Looking ahead to 2025, AI browser agents, stablecoin innovations, expansion in the payment sector, optimization of decentralized exchanges (DEX), and asset issuance remain key trends.
Article author: Zeke.
Source: YBB Capital.
Article translated by: Mars Finance.
Introduction.
From the engraving craze to the birth of the first crypto president, 2024 is about to come to a close. This year, the cryptocurrency industry experienced an unusual 'bull market': altcoins performed poorly while meme coins took the lead, but ultimately all eyes refocused on BTC. Although there were also lows and lingering disappointments, overall, the cryptocurrency industry continues to move in a more positive direction. Looking forward to the upcoming 2025, there are many fields worth paying attention to. This article will briefly discuss my outlook for next year based on recent viewpoints.
About AI.
At the current stage, blockchain projects often overcomplicate their technical implementations in pursuit of conceptual perfection, ultimately affecting user interactions and experiences. Projects based on intent architecture tend to be more complex. Whether centralized (e.g., TG Bot), structured (combining on-chain and off-chain preprocessing), or distributed (e.g., Solver + Executor architecture), such intent-based projects typically face some common issues. For instance, users still need to have a certain level of understanding of DeFi, and the expression of 'intent' itself must be clear, accurate, and simple. When users present complex and ambiguous intents, current intent projects often struggle to complete them, and their execution range is very limited. Since the concept was proposed by Paradigm in mid-2023, most intent-focused projects have remained on paper and have not effectively promoted new user guidance or lowered barriers. However, based on the development trajectory of Ethereum Layer 2, the demand for such solutions is very urgent.
Looking back at the recent development of Layer 2, leading projects like OP Superchain continue to strengthen their alliance expansion efforts, while ZkSync's Elastic Chain and Arbitrum Orbit are also following this approach to form their own alliances. These alliances will achieve internal interoperability through clustering solutions to alleviate fragmentation and interoperability issues within the Ethereum Layer 2 ecosystem. In the future, as competition among dozens of chains begins to contract, ultimately, only multiple parties will remain in contention. However, from a more macro perspective, as the cryptocurrency market warms up, some new Layer 2 projects (such as Movement and Fuel) are also joining the competition after launching their mainnets to capture the limited liquidity in the altcoin market. For those projects that cannot enter the first tier, the phenomena of fragmentation and insufficient interoperability continue to intensify. Virtual machines based on different architectures may not even be able to have compatible wallet plugins with each other. For ordinary blockchain users, the entire Layer 2 ecosystem will become increasingly complex, which will inevitably create significant barriers for the implementation of non-financial applications.
To truly attract new users to Ethereum, ecosystem unification is crucial. An ecosystem that requires users to possess 'semi-geek' qualities is doomed to fail in achieving 'mass adoption'. In contrast, the counter-trend growth of Solana and Ton this year shows that they have lowered user thresholds and provided experiences closer to Web2, which has played a significant role in ecosystem expansion. To put it bluntly, these two ecosystems have actually just lowered the difficulty of asset issuance, allowing users to hardly feel the process of using blockchain. This indicates that Ethereum needs to integrate various parties and focus on user experience; however, given the core developers' long-standing open attitude, the likelihood of enforcing unification of the Layer 2 ecosystem is low.
I believe the solution to this problem lies in AI browser agents. Long ago, some envisioned that AI could completely transform the way applications interact, moving from point-to-point interactions to cross-application operations, forming a 'super application'. For example, when a user expresses a travel need, the AI can automatically handle flight bookings, customize travel routes, arrange dining, and preset times. If the AI also has long-term memory capabilities, it can arrange subsequent trips for the user.
Now, Google is about to launch an AI browser agent called Project Mariner, powered by Gemini. In a demonstration by Google Labs director Jaclyn Konzelmann, when users install the AI agent extension on Chrome, a chat window pops up on the right side of the browser. Users can issue instructions to the agent, such as 'put the items on this list into the grocery store's shopping cart', and the AI agent will automatically navigate to the grocery shopping platform, add the items to the cart, and proceed to the checkout page. After confirming everything is correct, the user only needs to complete the payment (the agent does not have payment authority). OpenAI will also launch a similar product next month.
It is worth noting that although Google's Project Mariner is currently only open to a small number of testers, I have already experienced similar agents aimed at ordinary users in some crypto projects. After hours of trial, I found that this agent has an accuracy rate of about 60%-70% for executing complex or ambiguous tasks, can autonomously complete token transactions on decentralized exchanges across multiple blockchains, and can transfer assets across various Layer 2 networks. Throughout the process, I only needed to express my intent to the agent and input my wallet password.
Of course, the platform still needs to use centralized model APIs. So, what role can Crypto play in this trend? I believe that AI browser agents will not only enhance the user experience of intent parsing but also promote the development of AI wallets, decentralized computing power, and decentralized data projects in the coming year.
Imagine a simple question: why has the beautiful vision of AI agents only begun to become a reality today, amidst years of rapid AI development? Looking back at the development history of OpenAI, it is not difficult to see that language models have been able to advance faster than image generation models because the internet itself, as a vast corpus, can provide a continuous stream of text training material. More limiting for the development of language models are issues of computing power and energy consumption. On the other hand, AI agents require a large amount of manual labeling and feedback, making their reasoning process costly. Crypto is naturally suited to acquire labor through token incentives: in such an economic system, upper-level users can earn tokens by providing a large amount of labeled data and feedback in a decentralized manner, while lower-level users can integrate decentralized computing power and data projects. Once training is complete, it can then be integrated with wallets and DeFi projects through SDKs to form a true AI wallet, ultimately achieving a closed loop. Based on this idea, other AI agent concepts are likely to emerge one after another, as any Web3 AI agent requires computing power, labeling, and feedback to 'grow'.
Stablecoins.
Stablecoins have always been a battleground in the cryptocurrency industry and are also a highly competitive arena. In terms of application value, they have also gained widespread recognition outside the industry. For example, this year several major players in traditional finance have begun to venture into stablecoin issuance, including PayPal's PYUSD, the collaboration between BlackRock and Ethena to launch USDb, and VanEck's AUSD targeting Argentina and Southeast Asia.
With Tether and Circle's continued strength in this field, new entrants are gradually differentiating into two main types: one is the issuers of fiat-backed stablecoins, who are beginning to turn to emerging markets in South America or specific application scenarios; the other is an increasing number of algorithmic stablecoins that use low-risk financial products as underlying assets, such as the aforementioned Ethena and Usual. From a trend perspective, next year may see more 'Delta neutral' stablecoins competing for liquidity for short-selling in centralized exchanges (CEX), while the scope of hedge assets will expand from BTC and ETH to higher-risk, less liquid public chain tokens in an attempt to capture the last remaining segment of the market. As for stablecoins like Usual that use short- to medium-term U.S. Treasury bonds as backing assets, I believe their focus will still be on innovation in protocol tokens and yield methods because, comparatively, short- to medium-term Treasury bonds are still the most secure underlying assets. For the liquidity-limited CEX market, the competitive pressure for such stablecoins is lower, theoretically allowing for a higher growth ceiling.
Overall, stablecoins are gradually evolving towards more stable underlying assets and decentralized governance. However, what I most look forward to is whether a truly fully decentralized stablecoin protocol without over-collateralization can emerge next year.
Payments.
With the relaxation of regulations on stablecoins and accelerated adoption across countries, the downstream payment sector of stablecoins will also become the focus of a new round of competition. Heterogeneous public chains with high TPS and low Gas fees, like Solana and Move, will become the main underlying infrastructure for payment applications. Traditional payments have long been a mature and fiercely competitive red ocean market; what innovations can blockchain bring to payments? The most frequently mentioned points are twofold: first, optimizing cross-border payments to avoid the need for pre-funding, making cross-border remittances faster, cheaper, and simpler, thereby solving the problem of trillions of dollars being occupied due to pre-funding in the traditional system; second, serving emerging markets. As I discussed in my previous articles, the application value of stablecoins in regions like Asia, Africa, and Latin America has been preliminarily validated. With the strong financial inclusiveness provided by stablecoins, residents of third-world countries can more effectively cope with high inflation caused by government turmoil, participate in global financial activities, and subscribe to the world's most cutting-edge virtual services.
The 'PayFi' concept proposed by Solana Foundation manager Lily Liu at the 7th EthCC conference has brought more possibilities for the combination of blockchain and payments. Its core includes two points: first, real-time settlement, or T+0 settlement. PayFi can achieve multiple settlements on the same day, freeing itself from the delays and complexities of traditional financial systems and significantly increasing the speed of fund turnover; second, the Buy Now, Pay Never (BNPL) model. For example, a user deposits $50 in a lending product to buy a $5 cup of coffee. When the accumulated interest reaches $5, it is used to pay for the coffee, and then the funds are unlocked and returned to the user's account.
Based on this, many use cases can be derived. For example, in the aspect of financing needs for emerging projects, PayFi can provide a safer and more transparent entry and exit channel; when traveling abroad, there is no need to go through various physical financial institutions for currency exchange; the timing for payments and receipts can be more flexible (delaying receipts to earn interest, paying in advance to enjoy discounts); in addition, the yield methods will also become more diverse. Besides earning interest by depositing stablecoins into lending products, I personally believe that different types of stablecoins should also allow easy interchange. In the future, as more emerging stablecoins flood into the market, users can choose the most suitable type of stablecoin based on their risk preferences and simultaneously obtain stablecoin protocol tokens and higher stablecoin interest. If this payment system can go mainstream, its potential in the DeFi sector will be very considerable.
Decentralized exchanges (DEX).
As mentioned earlier, the fragmentation and lack of interoperability in Layer 2 remain painful issues. However, this developmental path faces another problem: excess blockchain space. The speed of infrastructure construction far exceeds the growth of decentralized applications (DApps) themselves. This issue is likely to naturally eliminate many long-tail public chains in the coming years and has also become a major challenge for Ethereum in obtaining positive feedback on its data availability (DA) pricing mechanism.
Looking back at those public chains that gained development opportunities counter-cyclically, most relied on strong communities, robust ecosystems, and marketing advantages to bring opportunities to asset issuance platforms, thereby rapidly increasing overall TVL. Not every Layer 2 can replicate this 'attention economy' model. The lack of killer applications will continue into next year. Following this line of thought, in addition to the previously mentioned potential demand for AI agents, clearer short-term directions include on-chain order book DEX, privacy solutions, payment-related tech stacks, and decision-making tools.
Personally, I am quite optimistic about on-chain order book DEX becoming the mainstream of the next generation of decentralized exchanges. After all, from the development history of AMM, its technical path is becoming increasingly complex, but the marginal efficiency improvements are becoming more limited. We mentioned this in our previous article about Uniswap. However, for most Layer 2 solutions, limitations in performance and Gas costs remain evident. Therefore, innovations in matching algorithms and Gas solutions will be key challenges.
Asset issuance remains mainstream.
From 2023 to now, there have been many hot topics surrounding asset issuance, from engraving to the current AI Meme platforms. If we extend the timeline a bit, 'asset issuance' has been the core proposition of the entire cryptocurrency circle since the ICO era, although the packaging and thresholds are constantly changing.
From a positive perspective, the demand for market participation from users has driven the early development of underlying infrastructure and DeFi. When this technology gains recognition and widespread application, blockchain begins to move towards the mainstream and integrate with the real world. On the negative side, market competition is becoming increasingly pure and absurd. The lower the threshold for asset issuance, the more dangerous this 'dark forest' becomes. With just a few clicks, a few images, and a few lines of text, a large-scale zero-sum game can be initiated. However, why not direct this model towards a positive side to drive industry progress?
For example, there are already some AI Meme projects that are beginning to evolve into practical AI agents, rather than the previous type of 'paper AI' that only produced random outputs. Additionally, the recent hot topic of DeSci can also be seen as a 'scientific version of ICO'. Although its core still focuses on models at present, in the long term, injecting the advantages of blockchain into traditional scientific research can make research more transparent, easier to disseminate, easier to obtain funding, and more conducive to collaboration. Whether it can be implemented in practice or how it will evolve remains uncertain.
In fact, the underlying thinking behind DeSci is similar to my views on GameFi, which is how blockchain can effectively promote the development of independent games in the context of small and medium-sized game studios lacking funds and manpower. However, the main issues faced by financing under the blockchain model remain the low issuance threshold, insufficient restriction clauses, and excessively strong financing capabilities (which can also be attributed to the extremely low entry barrier of blockchain itself). How to constrain the use of funds through rules and force project parties to continuously create products with actual value is the focus we should pay attention to.
Let speculators compete and let builders move forward. This is the premise for the continuous development of blockchain. Next year, there may be more iterative versions of 'ICO' emerging, but what I look forward to is witnessing the rise of the next 'DeFi Summer' in this grand game.
