A16Z Outlook 2025: What Industry Changes Will AI, Decentralization, and Stablecoins Bring?

Original Title: A few of the things we're excited about in crypto (2025)

Original Authors: Dan Boneh, Sam Broner, Andrew Hall, Mason Hall, Maggie Hsu, Miles Jennings, Scott Duke Kominers, Eddy Lazzarin, Chris Lyons, Daren Matsuoka, Joachim Neu, Daejun Park, Brian Quintenz, karma (Daniel Reynaud), Aaron Schnider, Carra Wu

Compiled by: Daisy, Mars Finance

Editor's Note: a16z has released its annual "Major Ideas" comprehensive list based on insights from its partners in artificial intelligence, U.S. dynamics, bio/health, crypto, enterprise technology, fintech, gaming, infrastructure, and other fields, for tech builders to explore in the coming year. Below are some expectations from our a16z crypto team regarding future developments. For more on policies, regulations, and more in 2025, please refer to related articles in November 2024.

AI Needs Its Own Wallet to Achieve Autonomous Behavior

As artificial intelligence transitions from "non-player characters" (NPCs) to "protagonists," they will begin to act as agents. However, until recently, AI has not been able to act truly autonomously. Currently, they still cannot participate in market activities in a verifiable autonomous manner (i.e., without human control), such as exchanging value, revealing preferences, or coordinating resources.

As we have seen, AI agents (like @truth_terminal) can trade using cryptocurrencies, which opens up various opportunities for creative content. However, AI agents have even greater potential, not only to better meet human intentions but also to become independent network participants. When AI agents start managing their own crypto wallets, signing keys, and crypto assets, we will see many interesting new use cases emerge. These use cases include AI operating or verifying nodes in decentralized physical infrastructure networks (DePIN), such as playing a role in distributed energy systems. Other possibilities include AI agents becoming true high-value players in games, and there may even be the first blockchain owned and operated by AI in the future.

Entering the "Decentralized Autonomous Chatbot"

In addition to AI having wallets, there is also an AI chatbot running in a Trusted Execution Environment (TEE). A TEE provides an isolated environment in which applications can be executed, enabling a more secure design for distributed systems. In this case, the TEE is used to prove that the chatbot is autonomous and not controlled by human operators.

Further expanding, the next big innovation will be what we call Decentralized Autonomous Chatbots (DACs, distinct from Decentralized Autonomous Companies (DACs)). Such chatbots can attract fans by publishing engaging content (whether entertaining or informative). They will build a fan base on decentralized social media; earn income from audiences in various ways; and manage their assets in cryptocurrencies. The relevant keys will be managed in a TEE running the chatbot software, meaning that no one can access these keys except for the software itself.

As risks develop, regulatory frameworks may be needed. But the key point here is decentralization: the chatbot operates on a set of permissionless nodes coordinated by a consensus protocol, and it may even become the first truly autonomous billion-dollar entity.

As More People Use AI, We Need Unique "Proof of Identity"

In a world filled with online impersonation, scams, multiple identities, deepfakes, and other real yet deceptive AI-generated content, we need "proof of identity"—a way to help us confirm that we are interacting with real people. However, the new issue here is not the false content; rather, it is that we can now produce this content at a lower cost. AI has significantly reduced the marginal cost of producing content that contains all the cues we use to judge whether something is "real."

Therefore, there is a greater need than ever for methods that can privately associate content with personal digital identities. "Proof of identity" is a crucial cornerstone for establishing digital identities. But here, it becomes a mechanism that increases the marginal cost of attacking individuals or undermining network integrity: for humans, obtaining a unique ID is free, but for AI, it is expensive and difficult.

This is why the privacy-preserving "uniqueness" attribute becomes the next big innovation for building networks we can trust. It is not just about proving identity; it fundamentally changes the cost structure for malicious actors to attack. Thus, the "uniqueness attribute"—or "Sybil resistance"—is an uncompromising feature of any proof of identity system.

From Prediction Markets to Better Information Aggregation Mechanisms

Prediction markets took center stage in 2024 with the U.S. elections, but as an economist studying market design, I believe that the transformation in 2025 will not be about prediction markets themselves. Instead, prediction markets will pave the way for more distributed, technology-based information aggregation mechanisms—mechanisms that can be applied across various fields, from community governance and sensor networks to finance.

The past year has proven this concept, but it is important to note that prediction markets are not always the best way to aggregate information: even for global, "macro" events, they can be unreliable; for more "micro" issues, the prediction pool may be too small to yield meaningful signals. However, researchers and technologists have had decades of design frameworks to incentivize people to (truthfully) share what they know in different information environments—from data pricing and purchasing mechanisms to "Bayesian truth serum" used to guide subjective assessments, many of these methods have already been applied in crypto projects.

Blockchain has always been a natural choice for implementing these mechanisms—not only because they are decentralized but also because they facilitate open, auditable incentive structures. More importantly, blockchain can make results public, allowing everyone to interpret these results in real-time.

Businesses Will Increasingly Accept Stablecoins as Payment

Stablecoins found product-market fit over the past year—which is not surprising, as they are the cheapest way to send dollars and enable fast global payments. Stablecoins also provide entrepreneurs with a more accessible platform for building new payment products: no intermediaries, minimum balance requirements, or proprietary SDKs. However, large enterprises have yet to realize the significant cost savings and new profit margins that can be gained by switching to these payment systems.

While we have seen some businesses show interest in stablecoins (and early adoption in peer-to-peer payments), I expect a larger wave of experimentation in 2025. Small and medium-sized enterprises with strong brands, fixed customer bases, and high payment costs—such as restaurants, coffee shops, and convenience stores—will be among the first to abandon credit cards. They cannot benefit from credit card fraud protection (since transactions are face-to-face) and are most affected by transaction fees (a 30-cent fee on a cup of coffee means significant profit loss!).

We should also expect large enterprises to begin adopting stablecoins. If stablecoins truly accelerate the evolution of banking history, then businesses will attempt to bypass intermediary payment providers—directly adding 2% to their profit margins. Companies will also start seeking new solutions to address current service issues provided by credit card companies, such as fraud protection and identity verification.

Countries Exploring On-Chain Government Bonds

Putting government bonds on-chain will create a government-backed, interest-bearing digital asset—without the surveillance concerns associated with central bank digital currencies (CBDCs). These products can unlock new sources of demand for collateral use in decentralized finance (DeFi) lending and derivatives protocols, thereby adding more integrity and robustness to these ecosystems.

As governments around the world continue to explore the benefits and efficiencies brought by public, permissionless, and immutable blockchains, some countries may attempt to issue on-chain government bonds this year. For example, the UK has explored digital securities through its financial regulator FCA (Financial Conduct Authority) sandbox; its Treasury and Exchequer have also expressed interest in issuing digital vouchers.

In the U.S.—given that the SEC plans to require the clearing of government bonds through traditional, cumbersome, and costly infrastructure next year—more discussions are expected on how blockchain can enhance the transparency, efficiency, and participation of bond trading.

We Will See Broader Adoption of "DUNA," a New U.S. Blockchain Network Industry Standard

In 2024, Wyoming passed a new law recognizing decentralized autonomous organizations (DAOs) as legal entities. DUNA (Decentralized Unincorporated Nonprofit Association) is specifically designed to support decentralized governance of blockchain networks and is the only viable structure for domestic projects in the U.S. By incorporating DUNA into decentralized legal entity structures, crypto projects and other decentralized communities can provide legal legitimacy to their DAOs—facilitating greater economic activity, protecting token holders from liability, and helping manage tax and compliance needs.

DAOs—communities managing the affairs of open blockchain networks—are essential tools for ensuring that networks remain open, non-discriminatory, and do not unfairly extract value. DUNA can unlock the potential of DAOs, and multiple projects are already implementing it. As the U.S. is expected to promote and accelerate the development of its crypto ecosystem in 2025, I anticipate that DUNA will become the standard for U.S. projects. We also expect other states to adopt similar structures (Wyoming is a pioneer; they were also the first to adopt the now-ubiquitous LLC) … especially as other decentralized applications outside of crypto (such as physical infrastructure/energy grids) emerge.

Liquidity Democracy Moving into the Real World

As dissatisfaction with current governance and voting systems grows, there is now a window of opportunity to experiment with new, technology-driven governance methods—not just online, but in the real world. I have previously written about how DAOs and other decentralized communities allow us to study political systems, behaviors, and rapidly evolving governance experiments on a large scale. But what if we could apply these learnings to real-world governance through blockchain?

We can finally leverage blockchain for secure, private election voting, starting with low-risk pilot projects to alleviate cybersecurity and auditing concerns. More importantly, blockchain will enable us to conduct "liquidity democracy" experiments at the local level—a way for people to vote directly on issues or delegate their votes to others. This idea was initially proposed by Lewis Carroll (author of "Alice's Adventures in Wonderland" and a prolific researcher of voting systems); however, it was not practical at scale… until now. Recent advancements in computation and connectivity, along with blockchain technology, make new forms of representative democracy possible. Crypto projects are already applying this concept and generating a wealth of data on how these systems operate—please check our recent research findings—data that can inform local governments and communities.

Builders Will Reuse Infrastructure Rather Than Just Reinventing It

Over the past year, teams have continued to "reinvent the wheel" within the blockchain tech stack—another custom validator set, consensus protocol implementation, execution engine, programming language, RPC API. While these results have improved certain specific functionalities, they often lack broader or foundational capabilities. For example, a programming language specifically for SNARKs (Succinct Non-Interactive Arguments of Knowledge): while an ideal implementation might yield more efficient SNARKs for ideal developers, in practice, it may fall short compared to general-purpose languages in areas like compiler optimization, developer tools, online learning materials, and AI programming support (at least for now), and may even lead to poorer performance of the generated SNARKs.

Therefore, I expect more teams in 2025 to leverage others' contributions and reuse existing blockchain infrastructure components—from consensus protocols and existing staking capital to proof systems. This approach will not only help builders save significant time and effort but also allow them to focus on continuously enhancing the differentiated value of their products/services.

Today, the infrastructure is finally in place to build mainstream Web3 products and services. Like other industries, these products and services will be built by teams capable of successfully navigating complex supply chains, rather than those that dismiss "not invented here."

Crypto Companies Will Start with End-User Experience Rather Than Letting Infrastructure Determine User Experience

While blockchain technology infrastructure is interesting and diverse, many crypto companies are not just choosing their infrastructure—rather, to some extent, the infrastructure is making decisions for them, impacting the end-user experience (UX). This is because specific technical choices at the infrastructure level are directly linked to the user experience (UX) of blockchain products/services.

But I believe the industry will overcome this potential ideological barrier: that technology should dictate the end-user experience, rather than the other way around. By 2025, more crypto product designers will start with the end-user experience they want and then choose the appropriate infrastructure from there. Crypto startups will no longer need to overly focus on specific infrastructure decisions before finding product-market fit—they can concentrate on genuinely finding product-market fit.

We will no longer be bogged down by specific EIPs (Ethereum Improvement Proposals), wallet providers, intent architectures, etc., but can abstract these choices into a cohesive, full-stack, plug-and-play manner. The industry is ready for this: rich programmable block space, mature development tools, and chain abstraction are beginning to enable more people to design crypto products. Most technical end-users do not care what language a product is written in; they only care about how to use the product daily. This will also begin to happen in the crypto industry.

"Hidden Lines" Will Help Lead to Killer Apps for Web3

The technical superpowers of blockchain set it apart, but these capabilities also somewhat hinder mainstream adoption. For creators and fans, blockchain unlocks possibilities for connectivity, ownership, and monetization… but the jargon within the industry (like "NFTs," "zkRollups," etc.) and complex designs create barriers for those who would benefit most from these technologies. I have deeply felt this when discussing Web3 with executives from many media, music, and fashion industries.

Many consumer technologies have undergone a similar path to widespread adoption: starting with the technology; certain iconic companies/designers abstracting away complexity; this process helps unlock certain breakthrough applications. Think about how email initially started—SMTP protocols were hidden behind the "send" button; or credit cards, where most users today do not care about the payment gateway. Similarly, Spotify revolutionized the music industry by bringing song playlists to our fingertips rather than showcasing file formats. As Nassim Taleb observed, "Over-engineering leads to fragility. Simplicity scales."

Therefore, I believe that by 2025, our industry will adopt this concept of "hidden lines." The best decentralized applications have already begun focusing on more intuitive interfaces, designed to make usage as simple as tapping a screen or swiping a card. In 2025, we will see more companies design simple, clear communication; successful products do not explain; they solve problems.

The Crypto Industry Finally Has Its Own App Store and Discovery Channels

When crypto applications are blocked by centralized platforms like the Apple App Store or Google Play, it limits their top-end users' access. However, we are now seeing some new app stores and marketplaces providing such distribution and discovery channels without setting barriers. For example, Worldcoin's World App marketplace—not only storing identity verification information but also allowing access to "mini-apps"—has attracted thousands of users to multiple applications in just a few days. Another example is the fee-free dApp store for Solana mobile users. These two examples also demonstrate that hardware, not just software—like phones and spherical devices—could be key advantages for crypto app stores… just as Apple devices were an advantage for early application ecosystems.

Meanwhile, other stores offer thousands of decentralized applications and Web3 development tools, covering popular blockchain ecosystems (e.g., Alchemy); and the blockchain itself acts as a publisher and distributor of games (like Ronin). However, this is not always fun: if a product is already distributed on existing platforms—like messaging apps—migrating it on-chain is very difficult (exception: Telegram/TON network). Similarly, this is true for applications with significant Web2 distribution channels. But we may see more such migrations happening by 2025.

Cryptocurrency Holders Transitioning to Cryptocurrency Users

In 2024, cryptocurrency made significant strides as a political movement, with key policymakers and politicians expressing positive views. We also saw its development as a financial movement (e.g., Bitcoin and Ethereum ETPs expanded investor participation). By 2025, cryptocurrency should further evolve into a computational movement. But where will these new users come from?

I believe it is time to re-engage the current "passive" holders of cryptocurrencies and transform them into more active users, as only 5%-10% of cryptocurrency holders are actively using cryptocurrencies. We can bring the 617 million people who already own cryptocurrencies into the blockchain, especially as blockchain infrastructure continues to improve, transaction costs decrease, and user experiences enhance. This means new applications will begin to emerge, attracting both existing and new users. Meanwhile, some early applications we have already seen—covering stablecoins, decentralized finance (DeFi), NFTs, gaming, social, decentralized physical infrastructure networks (DePIN), decentralized autonomous organizations (DAOs), and prediction markets—are also becoming more accessible to mainstream users as communities push for the adoption of these applications while focusing on user experience and other improvements.

Various Industries May Begin Tokenizing "Unconventional" Assets

As the infrastructure of the crypto industry matures and is propelled by other emerging technologies, the practice of tokenizing assets will spread widely across various industries. This will enable previously considered untouchable assets—whether due to high costs or a lack of recognized value—to not only achieve liquidity but, more importantly, participate in the global economy. AI engines can also use this information as unique datasets.

Just as the shale gas revolution unlocked previously deemed unreachable oil reserves, tokenizing unconventional assets may redefine income generation in the digital age. Seemingly sci-fi scenarios are becoming more plausible: for example, individuals could tokenize their biometric data; then lease this information to companies via smart contracts. We have already seen some early examples, such as through decentralized science (DeSci) companies, which leverage blockchain technology to bring more ownership, transparency, and consent to medical data collection. These types of developments enable people to utilize previously undeveloped assets in a decentralized manner—rather than relying on governments and centralized intermediaries to provide these resources.