Introduction: David George, a partner at a16z Growth, engages in a dialogue with Chris Dixon, a partner at a16z crypto, discussing their vision for the new internet, including decentralized AI infrastructure for cryptocurrency; initiating network effects, where AI will become the native media form of this era, and more. This conversation also explores why the original business models of the internet are unraveling and how the new internet can introduce entirely new business models for creators.
How Technology Evolves
"David George": You are currently focused most of your time on the crypto space. What is your view on the relationship between crypto technology and AI?
Chris Dixon: My macro view is that waves of technology often appear in pairs or threes. Fifteen years ago, mobile internet, social networks, and cloud computing were the three major trends. Mobile internet grew the number of users with computing devices from hundreds of millions to billions; social networks were the "killer apps" that attracted users; and cloud computing was the infrastructure that supported all of this. These three are interdependent; you can't have one without the others. At that time, people debated which was better, but it turned out they were all important.
"David George": That's right, they are all necessary.
Chris Dixon: I believe that AI, crypto technology, and new types of devices (like robots, self-driving cars, and VR) are the three most interesting trends today. They also complement and develop together. Crypto technology is something new (which is the subject of my book); it provides a whole new way to architect the internet for building networks. It has unique characteristics that make things possible that were previously unachievable. When many people think of crypto, they think of Bitcoin or meme coins. But for me and many professionals who truly understand crypto, its essence goes far beyond that. There are many intersections with AI. One of the most fundamental ways to combine them is to use crypto architecture to build AI systems. We have invested a lot in this direction.
We have discussed a core question internally: Will the future of AI be controlled by a few large companies, or will it be managed by a broader community? The primary question here is: Is AI open source? It shocks me how closed the AI field has become. Ten years ago, all AI research was public and published in papers. But then, the industry suddenly became closed. They claim it's for safety reasons, but I believe it's for their own competitive advantage. Fortunately, there are still some open-source projects, like Llama, Flux, and Mistral. But I am a bit worried that this open-source model is somewhat fragile because many projects do not publicly disclose their model weights. Can these really be considered open source? Some models are open source, but their data pipelines are not. Can they really be freely reproduced? They might change the model tomorrow, and you wouldn't be able to do anything about it. These AI models are advancing every month, but if they no longer remain cutting-edge, I don't know what to do.
"David George": At least for now, AI is heavily reliant on large companies.
How Cryptocurrency and AI Interact
Chris Dixon: Some of the projects we invest in focus on establishing a decentralized internet service architecture suitable for the AI ecosystem. For example, there is a project called Jensen that is building a decentralized computing resource network. Its model is similar to Airbnb, allowing users to submit computing tasks and allocate them to idle computing resources worldwide, thus optimizing supply and demand for computing power. This network acts like an economic ledger, managing the supply and demand of computing resources.
Another example is Story Protocol, which is a new way to register intellectual property. Suppose you are a creator; you can register images, videos, or music on the blockchain, which will record the media and its ownership rights. It uses existing copyright laws to clarify copyright ownership. This way, anyone can use this content as long as they comply with the agreement; anyone can come in and say, "You can use this mix, you can create derivative works, but you must pay me 10% of the revenue."
"David George": … or any proportion.
Chris Dixon: On the blockchain, you can set terms and create an open market. But in the current market, you can only contact companies yourself and try to negotiate. This leads to people either stealing content, not using it at all, or only large companies being able to reach copyright deals. For example, OpenAI paid $100 million to Shutterstock; the blockchain creates a broad democratic resource where small creators can set their own terms.
One core advantage of crypto technology is composability. The success of open-source software is largely due to its ability to allow developers to innovate by building on existing modules. Linux is a great example; it grew from nearly 0% market share in the 90s to over 90% of the server market today because of its composability. People contribute to the system (even small contributions) to make it better. This is similar to how Wikipedia functions as a knowledge integration system.
Returning to Story Protocol, it also allows creative content to be freely combined like Lego blocks. For example, someone creates a character, another person writes a story, and someone else uses AI to generate animations; you can create a new superhero universe as long as the funding flows back, and in the end, everyone can share in the profits.
"David George": The key to this model is that the flow of funds is transparent and fair.
Chris Dixon: This way, creators can use AI tools to enhance efficiency while also receiving economic returns, rather than being exploited for free. This is a great vision—it incentivizes people to use these new tools while providing an economic model. We often think about how to find new economic models for creative workers in an AI-driven world. This is the area at the intersection of AI and Crypto that excites me the most.
"David George": In the past, social platforms captured 100% of advertising revenue, while creators could only rely on traffic monetization. What we want to see is a new system where creators can freely price and trade. This can drive more innovation.
"David George": Because the economic incentives are aligned.
Chris Dixon: Based on this, we are seeing more of this 'crowdsourced' approach to doing AI. From the perspective of data, AI needs more data. The breakthrough of crypto technology lies in its ability to design new incentive systems. The key is how we can leverage these systems to collect more AI training data? Data can serve as input for AI, be used for model evaluation, or for other purposes. This is similar to what Scale AI is doing, but the difference is that we want to accomplish this in a decentralized way, rather than having a centralized company control the entire process.
One of the projects we invest in is WorldCoin, co-founded by Sam Altman. Its core idea is that in a world where AI can forge human identities and content, we need a way to prove that a person truly exists, and the best way to do this is through blockchain, using crypto technology for identity verification. WorldCoin has designed an incentive mechanism that allows users to register and obtain identity verification, such as a spherical scanner (orb) to scan irises, but this approach has sparked some controversy. Now they offer other methods, such as verifying identity through passports. Once you complete identity verification, you can receive a cryptographic credential on the blockchain, which can be used across various services.
A simple application scenario is verification (CAPTCHA). Current CAPTCHAs have become so complex that even humans may struggle to pass them easily. Compared to these cumbersome anti-fraud systems, we can use cryptographic verification methods. Users can receive a cryptographic code to prove they are human, and then add additional layers of verification on top of that. This is another interesting intersection.
There are many opportunities for decentralized AI at the infrastructure level, such as breaking down centralized AI systems to make them decentralized at both the code and service levels. There are also entirely new possibilities, such as Machine-to-Machine Payments, and so on.
I believe the most exciting part is exploring new business models in the AI era, especially business models for creators.
Breaking the Economic Contract of the Internet
"David George": You pointed out to me right after the ChatGPT moment, "Hey, we might be breaking the contract of the internet," and I think this is a very interesting question.
Chris Dixon: There is a chapter in the book about this, close to the end. I call it the new contract. If you consider incentive systems, one of the main reasons for the success of the internet is that it has a very clever incentive system. How do you get 5 billion people to join a system without a central authority? It's because of the incentive mechanisms of the internet.
ChatGPT has shown signs that the economic contract of the internet may be breaking down. Over the past 20 years, the internet has formed a kind of implicit economic contract: search engines and social platforms gain permission to access content, and in return, creators can gain traffic. For example, travel websites, recipe sites, illustrations, etc., would allow Google to scrape content in exchange for search traffic. This model supported the development of the internet. But now, AI generates content directly, and users don't even need to click links; Google no longer needs to direct traffic to websites. As a result, the income sources for creators are cut off, and the original economic model of the internet unravels.
In the past, Google would still distribute some traffic; for example, when users searched for questions, Google would display summaries but still guide users to visit websites for more information. But later, Google began to "cut off" traffic, such as displaying answers directly in search results for StackOverflow content instead of directing users to the original website. This led to a decline in traffic for many websites, affecting their monetization capabilities. Google is also doing similar things in industries like travel and dining (e.g., Yelp), even prioritizing its own content over independent creators' content. Although these issues have existed for a long time, the AI era has exacerbated the problem.
But if AI can directly generate illustrations, recipes, and travel suggestions, users no longer need to visit those content websites. This may provide a better experience for users, but it is a devastating blow to content creators. In the future, we may only have a few AI giants left, while the original independent websites and creators will lose their space for survival.
This is the question we need to think about: Can the internet in the AI era still support innovation and entrepreneurship? If we do not solve this problem, the internet may become like the television industry of the 1970s, with only a few giants controlling all content. This is not the future of the internet we want.
So how should new websites rise? How should new things be created? We have not truly thought through this question yet.
I don't think I have a unique answer, and the solution to this problem doesn't necessarily have to rely on crypto technology. But we need to recognize that this is undermining the original incentive mechanisms of the internet. Secondly, we need to consider: is this a good thing? I don't think so. We need to find the right solution—should we create new incentive mechanisms?
This is also why I have been focused on investing in and thinking about new incentive systems, such as projects like Story Protocol. We need to explore new ways to layer new economic structures on top of the existing systems to ensure that the internet can continue to innovate and develop.
From Mobile Internet, Social Networks, and Cloud Computing to Crypto, AI, and Hardware
"David George": One thing you mentioned is the simultaneous emergence of three technological products—generative AI, cryptocurrency, and new hardware platforms. How do you view the combination of these three?
Chris Dixon: The analogy is certainly mobile, social, and cloud computing. In the last wave, they promoted each other and jointly drove the development of the internet. We are already seeing some of these combinations today.
Now, we are in another wave of technological advancement, with the core technologies being AI, crypto technology, and new types of hardware, such as robots, self-driving cars, and VR. These technologies are not independent of each other; they complement each other to form a new ecosystem. New hardware devices, such as AR and VR glasses, rely on AI to provide better interactive experiences, like the intelligent assistants in the movie "Her." Self-driving cars, Tesla's robotics technology, and various humanoid robot projects are also deploying AI technology in physical applications in the real world. Meanwhile, crypto technology provides a new way for decentralized networks to support these AI applications. So one area I am interested in is DPIN—Decentralized Physical Infrastructure. A prominent example is Helium, a community-owned and crowdsourced telecommunications network project that is competing with traditional operators like Verizon and AT&T. Helium has designed an incentive mechanism that allows anyone to set up a node at home to support the network. These nodes function similarly to wireless signal transmitters, and currently, hundreds of thousands of people across the United States have installed these nodes.
Now, Helium has also launched network services, and compared to Verizon, it is much cheaper—only $20 per month, while Verizon's cost is $70. This is mainly because Helium's network is built by the community, without the need for traditional telecom companies to invest hundreds of billions of dollars in infrastructure.
How to Use Crypto Technology to Initiate Network Effects
Chris Dixon: Crypto technology has a significant advantage in solving the "cold start" problem.
Many network effect projects face a challenge in the early stages: how to attract enough users to get the network truly operational?
For example, Helium is built and operated by the community. But suppose there are only 10 nodes; it clearly won't work. Establishing network effects is a chicken-and-egg problem. If a new social network has only 10 people, it has no appeal for new users. But if it already has 1 million users, the value of joining for new users significantly increases.
The uniqueness of crypto technology is that it can incentivize early users through token economics, thus promoting the formation of network effects. Helium is just one example; other fields, such as weather data, self-driving data, electric vehicle charging stations, decentralized maps, and even scientific research, can build networks in similar ways.
Is AI Icing or Sugar?
"David George": Marc presented me with a metaphor I really like: Is AI "icing" or "sugar"? If AI is just "icing," then existing industry giants will win because they can simply add an AI chatbot to their existing products, leveraging their established distribution channels, sales capabilities, and customer relationships to continue dominating the market. But if AI is "sugar," meaning it is a core component, then you can't just "add it in"; you need to build the entire product from scratch. In that case, the AI field is more likely to be dominated by emerging companies.
Currently, we haven't seen a clear answer. The more a product follows traditional models (for example, just using AI to enhance existing businesses), the more it favors industry giants rather than startups.
Chris Dixon: We can look at this issue from Clayton Christensen's perspective. He introduced the concepts of "disruptive innovation" and "sustaining innovation." Many people misunderstand the meaning of "disruptive innovation"; it is not merely "new technology," but rather innovation that does not fit the existing business models of companies. This is precisely why even the largest enterprises struggle to cope with true disruptive innovation, as their core customers do not need it.
This aligns with Marc's concept of "icing vs. sugar"—if AI is merely "icing" on existing products, then industry giants will naturally dominate; but if AI fundamentally changes the business model, the situation is entirely different.
For example, today's database market is primarily dominated by traditional relational databases (SQL), while AI could bring about entirely different computing architectures, even completely overturning the concept of databases. If AI is only used to optimize SQL databases, then it is merely "icing," posing no threat to existing enterprises. But if AI fundamentally changes how data is stored and retrieved, rendering traditional databases meaningless, then it is "sugar," and it will disrupt the entire industry.
"David George": We haven't seen such cases yet. I've only seen impacts on pricing (for example, cheaper AI services), but that is not enough to bring about industry disruption.
Chris Dixon: Yes, that is the second layer of the issue. I usually use a framework to analyze the landing process of these emerging technologies, but before discussing that, we can first talk about consumer-level AI. Currently, I believe that the consumer-level AI field has not yet seen products that truly possess network effects. Although AI chatbots like Claude and ChatGPT have achieved success, they have not formed strong network effects. Users can switch AI tools at any time with almost no switching costs, which makes them easily caught in price competition.
"David George": We once thought that data network effects would become the moat for AI products.
Chris Dixon: Indeed, data network effects are a theoretically existing concept, but in practice, they often aren't that strong. Many people believe that the more training data AI has, the better the model will be, and users will rely on it more, thus forming barriers. But the reality is that the incremental contribution of data generated by individual users to AI training is actually quite small. In other words, the usage data from a single user does not significantly enhance AI's capabilities, making it difficult to form strong network effects. This leads to a significant risk for AI companies: market competition will intensify, and price wars are inevitable. While AI products like ChatGPT currently have strong brand recognition, the question is how to avoid entering pure price competition?
If the switching costs between different AI tools are low, then the ultimate market competition is likely to evolve into a "price war," with all companies forced to lower prices to attract users. In that case, these AI companies will not be "dominant" companies.
"David George": So do startups still have a chance?
Chris Dixon: If AI is only used to improve existing products, then startups will struggle to compete with large companies. But if AI serves as a core architecture to create entirely new business models, then it is a different story. Currently, many AI consumer applications we see, such as face-swapping and image enhancement, although they explode in popularity in the short term, are quickly replicated by TikTok or Instagram, ultimately causing startups to lose their competitive edge. If AI products do not have network effects, then once their functionalities can be replicated, it becomes very difficult to maintain competitive power in the long term. This is why, to establish truly successful AI startups, it is essential to find entry points that can form network effects, rather than just providing a single function.
For Tools, Come for the Network
Chris Dixon: A classic user growth strategy is: "Come for the tools, stay for the network." In other words, many users initially use a product because of a specific tool, but the reason they ultimately stay is due to network effects. For example, early Photoshop users may have just wanted an image editing tool, but later they discovered the powerful ecosystem of Photoshop and became long-term users. The rise of social networks is similar; many users initially joined for a specific feature (like a friend directory), but ultimately stayed because of the social relationship chain. AI can also adopt a similar strategy; for instance, an AI image generation tool can serve as an entry point, but what should ultimately form is a complete AI creative community, not just a tool software.
Imitative Technology vs. Original Technology
Chris Dixon: Before diving deeper, it is important to discuss how major technologies are phased in. The development of new technologies typically goes through two stages:
• Imitation Stage: New technologies imitate old technologies to make it easier for users to accept.
• Original Stage: New technologies create entirely different new experiences.
There is also a third stage: the broader changes brought about by new technologies. For example, after the invention of the automobile, we built highways, suburbs, and other infrastructures.
For instance, early websites were like electronic magazines, with all content being static and not much different. This imitation stage could last ten or even twenty years, from Mosaic in 1993 to YouTube and Facebook around 2005.
But as the internet developed, we began to see original internet products, such as social media, search engines, and online video platforms, which had no offline corresponding business models.
AI is still in the mimetic stage; the AI applications we see are mainly replacing human labor, such as AI customer service and AI writing assistants. But the real AI revolution will occur in original AI products, such as AI-generated game worlds and AI-generated interactive content. This is similar to when photography first emerged, and cultural critics worried about its impact on art. Walter Benjamin's famous essay "The Work of Art in the Age of Mechanical Reproduction" asked what would happen to artists when anyone could take a photo.
Today, similar questions exist in generative AI. If AI can create an entire movie, what will happen to traditional filmmaking?
"David George": We have already seen this point in images.
AI as a Creative Cornerstone
Chris Dixon: Yes, this trend has already started with images, and video may soon follow. When photography first emerged, people worried that it would replace painting, but ultimately, photography and painting each developed their own unique artistic styles. Fine arts shifted towards abstraction, moving away from photography. On the other hand, photographic technology spurred the rise of cinema. People realized that while machines could replace photography, they could also create a new form of art that had never existed before.
The same applies to generative AI; the negative viewpoint suggests that AI will replace human creativity, but in reality, AI may give rise to entirely new forms of art, providing a new canvas for human creativity, possibly in virtual worlds, games, or new types of films. This principle can also be applied to other fields beyond the creative industry, such as consumption and social networks.
When you create something new, broader changes will follow. Social networks are a great example. They emerged in the 2000s and peaked during the Obama elections in 2008 and 2012. At that time, news articles pointed out that social media had shifted from a secondary position to a primary one. Then we began to see unexpected social changes. These changes may unfold over the next 20 to 30 years.
Balancing Supply and Demand in AI
"David George": The technological stages you mentioned are very interesting. The development of the internet took a long time, partly because it required establishing a vast network. This involves supply and demand issues—the development of the internet needed to lay down fiber optics, cables, and other wireless infrastructure. What AI needs is computational resources, such as large-scale GPU clusters. But as AI moves from the "imitation stage" to the "innovation stage," the main limiting factor may not be technological capability, but rather human creativity and ideas.
Chris Dixon: I agree. The bottleneck in AI development is likely not in technology, but in the speed of human adaptation and the influence of policies and regulations, which are closely related.
"David George": In other words, the issues in AI development include both the supply side (computational power) and the demand side (user acceptance). But the key may still be the demand side?
Chris Dixon: Yes, the challenge on the supply side is to develop sufficiently powerful AI models and have enough computational support. But the real challenge is how to get users to accept AI and integrate it into their daily lives.
We are now seeing many entrepreneurs exploring how to use AI to solve real problems. But unlike 20 years ago, the entrepreneurial ecosystem has matured significantly. A decade ago, most smart people would choose to work for large companies rather than start their own businesses. But now, the entrepreneurial ecosystem is more complete, with financing, talent, and markets being more mature than before.
However, AI still faces a major issue regarding how people's work methods will change and how industries will adapt to AI.
How AI Will Change Industries
"David George": For example, how quickly will Hollywood adopt AI?
Chris Dixon: This is exactly what I am thinking about. When I was writing my book, I wanted to use AI to generate my own audiobook, but publishers and Audible explicitly prohibited the use of AI. Part of the reason is that the industry's unions are resisting AI, but there are also deeper reasons.
"David George": So, the capability of AI to generate content exists, but the industry is not yet ready to accept it. We can see that many potential applications of AI face regulatory barriers. For example, in the healthcare industry, the technological capability of AI diagnostics is already strong enough, but regulations still limit its widespread application.
Chris Dixon: In the next five years, judges in the U.S. may rule on whether AI training data falls under fair use, or Congress may enact laws to regulate AI training data. Currently, the legality of AI training data remains contentious. AI companies argue that AI training data is a "learning" of information, not "copying." But copyright holders argue that AI has used their content without permission, constituting infringement.
"David George": This is a question that almost all AI-related industries are debating.
Chris Dixon: Yes, ultimately, it may require legal rulings to determine the reasonableness of AI training; otherwise, this issue will remain unresolved.
"David George": In regulated industries, such as healthcare and finance, when will AI truly take root?
Chris Dixon: Currently, these industries are subject to very strict regulations, and it may take a long time for AI to enter these fields. However, in certain areas, such as self-driving cars, we have already seen significant progress.
"David George": Waymo is an example. Data shows that its safety is already 7 to 10 times higher than human driving, supported by millions of miles of real-world data.
Chris Dixon: Perhaps this is the model for the widespread application of AI—first achieving breakthroughs in a specific field (like self-driving cars) and proving that it performs better than humans, and then expanding to other industries.
What is the Ideal Future of the Internet?
"David George": What do you think the ideal internet should look like?
Chris Dixon: We are at a crossroads. The original vision of the internet was a decentralized network that communities could collectively own and manage, with the economic benefits of the network flowing more to users rather than a few large companies. But now, the flow of funds on the internet has changed, with more and more profits concentrated in the hands of a few tech giants.
"David George": Yes, advertising revenue on social platforms has reached hundreds of billions, but creators only receive a tiny portion.
Chris Dixon: Currently, the top five internet companies by market capitalization may already account for over 50% of the entire industry’s market share. The internet has become a closed ecosystem dominated by a few companies.
"David George": So now tech companies have control over users and are finding ways to keep users spending more time on their platforms.
Chris Dixon: Yes, they have climbed to the peak of the internet and then kicked away the "ladder," preventing new competitors from entering. This is why we are so focused on the construction of blockchain and decentralized networks. If the future internet is entirely controlled by a few companies, then the space for innovation will be greatly compressed. Building a business on centralized platforms is like building on quicksand, which could collapse at any moment. True innovation should be built on an open ecosystem, not controlled by a few companies.
"David George": So, the focus we need to pay attention to is how to allow small tech companies to survive and grow in this ecosystem. I remain optimistic about the future. Through your efforts and the push from the entire industry, decentralized technology and open-source AI are being accepted by more and more people. Today's discussion has been great; thank you for your participation.
Chris Dixon: Thank you for the invitation.
All content on the Coinspire platform is for reference only and does not constitute an offer or advice for any investment strategy. Any personal decisions made based on the content of this article are the sole responsibility of the investor, and Coinspire is not responsible for any profits or losses arising therefrom. Investment involves risks; decisions should be made cautiously!
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
