AiPIN: Is AI+DEPIN unlocking the door to mass adoption of Crypto?
This is the best of times, and it is also the worst of times. With the approval of Bitcoin spot ETFs in the U.S. at the beginning of the year and Trump's victory at the end of the year, cryptocurrencies represented by Bitcoin are increasingly becoming mainstream dollar assets, while we are experiencing the most tedious bull market. Aside from Bitcoin's price increase, the crypto industry has yet to see innovations and phenomenal projects like DeFi, NFTs, and GameFi from the last bull market. There is widespread disappointment in the innovation and returns across various sub-sectors, forcing the community's attention to shift towards short-term speculation on MEME and the frenzy of FOMO. However, amidst the distortions and despair of this structural bull market, a brand new era is quietly approaching.
As primary investors, the greatest excitement lies in finding new variables and structural opportunities. In this cycle, we believe one of the biggest opportunities is the combination of DePIN and AI in Crypto, tentatively called AiPIN. DePIN has been viewed as a sub-sector capable of traversing bull and bear markets due to its strong implementation capability and good cash flow, and it is also considered one of the hopes for driving mass adoption. In this cycle, the biggest beneficiaries of DePIN are likely to be AI projects, as DePIN will feed AI across three dimensions: data, commercialization, and liquidity.
1. Opportunities and Challenges Facing AI
Opportunities
After the rise of ChatGPT, AI has shown the world the opportunity to usher in a new era. From a micro perspective, the introduction of AI in traditional scenarios can not only integrate with existing products to optimize business processes but also have a profound impact on countless trillion-dollar markets, even reshaping these industries. For entrepreneurs, the addition of AI can effectively expand project valuation space and bring higher growth potential. From a broader perspective, AI is expected to create entirely new "species," opening a new era with unprecedented forms of software and hardware applications. Although the success rate of such innovative models is akin to winning the lottery, it still attracts countless brave individuals to forge ahead.
Challenges
Large models are the jewels of the AI pyramid, and numerous AI companies are rushing to join this battle of hundreds of models. This battle is protracted and extremely costly, as it requires massive computational resources to train algorithms, with high costs and continuous investment from training to service. The shortage of computing power is a dilemma faced by the entire industry, with giants like Microsoft and Facebook entering the arms race, while startups struggle to survive in the cracks. To avoid the brutal competition of bleeding cash, startups need to steer clear of models driven by financing that neglect cash flow. At the same time, the robustness and generalization ability of algorithms still have significant room for improvement, and the road ahead is long and arduous. The likely outcome may be that closed-source large models are monopolized by a few giants, while open-source and small models flourish, with opportunities for entrepreneurs possibly lying in edge computing and endpoint sectors.
Behind the massive investment is the commercial struggle of AI startup projects. Despite the influx of numerous companies, truly profitable cases are rare. Advertising in this rapidly changing industry comes too slowly, while subscription models contradict the need to attract a large user base and achieve scalability. Those technologies and products that truly understand user pain points, delve into niche scenarios, and can achieve end-to-end implementation have become the focus pursued after the restlessness subsides. After all, even the most advanced technology needs users to adopt it.
2. AI + Crypto: Ideals and Reality
In the crypto industry, the AI concept sector has also attracted funding and investor favor. However, despite the seemingly bright prospects of combining AI with Crypto, the projects that integrate AI and Crypto in the industry so far resemble conceptual marketing tools, or even more like "Meme"—creating a sensational hype effect but failing to deliver substantial innovation or practical applications.
When it comes to AI + Crypto, the hottest direction focuses on the "assetization" application scenarios in Crypto, which involves assetizing the three key elements of AI—computing power, models, and data—leveraging the decentralized characteristics of Crypto to enhance supply-demand matching efficiency. Assetizing computing power can introduce unused computing resources, facilitating the matching of supply and demand for idle computing power, especially against the backdrop of computing power scarcity. Decentralized GPU networks like io.net and rendering-focused Render (RNDR) are typical examples in this direction. In terms of AI model assetization and decentralized model operation, Bittensor (TAO) is a representative in this field. Decentralized data assetization may also help ensure privacy and reduce costs to some extent. However, the decentralized nature of Crypto has yet to genuinely alleviate the supply-demand contradiction of AI elements. Simply trading ownership or usage rights does not constitute true technological innovation; such model innovations often overlook engineering difficulties. For instance, decentralizing computing power to activate idle computing resources is not easy; large model training requires stability, and interruptions can lead to high sunk costs. Due to the complex technical details of computing power delivery, bilateral scheduling models similar to Uber and Airbnb fail here. Additionally, Nvidia's CUDA software environment and NVLINK multi-card communication make alternative costs extremely high, with NVLINK's physical distance limitations requiring GPUs to be concentrated in the same data center. In this context, the commercial model of decentralized computing power supply is difficult to realize, reducing it to mere narrative, with many computing power projects forced to abandon the training market and shift to serving the inference market. However, in the absence of large-scale application outbreaks, the demand for inference is insufficient, and large enterprises appear more stable and cost-effective by building their own solutions to meet inference needs.
On the other hand, aside from supplementing some real application scenario projects, many projects attract investment or users through the hype of AI and Crypto integration but lack actual technical landing products and market demand support. This results in some AI + Crypto combination projects having strong short-term "speculative" characteristics but lacking long-term technological value and user stickiness. In other words, many projects may seem exciting conceptually, but in practice, they often overlook numerous technical details and complexities, making it difficult to sustain development and achieve expected results.
3. DePIN: Hardware + Crypto, Challenges of Economic Models
Unlike the relatively "virtual" combination with AI, the integration of DePIN (Decentralized Physical Infrastructure Networks) and Crypto offers more practically operable application scenarios. DePIN introduces hardware devices and token incentive models, attempting to build a decentralized network reliant on cryptographic technology and IoT devices. This model drives individual participation through economic incentives, allowing them to deploy IoT devices, collect and share data, and earn rewards in the form of crypto tokens.
However, the hardware + Crypto model also faces numerous challenges. The advantage of DePIN lies in having hardware products, enabling many projects to land and commercialize, achieving revenue and cash flow, and even possessing the ability to traverse bull and bear markets. The problem is that most projects' narratives appear somewhat outdated, and product experiences often fail to compete with Web2. If they solely rely on token incentives to attract users, they may fall into a "death spiral" once the incentive model collapses. Due to its reliance on both hardware sales revenue and token models, the stability of the economic system is crucial. If the token value fluctuates too much or the costs of hardware deployment and maintenance are too high, the entire system's economic incentives will be difficult to maintain, potentially leading to user attrition and even network paralysis.
4. AiPIN: Opening the Door to Large-Scale Adoption of Crypto?
Recently, the category of AiPIN has garnered significant attention. AiPIN integrates AI technology with DePIN hardware, not only revolutionizing user experience and showcasing tremendous technological potential but also prompting a rethinking of the new interaction possibilities between humans and machines. AI represents productivity, Crypto represents production relations, and DePIN represents means of production. The convergence of these three trends in technology may open the door to large-scale applications of Crypto.
DePIN's main contribution to AI is primarily reflected in data. AI is essentially an intelligent system trained on massive amounts of data, often referred to as the "oil" of AI. Unlike the past, where the internet primarily served humans, in the new interconnected era, all devices will connect to the network, becoming data nodes. The number of machines will far exceed that of humans, forming the foundation for the Internet of Everything and laying the groundwork for a programmable society. Many AiPIN projects capture data through hardware sensors and then utilize AI to optimize data processing capabilities, achieving end-to-end process automation at the application level and unlocking the potential of niche industry scenarios. Some terminals are traditional non-networked devices, such as bicycles—before Mobike, people never thought bicycles would become part of travel data. Incremental devices like robots and autonomous vehicles also serve as new sources of data.
Investing in AI hardware is essentially an exploration and layout of execution devices. DePIN devices collect data seamlessly and establish rights, thereby aiding algorithm training. Based on data, technology, and application scenarios, there is hope to establish an ecosystem in significantly generalizable and scalable fields, resisting the impact of bubbles. Intelligent multi-modal capabilities will breathe new life into DePIN. The landing and scaling of AI technology rely on the organic collaboration of algorithms, computing power, and data, while AiPIN combines the technological applications of AI, possessing both technical advantages and commercial value. By leveraging Crypto to capture value, it completes a cold start and subsequently builds a prosperous self-operating ecosystem by distributing ownership and value to contributors.
The rich edge data collected by IoT devices in DePIN provides AI with extensive training and application scenarios. At the same time, the addition of AI makes DePIN smarter and more sustainable. AI can improve the notoriously poor user experience of Crypto products by enhancing device efficiency and optimizing network resource allocation through deep learning and prediction. It can also assist in auditing smart contracts, providing personalized services, and even dynamically adjusting economic incentive models through algorithms.
We see some smart hardware, such as rings, bracelets, and watches, utilizing sensor motion tracking to monitor health data like sleep, fitness, and heart rate, with a relatively mature supply chain. When combined with smartphones, these devices can lower interaction costs and create network effects. However, due to varying computing capabilities and user stickiness, many products ultimately become mere tools for receiving messages and call notifications.
New social smart hardware, characterized by comfort in wearing, long standby times, and no need for charging, facilitates social sharing. Whether visually or audibly, they attempt to occupy human senses, providing an excellent wearing experience while AI's multi-modal capabilities and interaction interfaces offer greater imaginative space for hardware. Users can experience and interact in more natural ways using voice, visuals, gestures, etc. These new products are somewhat like AI toys, providing not only emotional value but also continuously generating rich and imaginative content and data.
Smart glasses are another important category. Although the Vision Pro has not met expectations due to its high price and may be discontinued, lighter products (under 100g), priced lower and integrated with Crypto, will continue to explore and may become bestsellers, showcasing the powerful capabilities of mature computing platforms and display interaction terminals. In the consumer electronics field, AR is expected to explode in 2024 after experiencing a winter in 2017. As Zuckerberg said, the push from AI may make the moment for smart glasses to become computing platforms arrive sooner than expected.
There are also some AI-native hardware like Rabbit, which, despite currently underperforming in the market, provides valuable attempts to define new product forms. We look forward to new terminals and interaction methods achieving revolutionary user experiences. The disruption of human-machine collaboration has just begun, and many projects and teams are still exploring; the best ideas have yet to emerge. We anticipate AiPIN's "iPhone moment."
Creating phenomenal products has never been a smooth path. Hardware faces challenges such as fundraising and supply chain management, needing to balance aesthetics, comfort, and battery life in consumer electronics. Moreover, success requires not only the design and supply chain management capabilities of hardware products but also the ability to define AI applications, endpoint computing power, and strong sales promotion capabilities. For entrepreneurial projects, the primary task is to find product-market fit (PMF), then shift from sales-driven growth (SLG) to product-driven growth (PLG), and eventually open the second curve of growth through a data flywheel.
By integrating DePIN, AI can be nourished from both liquidity and data perspectives. Crypto can provide bottom-up funding opportunities for experimental projects to support the commercial landing and cold start of AI products, offering better solutions for the capitalization and liquidity of AI projects. Through effective data utilization by AI, DePIN may evolve from merely being a hardware (IoT) combination based on Crypto incentives to an intelligent adaptive ecosystem. By establishing a new data flywheel, multi-modal data may give rise to new super applications. Can the combination of AiPIN unlock the door to large-scale adoption of Crypto?
Copilot Venture Studio
Copilot Venture Studio is an empowering investment community composed of WEB3 investment institutions, seasoned AI corporate investors, and hardcore open-source hardware geeks. We are dedicated to discovering and supporting AiPIN startups with innovative potential at the intersection of AI and web3, addressing entrepreneurs' pain points while providing new options for investment. As an empowering investment community, we are not just investors but also builders and connectors of ecosystems. In addition to providing early-stage funding, we work alongside teams like co-founders to define business strategies, provide financing, early talent recruitment, daily operations, sales, and more, and are committed to connecting an open and innovative ecosystem that brings together top public chains, AiPIN projects, supply chain hardware resources, overseas channels, and traffic resources to jointly promote the integrated development of blockchain and AI. Build things with great people!