Dragonfly Managing Partner 2025 Prediction: The AI Agent Boom Will Continue, and Stablecoin Usage Will Experience Explosive Growth

Author: Haseeb ＞|＜

Compiled by: Deep Tide TechFlow

These predictions will either make me look like a prophet or make me seem quite ignorant, but one thing is for sure: my views may upset quite a few "holders."

I have divided my predictions into six parts: L1 and L2, Token Issuance, Stablecoins, Regulation, "AI Agent," and the combination of Crypto and AI.

Future Trends of L1 and L2

The lines between L1 and L2 are becoming blurred. Users are no longer concerned about the differences between the two (in fact, they may have never really cared). Today's blockchain ecosystem, including L1 and L2, has become overly crowded, and a wave of consolidation is expected in the future. The key to this consolidation is not technological superiority but the ability to find a unique market position and establish user stickiness through effective go-to-market strategies (GTM).

Despite the strong performance of SVM and Move technologies, the market share of EVM will continue to grow through 2025. This is primarily driven by projects like @base, @monad_xyz, and @berachain. This growth is no longer due to compatibility but because EVM and Solidity have a wealth of training data. By 2025, large language models (LLMs) will dominate the writing of application code, and EVM has already accumulated a large library of verified crypto contracts, which will become a significant advantage.

Solana's low-latency performance will prompt more blockchains to optimize response times. The blockchain industry will shift from a competition of "transactions per second (TPS)" to a competition of "latency"—for example, infrastructure like @doublezero and ultra-low-latency L2s (like @megaeth_labs) will drive user expectations for blockchain experiences to approach Web2 response speeds. We will see more trends in applications regarding Optimistic UI, pre-confirmation, intent expression, email registration, browser-integrated wallets, and progressive security. Special thanks to @privy for their innovative advancements in this area.

@HyperliquidX has already proven that focusing on specific applications with dedicated chains is a viable model, especially when user experience and cross-chain operability are prioritized. In the future, more projects will emulate this model, and the idea of "one chain to rule them all" will become a thing of the past.

New Trends in Token Issuance

The current model of large-scale airdrops through points programs has come to an end. In the future, two main token distribution models will emerge:

1. For projects with clear core metrics (such as exchanges or lending protocols), they will distribute tokens entirely based on points. These projects do not mind whether they will be "farmed" or "gamed," as the distribution of tokens is essentially a feedback or discount mechanism for core metrics, and those who participate in airdrops are, in a sense, their real users.

2. For projects without clear core metrics (such as L1 and L2), they will shift more towards crowdfunding sales. There may be small-scale airdrops to reward social contributions, but most tokens will be distributed through crowdfunding. Airdrops aimed at vanity metrics are outdated, as these tokens do not genuinely flow to users but rather to professional airdrop hunters.

Additionally, the market share of Memecoins will gradually be replaced by "AI Agent" themed tokens. This change can be seen as a shift from "financial nihilism" to "financial over-optimism."

Explosive Growth of Stablecoins

The usage of stablecoins is expected to explode by 2025, especially among small and medium-sized businesses (SMBs). Their application scenarios will no longer be limited to trading and speculation; more businesses will use on-chain dollars for instant settlements.

Banks are also beginning to pay attention to this trend: By the end of 2025, it is expected that banks will announce the issuance of their own stablecoins to avoid being left behind in the industry. However, under Lutnick's tenure as Secretary of Commerce, Tether will still maintain its dominant position in the market.

Meanwhile, @ethena_labs is expected to attract more capital, especially in the coming year as treasury yields continue to decline. As the opportunity cost of capital decreases, the returns from basis trading will become more attractive.

Regulation

In 2025, the U.S. is expected to pass legislation related to stablecoins, while broader market infrastructure reforms (i.e., the FIT21 Act) may be delayed. The adoption of stablecoins will accelerate significantly, but Wall Street's integration of crypto, asset tokenization, and related advancements in traditional finance (TradFi) may lag behind.

Under the Trump administration, Fortune 100 companies may become more proactive in offering crypto services to consumers, while tech companies and startups will exhibit a higher risk tolerance. Trump's inauguration may bring about a brief "regulatory vacuum," during which the market will adopt a more lenient attitude toward the integration of crypto technologies due to the lack of clear rules and enforcement priorities. It is expected that during this window, crypto technology will see a massive expansion of applications within Web2 platforms.

AI Agent

(This section is longer because my views may spark controversy—please read patiently!)

The craze for "AI Agents" is expected to last throughout 2025 but will gradually fade. This is not a true long-term disruption brought by AI, but rather due to its social attributes, making it a focal point for the crypto community (CT).

Current "AI Agents" are not true agents. They are essentially chatbots that come with Memecoins, having almost no other autonomous capabilities apart from posting on Twitter. Furthermore, most existing "AI Agents" are like "the Wizard of Oz"—controlled by real people behind the scenes to ensure the AI does not make mistakes. This situation will not change in the short term, as the current stage of agent technology still has many issues (even Fortune 100 companies have not yet deployed agents in production environments). For example, these agents can easily be manipulated, potentially making inappropriate statements that could harm brand reputation, or they could be hacked to steal their resources. True autonomous AI can be referenced in the case of @freysa_ai—if an AI has not been hacked, it is likely due to human intervention behind it.

Nevertheless, I believe this trend will accelerate. Chatbots do have the potential to replace many internet celebrities, as they do not need rest, consistently deliver information, and are "more economical" than humans. Additionally, most internet celebrities are not particularly known for their originality. The collection and dissemination of real-time information can already be easily achieved through algorithms (e.g., @aixbt_agent).

Currently, these chatbots are intriguing because their concept is very unique; it feels like watching an elephant paint. When you see it for the first time, you may not care whether the painting is beautiful, as the process itself is astonishing. But after seeing it a thousand times, that novelty will gradually fade. I believe this will happen when the technology of chatbots stabilizes.

Take aixbt as an example; it is now very adept at aggregating data from different projects. By next year, with the emergence of the next generation of agents, aixbt may reduce the generation of misinformation (i.e., "hallucinations"), analyze more deeply, and provide more insightful perspectives. However, for most users, these improvements may not seem particularly significant, and they may feel little different from now.

I believe this novelty and market enthusiasm will last throughout 2025, as the crypto industry typically maintains a long-term interest in new things. However, by 2026, I anticipate a mutation: chatbots will become so ubiquitous that users will start to tire of them. Public opinion may reverse. When users see their favorite human key opinion leaders (KOLs) losing their livelihoods due to competition from chatbots, a sense of "class consciousness" may be triggered. Users will gradually lean towards supporting human KOLs, even if the quality and consistency of their content may not match that of chatbots.

To address this preference for human content, future chatbots may hide their AI identities, attempting to masquerade as humans to capture more attention market share. Unlike today's reliance on Memecoins for monetization, future chatbots may adopt profit models similar to human KOLs, such as sponsorships, affiliate links, and promoting their own tokens. At that time, incidents of KOLs being accused of being chatbots may occur frequently, and scandals involving the revelation of AI identities may arise. This trend could become very complex and bizarre.

However, there is a darker trend behind this. Currently, large language models (LLMs) excel in text processing but are not yet mature in other areas. In the crypto field, one of the easiest ways to monetize text capabilities is to become an influencer, while another is to become a scammer. In the future, as technology advances, we may see a surge of autonomous scam bots (scambots). This situation could become a serious social issue, akin to the explosion of ransomware and cryptojacking after 2017.

While chatbots may still be the focus of attention in 2025, the long-term disruptive impact of AI will not manifest at the social level.

Similarly, the long-term impact of AI will not appear in the trading domain. AI will not give everyone a "trading agent" or mini hedge fund. While AI can indeed enhance individual capabilities, this enhancement is proportional to the user's capital, data, and infrastructure. Therefore, we can expect that AI will further strengthen existing large trading firms, as they have greater capital and data advantages. In other words, large trading firms will become even better at making profits. Additionally, AI will narrow the technological gap between trading firms, as all companies can utilize "cloud-based advanced quantitative analysis tools."

Over time, AI will make markets extremely efficient—even including some niche markets. This will lead to ordinary traders having almost no advantage, even if they have their own assistant AI. The value of original research will significantly decline as a result. However, for ordinary users, increased market competition and liquidity may be good news, indicating more trading opportunities and a more active market. (For example, @Polymarket may achieve higher liquidity across all domains!)

If the future hot topic is neither chatbots nor trading bots, then what else is worth looking forward to? Here are my core views, which are hardly mentioned by anyone at the moment: truly disruptive AI Agents will emerge in the field of software engineering.

Why is this so important? Ask yourself: what is the most important input in our industry? What expensive resources limit the emergence of more applications, wallets, and higher-quality infrastructure? The answer is software. If AI Agents can significantly reduce the cost of software development, it will change the landscape of the entire industry.

In the post-AI era, seed funding may no longer require raising millions of dollars. With just $10,000 in AI cloud computing costs, you can launch an application. Self-funded projects like Hyperliquid and Jupiter will shift from rare exceptions to the mainstream. On-chain application development and innovative attempts will see explosive growth. For a software-driven industry, this cost reduction shock will trigger a wave of innovation in the blockchain field.

The impact of this change on security will also be profound. AI-driven static analysis and monitoring tools will become ubiquitous, making security more accessible. These AIs will be optimized for codebases like EVM/Solidity or Rust and trained based on extensive databases of security audits and attack cases. They will also enhance their capabilities through reinforcement learning (RL) in simulated adversarial blockchain environments. I increasingly believe that, in terms of security, AI tools will ultimately favor defenders rather than attackers. AI will continuously conduct "red team testing" on smart contracts, while other AIs will focus on reinforcing contracts, formally verifying their properties, and enhancing incident response and remediation capabilities.

Meanwhile, while you can continue to trade those Memecoins with an AI flavor, true agents will be far more than just tweeting and hyping tokens; their impact will be much deeper.

True Crypto x AI

Above, we mainly discussed the impact of AI on the crypto industry (which is the primary direction of influence), but crypto technology will also have a reverse effect on AI.

In the future, truly autonomous agents may use cryptocurrencies for mutual payments. Once regulatory policies for stablecoins become more lenient, this trend will become more apparent—even large companies running AI Agents may choose to use stablecoins for payments between agents, as this method is more convenient than traditional bank accounts.

Additionally, we will see more large-scale experiments around decentralized training and inference. Emerging projects like @exolabs, @NousResearch, and @PrimeIntellect will provide genuine alternatives to centralized training and company-exclusive models. @NEARProtocol is also working hard to create a trustworthy, neutral, and permissionless complete AI technology stack.

Another intersection of Crypto and AI is in user experience (UX). Wallets in the post-AI era will undergo a complete transformation—an AI-driven wallet will be able to automatically handle cross-chain bridging, optimize transaction paths, minimize fees, resolve interoperability issues or front-end vulnerabilities, and help users avoid obvious scams or rug pulls. Users will no longer need to switch between multiple wallets, change RPCs, or rebalance stablecoins—AI will handle all of this automatically. This transformation may take until 2026 to mature enough to fundamentally change the user experience in the crypto industry. But when all this is realized, what impact will it have on the network effects of blockchain? What will happen when users no longer care about which chain an application runs on, or even fail to notice it?

This field is still in its early stages, but I am very optimistic about its future and hope to see it truly explode soon. In the long run (say by mid-2026), I believe that most of the market value in the "AI x Crypto" space will concentrate in this direction.

That concludes all my predictions. I promised to finish this article before reaching 100,000 followers, and although it took a bit longer, I managed to complete it before the New Year!

Happy New Year, everyone! I hope that by this time next year, I will have been replaced by AI and officially "unemployed"!

Disclaimer: The content of this article represents my personal views and does not reflect the position of Dragonfly; Dragonfly has investments in many of the projects mentioned in the article. This article is not financial advice; please do your own research (DYOR). As for whether I am an AI? I'll leave that question for you to judge.