The Broadband Inflection: Why Network Effects Will Remake Everything

The infrastructure buildout is complete. After burning through $2 trillion in capital between 1996 and 2001, after the spectacular implosions of Global Crossing, Worldcom, and hundreds of competitive local exchange carriers, after laying enough fiber to circle the earth 1,500 times, we finally have something to show for it: widespread, affordable, always-on internet connectivity.

The numbers tell the story. Broadband penetration in US households crossed 20% in the fourth quarter. Cable modem deployments are accelerating faster than anyone projected—Comcast alone added 735,000 subscribers last quarter. DSL, after years of telco foot-dragging, is finally gaining traction as incumbents realize they can't cede the entire market to cable. South Korea, our bellwether for what comes next, sits at 70% penetration.

This isn't just faster internet. This is the foundation for an entirely different computing model, and the companies that understand this distinction will build the next generation of franchise businesses.

Why This Time Is Actually Different

The skepticism is understandable. We've spent two years nursing wounds from the bubble. The Nasdaq sits at 1,350, down 75% from its peak. Pets.com, Webvan, eToys—the cautionary tales write themselves. "Infrastructure doesn't matter without sustainable business models" became the conventional wisdom.

But conflating bad business models with irrelevant infrastructure is precisely the wrong lesson. The dot-com era wasn't wrong about the future—it was wrong about the present. In 1999, when Kozmo.com was burning venture capital on one-hour delivery, only 3% of US households had broadband. When Webvan imploded, the infrastructure for profitable online grocery delivery simply didn't exist.

The constraint wasn't imagination. It was physics and economics.

Consider what dialup actually means as a computing platform. A user connects, waits for a page to load, disconnects to free up the phone line. Average session length: 29 minutes. The mental model is fundamentally transactional—you go online to complete a specific task, then you leave. Building businesses around engagement, around real-time interaction, around making the internet a default layer of existence—these were non-starters.

Broadband changes everything about the user posture. Always-on connectivity means the internet stops being a destination and becomes ambient infrastructure. You don't "go online"—you're just online. This shift in user behavior is what enables entirely new categories of application.

The Platform Economics of Real-Time

The most important implication of always-on connectivity is the economic viability of real-time, multi-user applications. When users are persistently connected, when latency drops from seconds to milliseconds, you can build businesses around synchronous interaction.

Look at what's emerging in the early adopter markets. In Korea, where broadband penetration hit critical mass two years ago, online gaming isn't a niche—it's the dominant form of entertainment. NCsoft's Lineage has 3 million subscribers paying monthly fees. Nexon's KartRider processes millions of concurrent racing sessions. These aren't single-player games ported online; they're persistent worlds that only work when network effects achieve density.

The business model implications are profound. Single-player software is a unit economics game—you sell a box, recognize revenue, and start over with the sequel. Multi-player platforms are network effects businesses—every additional user makes the service more valuable to existing users, creating exponential rather than linear returns.

This is why we're paying close attention to Blizzard's World of Warcraft, currently in beta. If they can translate their Warcraft brand into a persistent online world with subscription revenue, they'll have built something fundamentally more valuable than their packaged goods business—a platform where user engagement creates compounding returns.

The Communication Infrastructure Layer

Beyond entertainment, always-on connectivity enables a complete reimagining of how people communicate. The telephone network was designed for synchronous voice. Email was designed for asynchronous text. But broadband supports something in between—persistent presence with multiple modalities.

Instant messaging is the sleeper category everyone's underestimating. AOL has 195 million AIM accounts. Yahoo has 50 million Messenger users. Microsoft is aggressively pushing MSN Messenger. The conventional wisdom dismisses these as features, not businesses. But that's desktop software thinking applied to a network effects platform.

When your entire social graph is visible in a buddy list, when you can see who's online and available for real-time text or voice, the communication layer becomes stickier than email, more immediate than phone calls, and more social than either. The switching costs aren't about features—they're about the network itself.

This is why AOL's most valuable asset isn't content or dial-up access—it's the social graph embedded in AIM. This is why Microsoft is willing to lose hundreds of millions on MSN. This is why a small team at MIT called Skype is about to launch a peer-to-peer voice application that could make international calling free. The platform value of persistent presence communication is just beginning to be understood.

User-Generated Content as Moat

The second-order effect of always-on connectivity is the economic viability of user-generated content platforms. When publishing to the internet requires a dial-up connection and technical expertise, content creation remains the province of institutions. When everyone's always connected and the tools become simple enough, the entire model flips.

Blogger, acquired by Google last month for an undisclosed sum, represents the early edge of this trend. The acquisition price was likely modest—Pyra Labs was running on fumes. But Google isn't buying revenue; they're buying a platform where users create the content that makes the platform valuable.

Think about the business model implications. Traditional media companies face linear scaling problems—more content requires more editors, more fact-checkers, more overhead. User-generated platforms face inverse scaling economics—more users create more content which attracts more users. The platform owner's job is to provide tools and infrastructure, not content.

We're seeing this pattern emerge in multiple categories. Friendster, launched last year by Jonathan Abrams, has grown to 2 million users purely through viral growth. Users create profiles, establish connections, and the network becomes valuable through the density of social relationships. No content creation by the company required.

PayPal, which went public last February at a $1.2 billion valuation despite minimal revenue, demonstrated the same principle in financial services. The company didn't need to negotiate with banks or build payment infrastructure—they turned users into the distribution channel. Every email payment request was a growth vector.

The Wikipedia Experiment

The most radical experiment in user-generated platforms is Wikipedia, now 18 months old with 100,000 articles. Jimmy Wales's thesis—that you can build a comprehensive encyclopedia through open collaboration—seems preposterous by traditional publishing economics. Encyclopedia Britannica employs 100 full-time editors and 4,000 expert contributors. Wikipedia has no employees and anyone can edit.

Yet the wiki model works precisely because of network effects and broadband economics. When thousands of contributors can continuously update and improve articles, when vandalism can be reverted in minutes, when specialized knowledge becomes accessible to casual browsers, you get something qualitatively different from traditional reference works.

This matters beyond encyclopedias. The wiki model demonstrates that user-generated platforms can compete with—and potentially surpass—traditional institutional content in quality, not just quantity. That's a profound shift in how we should think about building information businesses.

The Darkside: Winner-Take-All Dynamics

Network effects create value. They also create brutal competitive dynamics. When the value of a service increases exponentially with each additional user, markets don't settle into comfortable equilibrium with multiple players. They tip toward monopoly.

We saw this in the first internet wave. eBay didn't win auctions by having better technology—they won by achieving liquidity first. Once they had the most buyers, they attracted the most sellers. Once they had the most sellers, they attracted the most buyers. By 1999, the market was over. Amazon achieved similar dominance in books by obsessing over selection and convenience before anyone else achieved scale.

In the broadband era, these dynamics will be more extreme. Real-time, persistent platforms are even more subject to winner-take-all outcomes because switching costs include the entire social graph or user-generated content library.

Consider instant messaging. The technical barriers to launching a messaging client are minimal. Any competent engineering team could build a better feature set than AIM or Yahoo Messenger. But you can't compete with 195 million AIM users. If your friends are on AIM, you're on AIM. The network is the moat.

This has profound implications for how we evaluate opportunities. In the broadband era, being second with better technology is worth nothing. Being first with sufficient quality to trigger network effects is worth everything. Speed matters more than elegance. Distribution matters more than features.

The Platform Control Question

The open question is who controls the platforms. AOL has the largest installed base in messaging but operates a closed garden. Microsoft has desktop operating system dominance and is aggressively pushing integrated services. Yahoo has massive reach but less stickiness. The open source protocols—IRC, XMPP—have the advantage of interoperability but no business model.

Our working hypothesis: the broadband era will see renewed attempts at vertical integration. Microsoft's .NET strategy is fundamentally about controlling the platform layer—making Windows the interface between users and internet services. AOL Time Warner's strategy, despite its execution problems, has the right architectural insight—combining content, community, and connectivity in a single offering.

But there's a counter-trend worth monitoring: the peer-to-peer architecture. Napster demonstrated that you could build services that route around centralized control. Kazaa, Morpheus, and Gnutella are proving the model wasn't a one-time phenomenon. If Skype successfully launches peer-to-peer voice, they'll have demonstrated that even telecommunications—the most centralized infrastructure in history—can be disrupted by distributed architectures.

The architectural question—centralized platforms versus distributed protocols—will determine who captures value in the broadband economy.

Enterprise Implications: ASP 2.0

The enterprise market missed the first internet wave entirely. The Application Service Provider model imploded because enterprises weren't going to trust mission-critical applications to services delivered over unreliable dial-up connections. Salesforce.com, launched in 1999 with the tagline "The End of Software," has struggled to gain traction precisely because CIOs aren't convinced that applications-over-internet can match installed software reliability.

Broadband changes the equation. When connectivity is always-on and latency is low, the reliability objection weakens. When the application lives in a data center rather than on distributed desktops, the maintenance and upgrade burden shifts from IT departments to vendors. When pricing moves from capital expenditure to operating expenditure, the cash flow implications become attractive to CFOs.

This is the argument Salesforce.com is making, and their recent growth suggests the market is beginning to believe it. They've crossed 75,000 users and are adding enterprise accounts at an accelerating rate. Marc Benioff's bet—that CRM delivered as a service over the internet would beat Siebel's installed software model—is starting to look prescient rather than delusional.

The broader implication: every category of enterprise software becomes vulnerable to "software as a service" disruption. HR systems. Expense management. Collaboration tools. Financial planning. If the application can be delivered over a network, if the always-on broadband connection is reliable enough, the economic advantages of the service model become overwhelming.

Investment Implications

So what do we do with these observations? The public markets remain skeptical of anything internet-related. The venture capital environment is the worst in a decade—total VC investment in 2002 was $21 billion, down from $105 billion in 2000. Most firms are in capital preservation mode, not deployment mode.

This creates asymmetric opportunity. The infrastructure is finally in place. User behaviors are changing. Network effects businesses are demonstrating real traction. But capital markets are still pricing in dot-com failure rather than broadband success.

Our investment framework focuses on three criteria:

First, real network effects. Not every internet business benefits from network effects. E-commerce is largely a unit economics game—each transaction is independent. But platforms where each additional user makes the service more valuable to existing users—communication, user-generated content, marketplaces, multi-player experiences—those are the businesses that can achieve durable competitive advantage.

Second, evidence of organic viral growth. In the broadband era, paid customer acquisition is a red flag. The whole point of network effects is that users become the distribution channel. If a business needs substantial marketing spend to grow, it's probably not a true network effects platform. Friendster, Skype, Kazaa—these services are growing explosively without material marketing budgets. That's the signal.

Third, founding teams that understand platform economics. Most entrepreneurs still think in product terms—build a better mousetrap, charge for it, scale linearly. The founders who will build franchise businesses in the broadband era are the ones who understand that their job isn't to build features, it's to orchestrate network effects. They're thinking about viral loops, about user-generated content flywheels, about how to make switching costs structural rather than contractual.

Sectors to Watch

Communication platforms. Instant messaging, voice-over-IP, persistent presence—this is where the largest networks will be built. AOL and Microsoft are obvious players, but there's room for disruption from companies building on open protocols or peer-to-peer architectures. Skype's pending launch is worth watching closely.

Social platforms. Friendster has proven the model works. The question is whether they can maintain their lead or whether better execution by competitors—including the recently launched MySpace—will fragment the market before network effects create a winner. The social graph is potentially the most valuable data structure on the internet.

User-generated content. Blogging platforms, wikis, photo sharing—any service where users create the content and the platform provides tools and infrastructure. These businesses have remarkable unit economics once they achieve scale. Google's acquisition of Blogger signals that the search giants understand this.

Marketplaces. eBay proved the model, but there are dozens of vertical markets where similar dynamics could create value. The key is achieving liquidity before competitors—once a marketplace tips, it's nearly impossible to displace.

Enterprise SaaS. Salesforce.com is the wedge, but the entire enterprise software stack is vulnerable to service-based disruption. The companies that figure out multi-tenancy, that build for internet delivery from the ground up rather than retrofitting client-server applications, will have structural cost advantages that incumbent vendors can't match.

The Next Thirty-Six Months

Broadband penetration will continue accelerating—we'll likely hit 40% of US households by 2005. International markets, particularly in Asia and Northern Europe, will reach saturation even faster. This means the platforms being built today will have access to hundreds of millions of always-connected users within three years.

The companies that achieve critical mass in that window will build compounding advantages that become insurmountable. The companies that lag will find themselves locked out of winner-take-all markets, regardless of their technical superiority.

This is the investment environment we're entering. The infrastructure is built. The user behaviors are changing. The network effects dynamics are becoming clear. What remains is execution—the ability to move fast enough, to understand platform mechanics deeply enough, to achieve viral growth before competitors.

The irony is that the market is treating this as a risk-off environment. Venture capitalists are being cautious. Public market investors are avoiding anything internet-related. But the actual risk is being under-deployed when the fundamentals are finally aligning.

The dot-com era wasn't wrong about the future. It was just early. Now the future is arriving, and the capital markets haven't noticed yet. That's our opportunity.