Cachet Technologies $50K CONTEST: MAY 1998 Danny Lewin, Jonathan Seelig, and Scott Tobin walked down Memorial Drive just out- side the MIT campus in Cambridge. The three now were thoroughly dejected and won dered what to do next. Their thoughts tumed to the outcome of the 1998 MET $50K tition that had just finished. Each year, the contest drew the best y products and services at the university. The audience had been ful of venture capitalists, and they knew that a win in the competition would have meant a [ot in terms of obtaining financing. Lewin and Seclig thought that they had asse1 bled one of the most compelling new business propositions with their entry, Cach ‘Technologies. The company proposed to establish a potentially revolutionary way to dis- tribute content over the Intemet. Lewin, an Israeli Ph.D. student who had worked for IBM in Haifa, was studying under Professor Tom Leighton, a world-renowned scientist in the Laboratory of Cor puter Seience (LCS) at MIT. Lewin had seen an adve for the business plan contest in the fall and had hopes of being able to pay off some of his mounting debts. ‘Considering his wife, family, and his substantial student loans, the $50,000 prize would have gone a long way to defraying his expenses. Lewin had brought both Seelig, a first- year Sloan MBA student, and Tobin, an associate at Battery Venture Partners and a long-time friend, into the project. A team of computer scientists had made tremendous progress on a prototype, and the team had identified potential partners and customers, Evidently, they were misguided First prize in the $50K Competition was shared by vo firms: Direct Hit, a co pany that had been established to develop software that improved the search bility of Internet search engines, and an Internet nonprofit named Volunteer Cor modity Connection, The rmner-up was Car Soft, a company that produced software that would enable individuals to run diagnostics on their automobiles by connecting a personal computer to their automobile’s computer. ig eapa- esearch Asadate Howard Reitz prepared this cave under the supervision of Professor Pal Campers as the ass for elas discussion rather than to state ether effective or ineffective baling of an administrative station, Copyright © 2000 by the President and Fellows of Hanard Calle. To oder copes or request permission to repronice materials, ell -W0-545-7685, write Business School Publishing. Boston, MA 02163, ‘or goto htp/swwhhsp harvan edu. No pat of ths publication may be reproduced, sore in retriesa system, used ina speaisheet, or transmitted in any form or hy ay eans-—electro, mechanical phetocopying. recoding, or otherwise—ithout the permission of Harvard Business SeleBA + Chapter 4: Cachet Technologies How could a company like Cachet Technologies that had developed a revolution- ary new product to speed content around the Internet have lost to a company that hal developed merely an evolutionary new technology to speed up Internet searches such as Direct Hit? Losing to a nonprofit made it even worse, ‘The three friends walked along, recalling their experienee and cursing in Hebrew, asking themselves: Why didn the venture capitalists and entrepreneurs judging the contest understand the potential of Cachet Technologies? What had they not foreseen prior to the competition? Was Ca- chet Technologies’ software worthy of being commercialized? Should Lewin and Seelig just call it quits? Peshaps most important, what did they need to do to determine the answer to these questions? Lewin and Seelig had attractive offers to pursue forthe suan- mer and Tobin needed to continue sereening deals and making investments for Battery Venture Partners, Maybe they would all be better served by pursuing those other ‘opportunities? WORLD WIDE WEB The rapid evolution of the World Wide Web (WWW) in less than a decade: had been well documented by the late 1990s. The development of the programming language known as hypertext mark-up language (HTML) in 1992 by Tin Berners-Lee enabled content to be displayed easily and accessed universally. Previously, the Internet hacl been used primarily as a means of transferring files. However, the development of HTML enabled programmers to create individual Web sites featuring the rich content and multimedia applications common in many complex Web sites. The cre vanced graphical user interfaces and commercial browsers enabled personal computer users for the first time to access a broad range of information from a variety of sources, facilitating the rise of Internet, ushering in the possibility of e-commerce, and chang ingcforever how individuals transacted business. ‘The promise of the Internet was that it represented a global, interactive, and trans active new medium, No other distribution channel had those qualities. First-generation Internet sites were often one-dimensional, static, and lacked interaetivity. They did not actively engage users, nor did they exploit the information revealed by each user dur- ing every session, Second-generation Web sites promised to be more dynamic and en- abled, leveraging the interactive, and transaetive nature of the new medium. This re- quired a substantial investment in Internet infrastructure in order to handle the expansion in bandwidth needed to process this additional information. As recently as 1995, the Intemet liad only 5 to 10 million pages of content. By the year 2000, hos ever, the Internet was projected to have as many as one billion Web pages.” ‘As users moved to more sophisticated multimedia applications, the existing rout- ing system on the Internet had become overloaded. This created a delay in the system or the feeling of slow Web page access during peak periods, even for individuals with high-speed access lines. Event-driven supersites accessed by broadband placed a sys- temic load on the network that caused a spike in usage, dropping information packet and sometimes causing outages. Several examples of this phenomenon included th tempt to distribute the Starr Report, Victorias Secrets promotional event during the Super Bowl, and the heavily trafficked Heaven's Gate’s Web site after the cult made the his critical problem, called the “hot spot problem” by Bemers-Lee, was n of ade " sephen Mabey, “Tlkn’ hoot « Net solution,” Salomon Smith Barney: December 8, 1909, p. 20 2 thid,p. 2,Current Solutions + 55 caused when too many users accessed the system during peak periods or during a spe- cial event and reduced user access. The costs to companies were even higher with erit- ical applications like online auction, e-commerce, and trading sites. New applications ‘were planned to be introduced into the marketplace that would place even greater loads on the existing network, including advanced streaming media, voice-over-the-Internet, Virtual private networks, and Internet roaming services. CURRENT SOLUTIONS Intemet service providers (ISPs), that is, the companies that actually connected cus- tomers to the Intemet, and other Internet infrastructure companies, were addressing the problem of net congestion in a number of ways, including building out their sys- tem to accommodate larger network loads ancl ereating server farms that enabled them to host content for their customers. These investments in infrastructure were promi ing and were fueling the dramatic growth in new Internet infrastructure companies. ‘They were, however, very expensive and time consuming to implement. In order to rake these investments, Internet infrastructure companies had to make significant eap- ital expenditures. Many of these companies, however, earned relatively low returns. [See Exhibits 4-1 and 4-2 for detailed information on ISPs.] Furthermore, Internet service providers expected substantial competition in the future, Once the large teleos gained entry into the long-distance market, they would be able to offer bundled services of lo- cal and long-distance phone service as well as Internet access to their customers. This created a market for a whole new class of Internet service providers. ‘The most rapidly growing segments of the market were companies that provided hosting services to con tent providers One of the most promising solutions to net congestion was mirroring, whereby the entire content of a particular Web site would be replicated on a number of different servers at different locations. Mirroring companies provided their services to bandwidth intensive, e-commerce companies that were looking to improve both the speed and re- liability of their Web sites. Mirroring enabled content providers to spread out their traf- fic over multiple servers at different locations. It was, however, costly, time consuming, and inefficient, Content providers were required to duplicate the entire site at every server location, regardless of whether all of the information was requested by a user A mirroring company made its snargin by purchasing large volumes of bandwidth, n gotiating discounts of as much as 30% to 50% and reselling bandwidth to content providers. Early purchasers of mirroring services were bandwidth: intensive online aue- tion, e-commerce, and trading sites, Investment analysts believed that these companies ‘would continue to outsource their content delivery requirements as their business panded because no one could do it a cheaper than mirroring companies:? Companies such as Alteon, Bright Tiger, F5 Labs, and Resonate were developing software and hhard- ware that would help keep sites synchronized and load balanced. These solutions, how- ever, did not provide large a replication. Although the software solution was helpful to content providers, it did not address the issue of scalability. It still cost more than twice as much to maintain two sites as it did to maintain a single site. [See Exhibit 4-3 for information on the financing of competitor firms] Another promising solution to net congestion was caching. Caching involved the temporary data replication and storing of a Web site on servers that were closer to the Dab, p22