Skip to main content

Investing in Enterprise Software

By March 18, 2013Article

In 1969, under incipient assault from the Antitrust Division of the U.S. Department of Justice, IBM unbundled software and an independent industry was born. Over the following decades, a host of startups were founded to deliver systems software and applications to enterprise customers. The enterprise software industry thrived through the great Bubble of 1998-2001, enabled by two critical factors: a unique revenue model and continuing access to the public equity markets.
My own active investing career encompassed these years, as recorded in my recent book “Doing Capitalism in the Innovation Economy.” My colleagues and I and the entrepreneurs we backed were all beneficiaries of this industry’s distinctive attributes.  Today, the economics of investing in enterprise software requires radical reconsideration in light of the concurrent rise of the Software-as-a-Service revenue model and the effective closure of the IPO window for more than a decade.
The origins of the “perpetual, non-exclusive license plus annual maintenance” model are lost in the mists of time (one story gives priority to Software AG’s successful marketing of its Adabase in the 1970s). Whoever were the pioneer vendors and customers, for some 30 years this mode of selling endowed the industry with a powerful financial privilige. In abstract, mathematical terms, enterprise software companies were able to get paid for the estimated integral of all the future value that customers could be convinced they would receive from deploying the code. In hard cash terms, this meant that the rich enterprise customer was financing the poor startup, thereby reducing dollar-for-dollar the quantum of expensive equity required from venture capitalists. No wonder that the successful enterprise software salesperson came to be instantaneously recognizable by the gold Rolex sported on the wrist.        
The perpetual license model accelerated cash flow to the emergent startup, bringing closer by years achievement of that equally golden bridge to “positive cash flow from operations” which, as I was educated by my remarkable mentor Fred Adler, constituted “corporate happiness.” And, in turn, early achievement of positive cash flow not only reduced the cumulative burn that had to be financed by venture capital. It also provided two more economic rewards: a hedge against any transient hiccup in access to the IPO market and an “outside option” when approached by a possible strategic acquirer. The happiness conferred by positive cash flow meant that management and owners could afford to wait on the IPO market and to play tough from a position of financial self-sufficiency vis-à-vis potential acquirers.
Of course, dependence on large up-front license sales carried with it both volatility in reported results and a consequent temptation to push the envelope on revenue recognition. Both these factors entered into the valuation methodologies of venture capitalist and public market investors alike. But the positive economic and financial benefits, backed by strict governance enforced by inside investors with skin in the game, were compelling.  For 20 years they combined to limit the amount of risk equity capital needed to fund a well-managed enterprise software company to no more than about $20-30 million of mid-1990s dollars. They contributed powerfully to my own primary focus on enterprise software.
At a fundamental level, there is only one question to ask about the rise to dominance of the SaaS model in the enterprise software market: what took the customers so long? In those same abstract, mathematical terms, vendors of enterprise software are now receiving the derivative, the incremental value that the customer is actually receiving from running the code, whether the calculation is made based on number of active users per unit of time or on a transactional basis. I well recall reading the prospectus for Software.com’s IPO in 2004 and realizing that it had burned through more than $100 million of venture funding to reach the threshold of positive cash flow.
A partial offset on the cost side has emerged in parallel with the rise to dominance of the SaaS model. The proliferation of open source tools and utilities, along with the ability to rent processing and storage resources as required from Amazon and other vendors of cloud services, has radically reduced the cash required to launch a software project. Hence the endless supply of social media wannabes, projected onto the Web with minimal up-front investment. But if the target is the enterprise market, the startup will require a direct sales force with accompanying marketing and customer service teams.  A dedicated go-to-market channel cannot be outsourced, raising the capital needed and deferring achievement of positive cash flow.
Once that state of corporate happiness is reached, the public market’s embrace of the few, new-model enterprise software companies to have completed an IPO is both rich and well deserved. It is a function both of the radically easier challenge of selling on the SaaS model and of the increased predictability of results over the relevant investment horizon. But getting there typically requires recruitment of successive rounds of follower VCs whose returns on the marked-up valuations they face will depend on the step-function mark-up to public investors through that critical IPO. And here is where the second change in the environment bites.
The presumption that the IPO market would be there persisted from 1983 through the bursting of the Bubble in 2000. Since then, only in 2004 did the number of IPOs for venture-backed companies of all sorts exceed 20 per quarter, the threshold for a “normal” IPO market that my academic collaborator, Professor Michael McKenzie, and I derived from detailed analysis of the IPO market since 1980.
From the onset of the Global Financial Crisis in 2008, total venture-backed IPOs have rarely exceeded 10 on a quarterly basis. Moreover, the scale of the median IPOs has remained at the elevated level — more than $100 million — reached during the frenzy of 1999-2000. Driven in good part by the coincidental, radical consolidation of the investment banking industry, the simple math of IPOs means that to have a hope of going public a company must be generating on the order of $100 million in revenue.  This must be a particular challenge for the enterprise software company executing the SaaS revenue model.
The dominance of the SaaS revenue model and the closure of the IPO window combine to generate extreme frustration in the mind and temperament of this theorist-practitioner of capitalism at the frontier of technological innovation. For there need be no fear that entrepreneurs of enterprise software and the venture investors who back them have run out of opportunities. Two stand out: the emergence of the fourth generation of computing environments and the belated integration of the two principal strands of enterprise computing, transaction processing and analytics.
Over three decades, enterprise computing progressively moved from (1) mainframes accessed by dumb terminals through (2) intelligent clients interacting with a range of specialized servers to (3) millions of PCs transacting with hosts of servers across the Internet. Now it is apparent that “mobile/cloud” represents a fourth generation of architecture. Like the ones that preceded it, the cloud/mobile environment will require all the “-ilities” that users of the original mainframes came to take for granted: reliability, availability, scalability and secure(il)ity.
The apotheosis of my own career as a venture capitalist at Warburg Pincus came by way of backing the extraordinary entrepreneurs who built BEA Systems into the source of this technology when the Internet emerged as a medium for commerce. The technical challenges of the delivering the –ilities to mobile/cloud will no doubt be even greater than those demanded by the commercial Internet. The rewards will likely be commensurate. (Full disclosure: I am an investor in Magnet Systems, whose founder and CEO is Alfred Chuang, the “A” in BEA.)
In parallel with deployment of the mobile/cloud architecture, the two styles of enterprise computing are finally coming together. Online Transaction Processing has been the core of enterprise computing since IBM introduced its Customer Information and Computing software (CICS) some 40 years ago. Orthogonal to OLTP computing, analytics — the extraction of information from archived transactional data — was a batch-processed, offline activity. The technical architectures and business cultures of each were distinct and grew to be deeply rooted. These silos are now dissolving.
As might be expected, Wall Street was the pioneer in applying analytics to the real-time streaming data generated through computerized trading systems, seeking opportunities for profitable arbitrage however transient. But the generation of Big — and overwhelmingly unstructured — Data through the multiple media of the social network has added a new challenge and created an enormous new opportunity.
Advances in the core of computer science and in statistical methods are making it possible to read unstructured data for meaning and intent. (Full Disclosure: Warburg Pincus is a substantial investor in and I am a director of Nuance Communications, a leader in natural language understanding technologies.) All enterprises that address large customer communities — not only the pioneers of the online world — are now challenged to integrate these new analytical capabilities with the patterns of actual transactions their OLTP systems capture. The power of such real-time, predictive analytics is just emerging. Its consequences will be huge.
So today, the enterprise software industry and those who invest in it face multiple challenges: capital requirements that have increased three to four times with the demise of the perpetual license model and the absence of that presumption of access to the IPO market needed to incentivize late-round investors. And with those challenges come opportunities at least as great: to participate in two transformational revolutions.  Rendering the mobile/cloud environment enterprise-capable is the first of these.  Delivering the promise of predictive analytics operating on Big Data, enterprise domain by domain, is the second.
As usual at the frontier of the Innovation Economy, vision and guts are required in equal measure.
Dr. Janeway’s book, “Doing Capitalism in the Innovation Economy,” is available on Amazon.com.
Dr. William H. Janeway is managing director and senior advisor at Warburg Pincus. Prior to joining the firm, Dr. Janeway was executive vice president and director at Eberstadt Fleming. Dr. Janeway is a director of Magnet Systems, Nuance Communications and O’Reilly Media and Roubini Global Economics. He is also chairman of the Board of Trustees of Cambridge in America, University of Cambridge and a founding member of the Board of Managers of the Cambridge Endowment for Research in Finance (CERF).