computer programs, especially when the cost is in little monthly nibbles rather than $495 up-front the way we do it today, and when you can always rationalise a purchase by saying, “Well, I can always cancel it if I don't like it”?
I believe that soon we're not going to buy computer programs either, we're going to subscribe to them. Programs are programs.
What companies stand out as the huge unassailable (for a while) monoliths of this century? In the United States I'd list:
| American Telephone & Telegraph | 1875–1980 |
| IBM | 1930–1970 |
| Xerox | 1960–1970 |
Now consider that during the time that each of these companies was in a position of total dominance of its market, it delivered its product, which was fundamentally a piece of hardware, as a service, almost entirely on a rental or subscription basis. This, combined with a dominant market share obtained either by getting there first (AT&T/Xerox) or by blowing away less-serious competitors with a massive sales organisation (IBM in computers after 1948), largely insulated the base revenue stream of these companies from business cycles and competitive threats. The annuity-like revenue base, in turn, allowed them to make large, long-term investments in technology relevant to their business (Bell Labs, IBM Research Labs, Xerox PARC) and in product development aimed at further distancing them from their competitors.
Note that in each of these cases the subscription/rental nature of the revenue stream allowed these companies to subordinate technological progress to the needs of the business. Unlike a free-for-all like today's RAM chip or hard disc market, where product generation times are measured in months, AT&T could introduce direct dialing, direct long distance dialing, electronic switching systems, etc. on a decades-long plan geared to optimising their profits. IBM was not forced to rush out the 7094 or 360 under the gun by competitive fears—they could switch their rental base to a new generation at a time of IBM's choosing, when the technology was ripe to increase their revenues and earnings. In short, when a company achieves a stable subscription base, it calls the technological shots in the market. Of course if a company is complacent, it will eventually be knocked out, but you have to be awfully complacent and/or incompetent to nullify the benefit of a 10 to 1 advantage in product development and marketing resources (Xerox, of course, demonstrates that it can be done, but in the other cases it took government action or fear of government action to displace the dominant player).
Redmond, Washington, March 20, 1992. Transcript of meeting of John Walker and John Forbes of Autodesk and Bill Gates and Todd Needham of Microsoft.
| WALKER: | So let me see if I understand where you're going with this, Bill. What you'd really like is if in, say, five years, everybody with a computer gets a Microsoft bill every month, just like a telephone bill, for each product they use. |
| GATES: | Precisely. |
Have you heard of the “Microsoft Developer Network”? I'm a member. Walk around and ask random programmers if they are as well (either company-paid or on their own account). There are 45,000 members as of the last time I looked, and I suspect the ranks are swelling rapidly. If you're a remotely serious Windows developer, you simply cannot afford not to join, because it's the only effective way to receive massive amounts of source code, internal technical documentation, beta copies of soon to be released Microsoft products, special development, debugging, and authoring tools, etc., etc.
How does it work? Well, it's a, er, subscription. I mail in a check for US$200 per year, and every 90 days they send me this, er, CD-ROM filled with 600 megabytes of ever-changing goodies. And as long as my subscription is current, the bits just keep on coming.
Have you noticed how Microsoft update marketing works these days? They've pretty much dropped even the pretense of running updates through the distribution channel except for mega-million blowouts like Windows 3.1 or DOS 6. No, as long as you've registered the product, right about the time the new release hits the cover of PC World, a little letter from Bill shows up with their little blue OCR form. You just tick the box and attach your check or credit card number for a readily-digestible fee (usually US$30 through US$150) and bung it in the post. A week later the update shows up at your door. For those who update faithfully this way (and I won't get into the many ways Microsoft forces you to stay current or pay a heavy price—they're masters at it), this amounts almost to a subscription—not on a consolidated basis like cable or satellite TV, but to a product, like a magazine subscription. And the economic dynamics strongly resemble those of magazine subscriptions.
Magazine subscriptions are typically sold through retail channels and/or highly discounted for an initial subscription; publishers don't usually make money on initial subscriptions. Renewals, however, are handled by direct marketing and all the recurring revenue for the subsequent years goes right into the publisher's pocket. It's the same for Microsoft Word; Gates has every incentive to discount initial sales of Word as much as possible to gain market share against Word Perfect, and to grant all kinds of incentives to his channels as long as he believes that each initial sale plugs him into a virtually guaranteed revenue stream of, say, $50 per copy per year. Of course to realise that revenue he has to keep the updates coming and provide enough added value in each one so people continue to stay current. But since I've never seen a wish list get shorter for any software product I've worked on, I hardly think that's a problem as long as you're willing to invest in development.
Microsoft, 2 Cable Giants Weigh Interactive TV Venture By John Markoff
New York Times Service International Herald Tribune, Monday, June 14, 1993NEW YORK—Three dominant technology and entertainment companies are on the verge of joining forces to create the equivalent of software for cable television—a system that would combine the worlds of computing and television and perhaps shape how much of popular culture is delivered.
Time Warner Inc., the largest entertainment company, Tele-Communications Inc., the largest cable television company, and Microsoft Corp., the largest software company are expected to announce by the end of the month that they will form a company, tentatively called Cablesoft. The companies hope the new venture will lead the way in establishing a standard for the transmission of a coming generation of interactive programs.
At stake is control of the unobtrusive cable box that sits atop many television sets. In recent months the box has become a battleground for computer, telephone, and cable companies.
…
Last month, for example, Intel Corp., the world's largest chipmaker, Microsoft, and General Instrument Corp. announced plans to develop a cable converter that would have a built-in personal computer. Last Monday Time Warner announced that Silicon Graphics Inc., a Silicon Valley computer maker, and Scientific Atlanta, a supplier of cable boxes, would supply hardware and software for its digital television trial in Orlando, Florida, which is scheduled for next year. A day before that announcement, Kaleida, a joint venture of IBM and Apple Computer, said it was joining with Motorola Inc. and Scientific Atlanta to develop a similar futuristic television controller.
…
Need I point out that just as soon as your cable box talks to your computer in almost any fashion whatsoever, the technological means exist to make subscription software as secure as subscription television. And the pirate TV decoder business seems to have been roundly wiped out.
The Time Warner/Tele-Communications/Microsoft deal is, of course, something that's looking out a few years and probably genuinely focused on interactive television. But purely as a side effect, something that just falls out, is the means to distribute and authorise subscription software world-wide. Controlled by Microsoft. This could have implications outside the entertainment world.
Ever since 1985 when I first I proposed the “AutoCAD Professional Subscription” as a means of finding the holy grail of recurring revenue, I have been following the evolution of the software business from a retail sales model to a subscription/service base. When I look at the convergence of the trends and events I've noted above, I cannot help but believe we are, if not already in that era, at least on its threshold.
Emerging from this period of transition with a large, stable, and growing subscription base will render the companies who succeed formidable, almost invincible, competitors compared to firms with smaller market share forced to generate their revenue entirely from new sales. Reinvestment of a stable revenue base can, if done wisely, further widen the gap between the dominant firm and the dwarves.
For some reason, when I discuss the subscription model of software distribution, many people get confused and think I'm talking about something in the medium to far future—“Well, yes, things may indeed go that way once we have interactive television/fiber to the home/data highway/…, but for now….”. But other than the Microsoft cable box venture, which is interesting but unnecessary, nothing I have discussed has any technological contents whatsoever; it is purely a question of marketing and distribution strategy. Bucks, not Buck Rogers.
To summarise and demonstrate that all the foundation pieces exist:
Unlike many major transitions in distribution strategies, moving toward a subscription model can be done, as far as I can tell, with little or no risk (effort and expense, yes; risk, no). In a business which concentrates primarily on new sales, the installed base is often an underperforming asset waiting to be discovered. Moving to direct marketing of “frequent, cheap” updates and upgrades, as Microsoft has done, is unlikely to alienate existing channels geared to selling new products. As Autodesk has discovered, as long as we keep new sales of AutoCAD firmly in the dealer channel, providing direct options for “the little stuff” may provoke grumbling, but seldom more grumbling than we hear about “unprofitable, time wasting update business”.
I believe a subscription strategy can be evaluated and planned relatively simply once we discover the answer to the following question:
How much revenue do we generate, per annum, from the average unit of AutoCAD, after its sale?
Now, I don't have the vaguest idea of this number, but let's play a little napkin engineering and make a wild stab. The wild and wooly R12 update generated $22 million in update revenue and the subsequent quarter $13.5 million (Pru-Bache report, May 24, 1993). Let's assume we hold the $13.5 million level (which Laura and folks don't expect, but I want to err on the high side). So we have $63 million in update revenue, liberally construed, in a year with a blockbuster update. Folding the napkin and continuing, we have about a million installed copies, but let's say they're, oh, 700,000 “active” copies, disregarding shelfware and people who haven't upgraded since Version 2.6. Well, that comes out to about US$90 per year per “active” copy. So, for example, if we could get half our “active” users to subscribe for, say US$250 per year, we would have a recurring revenue stream greater than our biggest update year ever, and without all the push and cost it takes us to launch an update. And given what Autodesk usually charges for updates, many users would probably consider this a bargain, particularly if it avoided all the hassle currently involved in updating a copy of AutoCAD.
It's also intriguing to divide the Pru-Bache FY 94 estimate of US$430 million by my “active base” of 700,000. That comes out to US$614 per active unit per year. So were we, for example, to move to a subscription for AutoCAD of about $100 per month for new sales, we would generate, month after month, year after year, revenue equal to what we largely derive today only from new sales—again assuming only 50% conversion of the already active base. This implies a more radical change in the way we do business which could be deferred until experience with the installed base upgrade/update program confirmed its viability, or simply put off forever, retaining different channels for first sale and subscription as in the magazine business and Microsoft's current practice.
If you've made it all the way through my arguments without either getting lost in my thorny prose or storming away in violent disagreement with my premises or deductions, you may be beginning to think I'm onto something here. But that raises the legitimate question, “Well, if even Walker's figured it out, why isn't everybody in the industry already doing it?”. Indeed, I've wondered quite a bit about that myself.
I think the answer lies in the observation that most companies who succeed in building self-sustaining subscription-based businesses start from a position of effective monopoly of their sector. In the case of AT&T, it was a combination of technology, patents, and government grants which conferred the monopoly. IBM built its first monopoly in tabulating equipment on the patent of the Hollerith card, then clawed its way to an effective monopoly in computers by out marketing and out-customer-servicing Remington Rand, Burroughs, and others. Xerox derived its monopoly from the patent on xerography.
Quite simply, to derive enough revenue from a subscription strategy to make the business run, you have to have the lion's share of the market, not a small slice. To get people to subscribe, you have to have demonstrated technological leadership that convinces them they'll get more value by paying you regularly than buying from somebody else outright, then replacing the product later on. And of course the central development engine needs to be big enough to keep generating the value that gives subscribers value for their money, year in and year out.
Which means that to pull off the transition to a subscription base, you have to start with a large market share lead, and therefore the only companies in the software business well-positioned to do this today are:
Microsoft
Autodesk (in CAD)
Word Perfect (in word processing, but slipping)
Lotus (in spreadsheets, perhaps, and slipping quickly)
Is software a business? Can a company make money selling software, consistently and reliably? In 1983 many people doubted that these statements were true. In 1993, we have one great success story but little confidence that other sectors and companies are safe investments for the long term with predictable chances for growth.
By 2003, I believe that everybody will know the answers to these questions: “Yes, and yes”. Within 10 years the software industry will have restructured itself from a costly and unpredictable bookstore/appliance dealer sale-oriented model to a cable TV-like subscription model. The companies who emerge from the turbulence of this transition will be the colossi of the industry, no more and no less inherently risky than television networks, book publishers, or regional telephone companies. Their revenues, measured in the billions to tens of billions will fund ongoing product development aimed and increasing their subscription base.
Ironically, they may cease to be viewed as “growth stocks”—once the constant revenue base comes to eclipse the near-term potential of a new product launches, their performance may appeal to those who buy utility stocks today. (But then when electricity use was growing exponentially in the early '50s, utilities were “growth stocks”.)
It's 2003. The little black box on the top of the TV is hardly big enough to hold all the logos printed on it: Microsoft, Time-Warner, Swiss PTT, SES/ASTRA, General Instrument, Dolby Labs, Intel, Motorola/Iridium, etc. A couple of wires hook it to the TV and the computer, and the ubiquitous smart card sticks out the front. Every month I get a bill for the programs I subscribe to:
| Astra Basic pack | 2.00 |
| Eutelsat Basic pack | 2.00 |
| HBO | 7.00 |
| The Dinosaur Channel | 0.50 |
| Canal Plus | 1.25 |
| TeleCine Romande | 4.00 |
| CNN International | 0.75 |
| Microsoft Basic pack | 4.00 |
| Microsoft Developer Network | 10.00 |
| Microsoft Project | 2.00 |
| Microsoft Visual C+++++ | 9.00 |
| Microsoft Producer | 4.00 |
| Autodesk AutoCAD | 75.00 |
| Autodesk Cyberspace Explorer | 8.00 |
| EuroFeed Internet News Link | 1.00 |
| BBC World Service Radio | 0.25 |
| … about 20 more … programs are programs |
And since the bill gets paid automatically, and it's only about 200 francs a month, I don't look at it too closely (other than that AutoCAD thang—wonder if they'll ever give me a break!). Each little nibble is so small, though, compared to when I had to shell out US$400 over the counter for some software I didn't know would even work when I installed it, that I pay and pay and pay.
This document is a work of education, not advocacy. I believe that wisdom consists of embracing change, not battling it. We win by enlisting the slow but inevitable forces of economics on our side, not by shoveling sand to halt the tide. All the evidence I can see convinces me the software business is finally making the transition to a recurring revenue model which will be its salvation. This transition will entrench those companies who leverage their large existing market share into a consistent and reliable base of recurring subscription revenue.
Among major software companies, Microsoft and Autodesk (in the CAD sector) are uniquely positioned to lead this transition and benefit from it.