Original author: JieXuan Chua, CFA
Translated by: Kate, Mars Finance
Key Points
• Over the past few months, interest in artificial intelligence (“AI”) has increased, as evidenced by Google search trends and surges in prices of AI-related tokens.
• Funding for AI-related web3 projects surged to $298 million in 2023. This is more than the total funding for AI projects from 2016 to 2022 ($148.5 million).
• AI-related tokens generally performed well in 2023, with the top five AI tokens by market capitalization performing significantly better than BTC and ETH, with gains ranging from 200% to 650% in 2023.
• We have observed several trends and real-world use cases arising from the convergence of AI and cryptocurrency. From driving the growth of decentralized physical infrastructure networks (“DePINs”) to creating more interactive consumer-facing applications, we highlight some of the notable developments in this report.
introduce
2023 is proving to be a milestone for artificial intelligence (“AI”) as the transformative power of AI becomes more evident, particularly with the widespread use of AI chatbots such as OpenAI’s ChatGPT, Google’s Bard, Microsoft’s Bing Chat, etc. ChatGPT in particular highlights the potential of AI, reaching the 100 million user milestone in just two months.
This achievement surpasses major social media platforms such as TikTok and YouTube.
Figure 1: ChatGPT is one of the fastest growing apps, reaching 100 million users two months after launch
Source: demandsage, Binance Research
What’s more, AI is also beginning to reshape the crypto space, both in terms of real-world use cases and in the strong interest in AI-related tokens. The convergence of these two disruptive technologies has quickly become a prominent topic within the industry. Building on our previous report revealing use cases for AI in crypto, we now revisit this changing landscape. Given the recent resurgence of interest in the space, we’ll take a look at the current state of the market and examine new developments.
Market conditions
In 2023, public interest in artificial intelligence has risen significantly, as evidenced by a significant increase in global searches for "artificial intelligence" on Google. This heightened interest demonstrates growing public engagement in AI-related topics. This surge can be largely attributed to the popularity of AI chatbots, the launch of new AI tools, and increased media coverage and desire to learn about AI.
Figure 2: Google search interest in AI surges in 2023, significantly surpassing “crypto” and “bitcoin”
Source: Google Trends, Binance Research, as of December 31, 2023
Note: Numbers represent search interest relative to the highest point on the chart for a given region and time.
In contrast, search interest in "crypto" remained relatively stable throughout the year. There was a slight downward trend from January to May, followed by a period of stability and a slight increase towards the end of the year. Search trends for "bitcoin" mirror those for "crypto," but with more pronounced fluctuations. The fluctuations in interest in Bitcoin may be related to several hot topics surrounding Bitcoin, including Ordinals/BRC-20, potential spot ETFs, and the upcoming Bitcoin halving in 2024. These events led to a rise in the price of Bitcoin, which reignited public interest.
Overall, search trends reveal a clear divergence between growing interest in AI and relatively stable interest in Bitcoin and cryptocurrencies, suggesting that AI has been capturing the public’s attention at an increasing pace, with no clear signs of abating interest to date.
Strong investor interest
The AI sector also showed strength in terms of investor interest in 2023. Despite an overall decrease in funding amounts, AI’s share of US startup funding grew by 230% to about 26%. This growth occurred against the backdrop of funding downturns experienced by both AI and non-AI sectors. However, AI has shown particular resilience compared to the overall market.
Figure 3: AI’s share of US startup funding to double by 2023*
Source: Crunchbase, Binance Research, as of August 29, 2023
*Note: Updated data for 2023 are not yet available. Readers are advised to take this limitation into account when interpreting the analysis.
Compared to 2022, absolute funding in non-AI fields decreased by 65%, while funding in AI fields decreased by a relatively small amount, only 6%.
Furthermore, when considering the number of funding rounds, non-AI industries experienced a 55% decrease, while the AI sector experienced a 45% decrease. The relatively small decline in AI funding and funding rounds suggests that investor interest in AI applications remains relatively high despite an overall downward trend in funding amounts since the peak in 2021. This may also indicate a continued belief in the long-term potential and viability of AI technologies and applications.
In addition, the AI sector of Web3 has experienced explosive growth in funding in 2023. According to Rootdata, the total funding for AI projects from 2016 to 2022 was $148.5 million, while the funding in 2023 alone reached $298 million. This figure in 2023 is twice the total funding in the previous seven years, reflecting the surge in the appeal of AI in this year.
Figure 4: Funding for AI projects in 2023 is $298 million, ranking 7th, accounting for 3.7% of total funding for Web3 projects
Source: Rootdata, Binance Research, as of December 31, 2023
Compared with other areas in the Web3 field, funding for artificial intelligence projects in 2023 is US$298 million, ranking seventh, exceeding NFT's US$293 million and DAO's US$42 million. This funding represents approximately 3.7% of total Web3 project funding in 2023. While 3.7% may not seem like much, considering that artificial intelligence will only begin to gain significant traction in 2023, this substantial funding increase highlights the growing recognition and value of the industry.
Strong performance
From a price perspective, AI tokens have also outperformed the overall market, experiencing a significant surge over the past quarter and year. The increased interest in the sector has contributed to the strong price performance of AI-related tokens.
Figure 5: AI tokens ranked as the second best performing category over the past three months
Source: Dune Analytics (@cryptokoryo_research), as of January 2, 2023, AI tokens include: AGIX, CTXC, FET, OCEAN, ORAI, RNDR
According to the Dune dashboard that aggregates the performance of representative tokens across different narratives/sectors, AI tokens are the second best performers over the past three months. Note that while the original dashboard included MEME tokens, we have excluded them from our analysis because their relatively low market cap resulted in disproportionately large percentage performance gains.
When comparing the top five AI tokens by market cap to BTC and ETH, it is clear that AI tokens will significantly outperform the major tokens in 2023.
The one-year performance of these AI coins ranged from 200% to as high as 650%. In comparison, BTC ended the year up 150%, while ETH was up 44%.
However, it is important to note that BTC and ETH have much larger market caps compared to these AI tokens. Therefore, it is natural that BTC and ETH have smaller gains in percentage terms. This comparison is mainly meant to highlight the strong performance and growing traction of AI tokens in recent months.
Figure 6: In 2023, the top five AI tokens by market cap significantly outperformed BTC and ETH, with gains ranging from 200% to as high as 650%.
Source: CoinMarketCap, Binance Research, as of December 31, 2023
Overall, AI has gained significant traction. The adoption of AI applications has been climbing at an accelerated pace, attracting continued interest from both investors and retail investors. Additionally, AI tokens have been performing strongly. In addition to these trends, there are a few emerging AI x crypto innovations worth discussing, as detailed in the next section.
AI x Crypto Development
The surge in interest in AI has fueled the growth of AI-related crypto applications, paving the way for continued innovation in the field. In this section, we take a deep dive into some of the trends and real-world use cases arising from the convergence of AI and crypto technologies. From driving the growth of decentralized physical infrastructure networks (“DePINs”) to creating more interactive consumer-facing applications, we highlight some of the noteworthy developments in the space.
AI x DePIN
Large language models, deep learning, and various AI applications rely heavily on the computing power of graphics processing units (“GPUs”). However, the surge in interest in AI over the past year has led to an outsized demand for GPUs, resulting in a shortage of chips. Without easy access to GPUs, the high cost of computing can be prohibitive for researchers and startups working on AI-related research. This is where decentralized computing networks (a subset of DePIN) come into play. They offer an alternative to existing solutions dominated by centralized cloud providers and hardware manufacturers. As a result, we have also witnessed strong growth in the industry driven by demand for GPUs.
Given that GPUs don’t always run at 100% capacity, decentralized computing networks seek to connect those with spare computing power to those who need it. This is achieved by establishing a two-sided market that allows suppliers of computing power to receive rewards from buyers. Examples of such networks include Akash, Render, Gensyn, and io.net. Additionally, decentralized computing networks are also priced competitively because there is no significant additional cost for providers to provide computing power to the network.
Figure 7: Decentralized computing networks are competitively priced
Source: Cloudmos, as of January 2, 2024
Note: Pricing is for 1 CPU, 1GB RAM and 1GB disk
Decentralized computing networks are riding the wave of AI growth by offering potential solutions to real-world problems, with more and more activity on their platforms.
Figure 8: The number of rendering scenes on Render Network increased in 2023
Source: Dune Analytics (@lviswang), as of December 31, 2023
Figure 9: Akash network active leases surge in Q4 2023
Source: Cloudmos, as of January 3, 2024
AI x Zero Knowledge
Smart contracts are known for their efficiency due to their code-based automation capabilities. However, their predefined nature can sometimes lead to a lack of adaptability, especially in unforeseen complex situations. This is where machine learning (ML), a subfield of AI, can provide significant improvements. Machine learning models are trained on extensive data sets and have the ability to learn, adapt, and make highly accurate predictions. Integrating these models into smart contracts can open up a wide range of adaptable and flexible capabilities.
A major challenge of this integration is the prohibitively high computational overhead of on-chain ML computations. This brings us to the concept of zero-knowledge machine learning (“ZKML”). ZKML combines zero-knowledge proofs and machine learning. In this setup, ML computations are processed off-chain, and ZK proofs are used to verify the integrity of these computations without revealing the actual data. Using ZKML, smart contracts can effectively harness the power of AI while maintaining the security and transparency of blockchain technology.
Figure 10: ZKML combines zero-knowledge proof with machine learning, performing off-chain computation first and then on-chain verification
Source: Binance Research
One notable development is the ZK Predictor launched by Upshot in partnership with Modulus Labs. This tool enables Upshot to leverage Modulus ZK circuits to verify asset valuations without revealing proprietary intellectual property. It can help develop automated market makers (“AMMs”) that optimize pricing for long-tail assets, AI-driven on-chain index funds with on-chain cryptographic proof of their operations, or prediction markets focused on specific themes that can enhance and verify the accuracy of crowd-source pricing signals. Other products of ZKML include price oracles. For example, Upshot feeds its AI models with complex market data to assess the value of long-tail assets such as NFTs. Modulus’ technology then verifies the correctness of these AI calculations, encapsulates them in proofs, and submits them to Ethereum for final verification.
These examples are just the beginning of the countless applications that ZKML can support. As the technology is still in its infancy, it is expected that more mature and widespread ZKML applications will emerge in the coming years.
AI x Consumer dApps
Over the past year, we have observed an increase in AI integration in consumer-facing decentralized applications (“dApps”) to increase interactivity and boost user engagement. This trend is changing the way users interact with platforms, providing personalization and interactivity. By leveraging AI, these dApps enable users to move from being mere users to active participants.
One example is an AI user-generated content (“UGC”) platform such as NFPrompt. As the name implies, AI UGC refers to content created by users with the help of an autonomous system. This can be achieved by setting a set of rules that can be automatically output and embedding some form of randomness in the algorithm. In other words, users can input a set of rules or constraints (e.g., patterns, colors, shapes) and the AI will generate content based on this framework. By involving users in the creative process, AI UGC platforms create a more participatory relationship between users and the platform, while also allowing users to come up with unique, one-of-a-kind content that is infinitely scalable.
Figure 11: Generating an NFT using a text prompt on NFPrompt
Source: NFPrompt
In addition to content generation, the integration of AI could have a profound impact on web3 games or virtual worlds, where game characters are more interactive and dialogues are more realistic. Insomnia AI's game "Him" and "Her" are good examples. Through the use of AI, the gameplay is characterized by a focus on customization and realistic communication. This provides a more personalized experience and cultivates a more authentic emotional connection, thereby improving user stickiness.
Figure 12: He and She use AI to provide an immersive experience
Source: Sleepless AI
AI x Data Analytics
Accurate market data is key to understanding industry trends and is essential for investors to make informed investment decisions. However, instances of real trading, such as wash trading, can artificially inflate sales and distort true sales volume. By integrating AI into the analysis, the noise is filtered out and data can be output more accurately. This is widely achieved through AI and machine learning ("ML"), where large amounts of data are used as input to identify wash trading patterns or trends. The end result is a more accurate depiction of market activity.
Take BitsCrunch as an example, an AI-based NFT data analysis platform that uses AI and machine learning to detect fake transactions or suspicious transaction patterns in real time, thereby providing accurate data. The use of AI/ML enables the platform to analyze large amounts of data with relative ease, allowing the platform to distinguish between real and inorganic transaction volumes. This in turn helps make informed decisions.
Figure 13: Wash trading indicators analyzed by BitsCrunch
Conclusion
The convergence of AI and crypto has sparked tremendous excitement about the potential of these cutting-edge technologies to redefine the digital landscape. The growing popularity of AI-centric tokens, and the growing interest reflected in online search trends, underscores the continued acceleration of the AI narrative.
Granted, we are not at the point of mass adoption yet. Many AI-driven crypto projects are still in the early stages of development, and others may cater primarily to niche audiences. However, the increase in tangible use cases is an encouraging trend and positive for long-term growth. With all this in mind, investors need to understand the risks of investing in such cutting-edge technologies while taking advantage of the AI hype.
