Chris Dixon on How to Build Networks, Movements, and AI-Native Products

The most successful founders and investors don’t just chase trends—they learn to identify and ride the overwhelming, exponential forces that ultimately determine which products reshape our world. In a conversation between a16z partners Chris Dixon and Anish Acharya, the discussion centers on these foundational dynamics, from the classic power of network effects to the new rules being written in the age of AI. They trace how seminal companies like Instagram mastered the “come for the tool, stay for the network” playbook, starting as a simple photo-filter app before its social layer became indispensable. Yet, in today’s landscape, they debate whether AI’s most powerful consumer products will follow the same path or if defensibility has shifted toward brand strength, capital advantage, and “externalized” networks that exist across the broader internet ecosystem.

The dialogue explores whether the current explosion of AI tools represents a skeuomorphic phase—merely digitizing old forms—or if a truly native medium is yet to emerge. Drawing parallels to how film evolved from staged plays or how YouTube created a new grammar for video, they speculate that the most profound AI-native experiences might still be ahead, pioneered by a generation that grows up with the technology as a native language. This leads to an examination of the “idea maze,” where founders must commit to a broad domain of exponential change, like AI in healthcare, while remaining agile enough to pivot within it as the technology and market evolve.

A critical thread is the changing nature of moats in software. While network effects remain the gold standard, the conversation highlights how potent new defenses have arisen, including massive capital requirements for training AI models and the ability to command surprisingly high consumer subscription prices. This points to a potential renaissance for paid software, where founders can build sustainably by delivering extraordinary, deep value to specific niches without relying on ads. The discussion closes with a cautiously optimistic outlook on open-source AI, viewing it as a crucial democratizing force, essential for startup innovation and consumer access, even if it perpetually lags behind the cutting-edge, closed-source frontier.

Surprising Insights

• Network effects can be “externalized” to the internet ecosystem. Defensibility might no longer require a proprietary social graph inside the app; it can come from a product being enmeshed in a web of external content, influencers, tutorials, and SEO, as seen with tools like Cursor or Midjourney.
• Consumer willingness to pay for software is reaching unprecedented levels. With subscriptions for advanced AI tools hitting hundreds of dollars per month, software is moving beyond a trivial budget line-item and competing with more traditional areas of discretionary spend.
• Capital itself has become a primary moat in AI. The immense cost of training frontier models creates a significant barrier to entry, leading to a “barbell” effect where a few well-funded giants coexist with ultra-niche, solo-founded businesses.
• The most transformative tech movements often begin in small, cult-like hobbyist communities. The early adopters in subreddits, forums, and Kickstarter campaigns—not the mainstream—are frequently the source of the next big wave, as was the case with Bitcoin, VR, and 3D printing.

Practical Takeaways

• For founders: Design for “single-player mode” first. Create a tool so useful that someone would adopt it even if they were the only user, then thoughtfully layer in network or social features that enhance the core utility, following the Instagram playbook.
• Seek out and engage with niche enthusiast communities. To spot the next wave, look for highly engaged, technical hobbyist groups with their own jargon and culture; these are often the breeding grounds for future exponential trends.
• Intentionally position your company within an exponential force. Don’t just build a product; ask whether it is leveraged by composability (open source), network effects, or a meta-technological curve like AI scaling laws. Your strategy should be to ride that curve.
• When building in AI, consider going extraordinarily deep on a specific domain. Instead of building a general AI wrapper, develop deep, proprietary expertise in a vertical (e.g., AI therapy for a specific condition) to build a defensible edge against broader “god models.”