In 2024, Crypto experienced an extremely unusual 'bull market'. Altcoins performed poorly, while Meme coins thrived, ultimately all streams converged back to BTC. Overall, despite some lows and frustrations, Crypto is indeed moving towards a more positive direction. In the upcoming year of 2025, there are many directions worth paying attention to, and in this article, we will briefly outline next year's prospects based on recent viewpoints.
1. About AI
At this current stage, projects focusing on chain abstraction often become overly focused on achieving conceptual perfection, leading to an exceptionally complicated technical implementation process that ultimately affects the user's interaction experience. Projects that include Intent architecture tend to be relatively complex in implementation, whether based on centralized (like TG Bot), structured (combining on-chain and off-chain preprocessing), or distributed (like Solver + Executor architectures). These Intent projects often share some common issues. For example, users still need to have a considerable understanding of DeFi, and the expression of intent must be clear, accurate, and simple. For complex and vague intents proposed by users, current intent projects show a sense of powerlessness, with a very limited range of implementation. Therefore, since Paradigm proposed this concept in mid-2023 to today, the so-called intent-centered projects have been more noise than substance, providing little assistance in guiding new users and reducing user operational thresholds. However, we are all very clear that from the development path of Ethereum Layer 2, the market's demand for both remains urgent.
Looking back at the development of Layer 2 over the past few months, the second-layer alliance represented by OP Superchain has been steadily growing. Zksync's Elastic Chain and Arbitrum Orbit will also form their own alliances along this path. These alliances can achieve direct interoperability in the future through solutions like interoperable clusters, alleviating the current issues of excessive fragmentation and lack of interoperability in Ethereum's Layer 2 ecosystem. The competition among dozens of chains will also narrow down to a competition among multiple forces. However, from a broader perspective, as the crypto market improves, Layer 2 projects with new architectures, such as Movement and Fuel, are also racing to launch their own mainnets to capture the scarce liquidity of altcoins. For projects below the first tier, fragmentation and lack of interoperability continue to worsen. Virtual machines based on different architectural designs may not even have wallet plugins that communicate with each other. Not to mention bringing in new users; for ordinary blockchain users, the entire Layer 2 ecosystem is incredibly complex, and the development of non-financial application chains will face tremendous resistance in such circumstances.
For Ethereum to attract new users, ecological alignment is the biggest prerequisite. An ecosystem that requires users to be half-geeks to get started will never achieve 'Mass Adoption'. Looking at the counter-trend performance of Solana and Ton this year, strategies to lower user thresholds and provide a more consistent, Web2-like user experience have clearly played an important role in ecological growth. To put it more directly, what these two ecosystems have done, aside from publicity, is simply to lower the difficulty of asset issuance and make the use of the chain feel less intrusive. Therefore, for Ethereum, a comprehensive solution prioritizing experience is essential. However, given the consistently open attitude of Ethereum's core developers, it is naturally impossible to align the entire Layer 2 ecosystem through coercion.
I believe the only solution that can first solve this problem is AI browser agents. Early in the emergence of ChatGPT, many people envisioned that AI would revolutionize app interaction by operating across multiple apps to form a comprehensive super app. Taking travel as a common example, once AI receives the user's travel needs, it can automatically complete ticket bookings, customize travel routes, arrange meals, and plan timing based on what the user specifies. If this AI also has long-term memory capabilities, it can arrange plans better suited to the user based on that memory.
Today, Google is about to launch the AI browser agent driven by Gemini, Project Mariner. In the example showcased by Jaclyn Konzelmann, director of Google Labs, after installing the AI agent extension in the Chrome browser, a chat window pops up on the right side of the browser. Users can instruct the agent to perform tasks like 'create a shopping cart from this list at the grocery store'. Then, the AI agent will automatically navigate to a grocery platform and add items to the cart for checkout. Once confirmed, the user will check out themselves (the agent does not have payment authority). Similar products will also be launched by OpenAI next month.
It is worth mentioning that although Google's Project Mariner is currently only available to selected testers, I have already experienced similar proxies developed for ordinary users in some crypto projects. From several hours of trial use, the current proxies can achieve an accuracy level of around 60-70% for implementing complex and vague intents (cursor operation speed is relatively slow) and can autonomously complete tasks such as token trading on various Dex within public chains, even cross-asset transfers from Ethereum to Layer 2. During this process, all I needed to do was inform it of my intent and input my wallet password.
Of course, this base still needs to call the API of centralized models. So what kind of collision can Crypto generate with it? I believe that AI browser agents, aside from being a better intent solution, will also drive the emergence of AI wallets, decentralized computing power, and decentralized data projects next year.
Consider a simple question: why has it taken until now for the beautiful concept of Agents to be realized during these years of rapid AI development? In fact, looking back at the development process of OpenAI, it is not difficult to find that the development of pure language models has always been faster than that of models like image generation. This is because the Internet itself is a huge corpus, providing endless textual material for training, with the limitations of language model development being more about computing power and energy. Meanwhile, agents require a large amount of human labeling and feedback, and the inference process is costly. Crypto inherently possesses the ability to obtain labor through incentives. In this economic system, upper-layer users can provide a large amount of labeled data and feedback through decentralized means to obtain tokens, while the bottom layer can integrate decentralized computing power and data projects. After training is completed, it can be integrated with wallets and DeFi projects through SDKs to achieve a true AI wallet, ultimately forming a closed loop. Ideas for other AI agents can also be derived from this, as any AI agent applicable to Web3 will need computing power, labeling, and feedback to 'grow'.
2. Stablecoins
Stablecoins will always be a battleground, and they are also a segment of Crypto with a very high barrier to entry. Regarding their application value, they have gained relatively widespread recognition even outside the industry. For example, this year, several giants in traditional finance have also ventured into the stablecoin market, including PayPal's PYUSD, BlackRock's collaboration with Ethena for USDb, and VanEck's AUSD (serving regions like Argentina and Southeast Asia).
As Tether and Circle deepen their dominance in this track, new entrants to the stablecoin issuance space are gradually dividing into two categories. First, issuers of fiat-backed stablecoins are starting to turn their attention to emerging markets, mainly in South America, and specific application scenarios. Algorithmic stablecoins are also generally shifting towards low-risk financial products as underlying assets, such as Ethena and Usual mentioned in the previous article. From a trend perspective, next year will see more Delta-neutral stablecoins competing for short-selling liquidity in centralized exchanges, while hedge assets will gradually expand from BTC and ETH to higher-risk, lower-liquidity public chain tokens to compete for the remaining sinking market. Meanwhile, stablecoins like Usual, which are backed by short- and medium-term U.S. Treasury bonds, are more focused on innovation in protocol tokens and yield methods, with no better options than short- and medium-term Treasury bonds in terms of RWA asset types. However, compared to the limited liquidity in centralized exchanges, competition for such stablecoins will be smaller, and there will be greater upside potential.
Overall, the development of stablecoins is gradually moving towards seeking more stable underlying assets and decentralization in governance. However, I hope that next year will see the emergence of completely decentralized and non-super-collateralized stablecoin protocols.
3. Payments
With the compliance and accelerated adoption of stablecoins in various countries, the downstream payment track of stablecoins will become a new competitive focus. Heterogeneous public chains like Solana and Move, which have high TPS and low Gas, will become the main infrastructure for payment applications. Traditional payments are already a very mature and highly competitive red ocean market. What changes can blockchain provide? First, there are two relatively simple and commonly mentioned points: one is to optimize cross-border payments, eliminating pre-financing requirements, making cross-border remittances faster, cheaper, and easier, addressing the issue of trillions of dollars in prepaid funds in traditional systems. The second is to serve emerging markets, which I mentioned in a previous article. In regions like Asia, Africa, and Latin America, the application value of stablecoins has already been demonstrated. Strong financial inclusion allows residents of third-world countries to effectively cope with high inflation caused by government instability. Through stablecoins, they can also participate in some global financial activities and subscribe to the world's most advanced virtual services.
The concept of 'PayFi' proposed by Solana Foundation manager Lily Liu at the 7th EthCC conference provides more imagination for the combination of blockchain and payments. This concept involves two core aspects: first, timely settlement, which means T+0 settlement. PayFi can achieve same-day settlement, even multiple settlements a day, eliminating the delays and complexities traditionally involved in the financial system, significantly increasing the speed of capital flow. The second is buy now, pay later (BNPL), which means 'Buy Now, Pay Never'. For example, a user deposits $50 into a lending product to purchase a $5 coffee. Once the accumulated interest reaches $5, that interest will be used to pay for the coffee, and the funds will be unlocked and returned to the user's account.
There are many ideas that can be extended from this, such as the financing needs of emerging projects in usage scenarios can form a safer and more transparent entry and exit through PayFi in blockchain, currency exchange during travel no longer needing to rely on various physical financial institutions, and the free control of payment and receipt time (delayed receipt to earn interest, early payment to receive discounts). The ways of earning will also become more diverse. In addition to the previously mentioned earning interest by depositing stablecoins into lending products, I personally believe that the types of stablecoins should also allow for easy conversion. In the future, with the surge of new stablecoins, users will be able to choose the most suitable type of stablecoin based on their personal risk tolerance at any time, allowing them to obtain both stablecoin protocol tokens and higher stablecoin interest. For DeFi, if this payment system can become mainstream, its growth potential will be unimaginably vast.
4. Dex
We have already mentioned the issues of fragmentation and lack of interoperability in Layer 2 in the first section. This development path also presents another problem: excess block space, where the development of infrastructure far exceeds the development of Dapps. This problem will lead to the natural elimination of a large number of long-tail chains within a few years, which is also a headache for Ethereum, where DA pricing mistakes do not receive positive feedback from Layer 2.
Looking back at this round of counter-trend growth, public chains fundamentally rely on their strong community, ecosystem, and promotional advantages, supplying these advantages to asset issuance platforms to achieve rapid overall TVL growth. Therefore, not every Layer 2 can replicate this attention economy, and the lack of super applications remains a reality to face next year. Following the trend, aside from the AI Agent-related demands we mentioned earlier, on-chain order book Dex, privacy, payment-related stacks, and decision-making tools are also quite obvious short-term trends.
I personally believe that on-chain order book Dex will become mainstream in the next generation of Dex. After all, looking at the development of AMM, the complexity of its technical path continues to increase, but the efficiency gains are becoming increasingly limited, which we also mentioned in articles related to Uni. However, for Layer 2, the limitations of performance and Gas are still very evident, and improvements in matching algorithms and innovations in Gas schemes will be key challenges.
5. Asset issuance remains the main theme
From 2023 to today, from inscriptions to the current AI Meme platforms, asset issuance methods have been a hot topic over the past year. If we extend this time span a bit, in fact, from the ICO era to the present, asset issuance can be considered the only main theme in the crypto circle. Only the external packaging and the threshold for issuance are changing. On the positive side, the demand for user gaming has driven the advanced development of infrastructure and DeFi. As this technology becomes known and recognized by the world, blockchain can step into the mainstream and integrate into reality. On the negative side, this game has become purer and more absurd. The decline in the difficulty of asset issuance also means that this dark forest is more dangerous. Nowadays, all it takes is a few clicks with some images and a few sentences to start a grand zero-sum game. Why not steer it back to a more positive side? To promote industry progress through gaming.
For example, some current AI Memes are beginning to shift towards practical Agent development, rather than the early versions of nonsensical AI Agents. The recently popular DeSci can also be called the 'ICO for scientific research'. Though currently driven by Memes, in the long run, combining various advantages of blockchain, DeSci could promote greater transparency, ease of dissemination, financing, and communication in traditional research. Whether it can ultimately be realized and how it will evolve still remains a question mark.