A fair question to ask in any investment outlook for software is the most basic one: Can there be new, breakout investments in a category that has matured, consolidated and evolved past the frenzied days of pervasive adoption? We remain convinced, but with substantial biases, that there will continue to be opportunities to build highly successful venture- backed companies in both application and infrastructure software. However, we think there have been some fundamental structural shifts in the software business that make the current “back to basics” conventional wisdom nave and in some cases pernicious. Given these shifts, we believe that there is an attractive and appropriate investment model for software companies going forward, one that is more capital intensive than you might expect. And, despite over-investment in some segments, we expect several segments to provide fertile ground for new breakout investments.
In the early 1990s, a common first meeting venture question was: “How many lines of code do you have?” This is surprisingly still fundamental to the triangulation analysis that includes other software basics, such as the number of people required to support the product. There is a meaningful historical precedent for the current phase of the adoption cycle for enterprise software. However, the often-heard phrase “back to basics” is simplistic. The next few years have strong analogies to the software industry during the client-server transition of the late 1980s and early 90s, when a relatively small number of software companies thrived in the infrastructure layer by providing tooling and deployment platforms for the emerging client-server ecosystem. Over time a second wave of software companies grew, providing application layer functionality that leveraged the new architecture.
A similar pattern can be seen today with Internet-based applications. Startups focused on tactical, ROI-oriented infrastructure to harden and automate the early customer built Internet applications are doing relatively well. For most six- and seven-figure license purchases, our experience indicates that leading corporations require an eventual cost savings of at least 10x annually. Plain, old-fashioned volume purchases in exchange for larger discounts are again the norm. Clearly, there is a renewed focus on capital efficiency and deeply defensible IP, and the investment community is backing more experienced entrepreneurs.
There are some key differences in both the revenue and cost side of the software business that require fresh thinking by investors and entrepreneurs. For example, there is often a fundamental disconnect between what customers will pay for software licenses and services vs. the classic enterprise direct sales cost model. Today the average deal size runs in the $150,000 to $250,000 range, in contrast to a historical range of $500,000 to $1 million. This deal size cap is likely to continue, given both the cost management pressures on enterprise customers and the emerging strength of internal IT organizations as a price stocking horse for purchased software applications.
The need to re-evaluate the fundamental assumptions behind the enterprise direct sales strategy creates an opportunity for startups vis-a-vis incumbents. Startup software companies are free to reexamine everything from sales force compensation to identifying one to two meaningful business development partners. Public companies often have problems re-casting their business models: They are constrained by their existing sales model and customers. Incumbents also have problems recasting their product platforms, both in terms of selling it in discrete modules vs. all-or-nothing suites and in terms of the software itself being developed in a more monolithic way. This makes it harder for the incumbents to easily parcel out discrete software modules for development offshore in the same way that a startup can tap into lower cost development for incremental, “niche” modules. With more fixed costs of doing business and capped/declining deal sizes, the incumbents inevitably face a margin squeeze while startups are free to innovate around both the cost and selling assumptions.
Another “back-to-basics” theory-that infrastructure is a good starting point layer for new software companies-has resulted in subcategories that are flooded with investment capital, such as storage software, systems management, security and Web services/integration. Unlike certain networking infrastructure markets, these types of software markets have shown that the best-case outcomes for a category are one to three successful companies. There are several structural reasons for this. Many of these categories are “niche” markets or capped in size at a few hundred million dollars vs. billion dollar-plus markets. Historically, there have also been a very limited number of strategic buyers for these software companies, so if you are not the No. 1 or 2 player with a shot at remaining a standalone public company, there is no one to bail you out. Not a single one of our top 10 software investment returns was the result of an acquisition.
One of the commonly held beliefs of “back-to-basics” thinking is that the financing model will revert, almost instantaneously, to the pre-bubble days. Proponents of this line of thinking believe that successful companies that meet their milestones will be able to go out to the pools of available capital and raise money at higher and higher prices as risk is reduced-just as they were able to do in the mid-1990s: They would also have you believe that those companies should get to cash flow positive with significantly less venture capital-say $10 million to $15 million-like in the “old days.” That would make the returns attractive, assuming $150 million or greater terminal values.
Not So Simple
While Accel sees merit in this “back-to-basics” ideology, we also see wishful thinking. In early 2002 we revisited the financing model with a clean sheet approach. We started with an assumption that not only have the sources of capital changed from the mid-90s (i.e., mezzanine investors, corporate investors and debt have all but evaporated), but the capital requirements of venture-backed software companies have also changed. We went back to our best entrepreneurs and our public companies and asked fundamental questions: How much does it cost to build a direct sales force? A core engineering team? A global services organization? What do current and projected sustainable revenue ramps look like?
Old Days’ Are History
Based on the feedback, we determined that, for the most part, venture capitalists won’t be able to build successful software companies for $10 million like they were able to in the “old days.” Our analysis indicates that the majority of high-potential software companies will require $20 million to $30 million to achieve sustained cash flow positive operations going forward. (This is not to say that there won’t be attractive companies that find ways to breakeven for less.) A significant assumption in our conclusion is the need to aggressively invest early in international sales and distribution. There is no “accidental-tourist” strategy that can fully capitalize on the enormous opportunities that exist in Asia and Europe. Japan alone represents enormous growth and more than a quarter of projected new license bookings in 2003 for several of our best-performing enterprise software companies.
For each new investment, we modeled three scenarios of capital intensity, interim-financing valuations in those scenarios and terminal values at early and mature stages of a company’s life. Terminal values are, of course, highly uncertain, but we assume that a category-leading company will trade at 2x to 5x forward revenues and have a 20 to 30 price-to-earnings ratio. Each new investment opportunity needs to demonstrate consistency with that model. Most do not. In those cases, interim financings do not support up rounds, increased capital risk is not rewarded with greater terminal values and the list goes on.
In our current environment, Accel has a two-part strategy. One, work from the top down to understand where major capital flows have occurred (in most cases a huge negative). Two, work from the bottom up with the goal of continuously developing a prepared mind of domain knowledge, relevant experience, and industry relationships in areas of high investment potential.
Today’s proactive software themes and some selected highly interesting private companies (from the portfolio of Accel and others) include the enterprise data center (Cape Clear, Rhapsody, VMWare, Webcohort and Wily); extended enterprise applications (Apexon and Motive Communications); product lifecycle management (Datasweep); and business process outsourcing (BrassRing and Salesforce.com).
Many of the most successful infrastructure software investments will be at the intersection of networking, computing and software. One of the best opportunities for building new categories of infrastructure software is the enterprise data center. It is the locus for new application development, running commodity Linux computers, J2EE application servers, SOAP, XML, and it is connected to the extended enterprise.
We also believe that we are in the early stages of adoption of extended enterprise application software. These companies are made possible by Internet technologies: They are inherently cross-system and cross-organization, combining basic integration with well-defined business processes and analytics. Capturing core product and production data and providing sophisticated business analytics is fertile ground.
Finally, in our software industry investment analysis, we put a disproportionate weight on the founding CEO. Our overwhelming empirical evidence is that for software companies the financial outcome is highly correlated with the ability of the founder to lead the company from inception well into public market success. We are continuously attempting to build as clear a profile as possible of a founding CEO’s success traits. These include technical depth, an innate feel for the collision of technology and customer needs and passion for providing extraordinary value to customers.
In conclusion, we do not see a fundamental improvement or shift in the IT spending outlook for 2003. Our long-term outlook, however, remains positive for building new standalone software companies that capitalize on core market fundamentals-as long as the emerging companies take a sober, creative and fresh look at both revenue and cost models.
Jim Breyer has been a venture capitalist for 15 years and has been a general partner with Accel since 1990. He has invested in more than 25 software and infrastructure companies that have gone public or been aquired, including Actuate, Agile Software, Foundry Networks, Macromedia and Walmart.com. Breyer sits on the boards of Wal-Mart Stores and RealNetworks.