Editor: Sonal Chokshi, a16z
Translation: Tim, PANews
Original link:
Original link:
01 Stablecoins, RWA, Payments, and Finance
Better and smarter stablecoin on/off ramps
Last year, stablecoin transaction volume was estimated to reach $46 trillion, continuously setting new historical records. Specifically, this is more than 20 times the transaction volume of payment platform PayPal; nearly three times that of Visa, one of the world’s largest payment networks; and is rapidly approaching the transaction scale of the US Automated Clearing House (ACH), the electronic network that processes financial transactions such as direct deposits.
Today, stablecoin transfers can be completed in less than one second, at a cost of less than one cent. However, the unresolved issue is how to connect these cryptocurrencies with the financial infrastructure people use in daily life. In other words, it’s about building exchange channels between stablecoins and traditional currencies.
A new generation of startups is filling this gap, linking stablecoins with mainstream payment systems and local currencies. Some companies use cryptographic verification technology to allow people to exchange local account balances for digital dollars. Others connect to regional payment networks, using QR codes, real-time payment systems, and other features to enable interbank transfers. Still others are building truly interoperable global digital wallet layers and card-issuing platforms, allowing users to pay with stablecoins at everyday merchants. These innovations collectively broaden participation in the digital dollar economy and are expected to accelerate stablecoins’ direct adoption as mainstream payment methods.
As these on/off ramps mature, digital dollars are beginning to directly connect to local payment systems and merchant toolkits, giving rise to entirely new behavioral patterns. Workers can receive cross-border salaries in real time, merchants can accept globally circulating digital dollars without a bank account, and payment apps can instantly settle value with users worldwide. Stablecoins will fundamentally transform from marginal financial tools to the foundational settlement layer of the internet.
——Jeremy Zhang, a16z crypto engineering team
Understanding RWA and Stablecoins in a More Crypto-Native Way
We have observed strong interest from banks, fintech companies, and asset managers in putting traditional assets such as US stocks, commodities, and indices on-chain. As more traditional assets are put on-chain, their tokenization often remains superficial, still confined to the current concept of real-world assets and failing to fully leverage crypto-native features.
However, synthetic products like perpetual contracts can provide deeper liquidity and are often easier to implement. Perpetual contracts also offer an easy-to-understand leverage mechanism, so I believe they are the crypto-native derivatives with the strongest product-market fit. At the same time, I also think that emerging market equities are among the best asset classes for perpetual contractization. (For some stocks, the liquidity of zero-day options markets often exceeds that of the spot market, which will provide an intriguing experimental case for perpetual contractization.)
Ultimately, this is a choice between “perpetual contractization and tokenization.” In any case, we expect to see more crypto-native RWA asset tokenization in the coming year.
Along similar lines, by 2026, we will see more stablecoins being “natively issued, not just tokenized.” Stablecoins will become mainstream in 2025, with the number of issued stablecoins continuing to grow.
However, stablecoins lacking strong credit infrastructure look like narrow banks, holding specific liquid assets considered ultra-safe. While narrow banks are a legitimate financial product, I believe they will not become the backbone of the on-chain economy in the long run.
Recently, many new asset managers, curators, and protocols have begun to provide asset-backed loans on-chain, collateralized by off-chain assets. These loans are usually initiated off-chain and then tokenized. I believe tokenization offers little benefit here, except perhaps for allocation to users already on-chain. That’s why debt assets should be originated on-chain, not tokenized after off-chain origination. On-chain origination can reduce loan management costs, back-office structure costs, and improve accessibility. The challenging parts will be compliance and standardization, but builders are already working to solve these issues.
——Guy Wuollet, a16z crypto General Partner
Stablecoins Initiate a Bank Ledger Upgrade Cycle and New Payment Scenarios
The software systems banks run are often unfamiliar to modern developers: in the 1960s and 1970s, the banking industry was a pioneer in large-scale software systems. The second generation of core banking systems emerged in the 1980s and 1990s (for example, Temenos’s GLOBUS and InfoSys’s Finacle). But these systems have aged and update slowly. As a result, the banking industry—especially the critical core ledger systems that record deposits, collateral, and other debts—still often runs on mainframes, programmed in COBOL, and interacts via batch file interfaces rather than APIs.
The vast majority of global assets rely on these decades-old core ledgers. While these systems are time-tested, trusted by regulators, and deeply integrated into complex banking operations, they also hinder innovation. Adding key features like real-time payments can take months or even years, and requires overcoming layers of technical debt and regulatory complexity.
This is where stablecoins come in. The past few years have not only seen stablecoins find product-market fit and enter the mainstream, but this year, traditional financial institutions have embraced them as never before. Stablecoins, tokenized deposits, tokenized treasuries, and on-chain bonds enable banks, fintech companies, and financial institutions to develop new products and serve new customers. More importantly, they don’t require these institutions to rewrite traditional systems that, while old, have run stably for decades. Thus, stablecoins provide a new path for institutional innovation.
——Sam Broner
Internet Banking
As agents emerge at scale and more and more business activities are automated in the background rather than completed by user clicks, the flow of money to value needs to change.
In a world driven by intent rather than step-by-step instructions, AI agents can mobilize funds by recognizing needs, fulfilling obligations, or triggering outcomes—value must flow as quickly and freely as information does today. This is where blockchains, smart contracts, and on-chain protocols come into play.
Smart contracts can already complete global dollar payment settlements in seconds. By 2026, emerging primitives like x402 will make settlement programmable and responsive: agents will be able to make instant, permissionless payments for data, GPU compute, or API calls without invoices, reconciliation, or batch processing. Developers will release software updates with built-in payment rules, limits, and audit trails—no fiat integration, merchant onboarding, or financial institution involvement required. Prediction markets will self-settle in real time as events unfold, with odds dynamically updating, agents trading freely, and global payout settlements completed in seconds—all without custodians or exchanges.
Once value can flow in this way, “payment flow” is no longer a separate operational layer but becomes a network behavior: banks become the foundational pipes of the internet, assets become infrastructure. When money becomes an internet-routable packet of information, the internet is not just the backbone of the financial system—it becomes the financial system itself.
——Christian Crowley and Pyrs Carvolth, a16z crypto GTM team
Mass Wealth Management
Traditionally, personalized wealth management services have been exclusive to high-net-worth clients at banks: providing customized advice and personalized portfolio allocation across different asset classes is not only expensive but also extremely complex. But as more asset classes become tokenized and accessible via crypto channels, personalized strategies powered by AI recommendations and collaborative systems can be executed and rebalanced instantly and at low cost.
This is not just about robo-advisors—now everyone can access active portfolio management, no longer limited to passive management. In 2025, traditional financial institutions will increase their exposure to crypto (directly or via ETPs), but this is just the beginning. By 2026, we will see platforms emerge that are built for “wealth growth,” not just “wealth preservation.” Fintech companies (like Revolut and Robinhood) and centralized exchanges (like Coinbase) will leverage their tech stack advantages to capture more market share.
Meanwhile, DeFi tools like Morpho Vaults can automatically allocate assets to lending markets with the best risk-adjusted returns, providing core yield-generating asset allocation for portfolios. Holding residual liquidity balances in stablecoins instead of fiat, and investing in RWA money market funds rather than traditional money market funds, can further enhance yield potential.
Finally, retail investors can now more easily invest in less liquid private market assets such as private credit, pre-IPO companies, and private equity. Tokenization helps unlock the potential of these markets while still meeting compliance and reporting requirements. As various assets in a balanced portfolio become tokenized (with risk ranging from bonds and stocks to private and alternative investments), portfolios can be automatically rebalanced without the need for fund transfers and other procedures.
——Maggie Hsu, a16z crypto GTM team
02 AI and Agents
From “Know Your Customer” (KYC) to “Know Your Agent” (KYA)
The bottleneck for the agent economy is gradually shifting from intelligence to identity authentication.
In financial services, the number of “non-human identities” already exceeds human employees by 96 times, but these identities are still ghostly and accountless. The missing key infrastructure here is KYA: Know Your Agent.
Just as humans need credit scores to get loans, agents (AI agents) also need cryptographically signed credentials to transact, linking agents to their authorizing principals, operational limits, and liability attribution. Until this mechanism is perfected, merchants will continue to block agents at the firewall level. The KYC infrastructure that took decades to build must now solve the KYA problem in a matter of months.
——Sean Neville, Circle co-founder, USDC architect, now CEO of Catena Labs
We Will Use AI to Complete Research Work
As a mathematical economist, in January this year, I found it difficult to get general AI models to understand my workflow. By November, I was already able to give models abstract instructions as if guiding a PhD student, and sometimes they even produced novel and correct answers. Beyond my personal experience, we are witnessing AI being applied in broader research fields, especially in reasoning: current models can not only directly assist scientific discovery but also independently solve Putnam Mathematical Competition problems (possibly the world’s most difficult university-level math exam).
It remains an open question which fields these research-assistive tools will benefit most and how they will function. But I expect AI research will give rise to and reward a new type of polymathic research mode: one that favors the ability to hypothesize connections between concepts and quickly extrapolate from more speculative answers. These answers may not be precise but can still point in the right direction (at least under some topology). Ironically, this is a bit like harnessing the power of model hallucinations: when models are “smart” enough, giving them abstract space for divergent thinking may still produce meaningless content, but sometimes it sparks breakthroughs—just as humans are most creative in nonlinear, non-explicitly directed thinking.
Reasoning in this way will require a new AI workflow—not just interactions between individual agents, but agent-nesting-agent patterns, i.e., using multi-layer models to help researchers evaluate early-stage ideas and gradually distill valuable content. I have used this approach to write papers, while others use it for patent searches, creating new art forms, or (unfortunately) discovering new smart contract attacks.
However, running such agent-nesting-agent research systems requires better interoperability between models and a mechanism to identify and fairly compensate each model’s contribution. These are exactly the two key problems that cryptography is poised to help solve.
——Scott Kominers, a16z crypto research team member, Harvard Business School professor
The Invisible Tax on Open Networks
The rise of AI agents is imposing an invisible tax on open networks, fundamentally disrupting their economic foundation. This disruption stems from the growing mismatch between the internet’s context layer and execution layer: currently, AI agents extract data from ad-dependent websites (context layer), providing convenience to users while systematically bypassing the revenue channels (such as ads and subscriptions) that support content creation.
To prevent open networks from being eroded and to protect the diverse content that drives AI development, we need to deploy technical and economic solutions at scale. This may include a new generation of sponsorship schemes, attribution systems, or other novel funding models. Existing AI licensing agreements are proving to be only stopgap measures, typically compensating content providers with only a fraction of the revenue lost to AI-driven traffic cannibalization.
The network needs a new techno-economic model where value flows automatically. The key shift in the coming year will be from static licensing to real-time, usage-based compensation mechanisms. This means testing and promoting relevant systems, possibly leveraging blockchain-enabled nano-payments and precise provenance standards, to automatically reward every entity that provides information for successful agent task completion.
——Liz Harkavy, a16z crypto investment team
03 Privacy and Security
Privacy Will Become the Most Important Moat in Crypto
Privacy is a key requirement for global finance to run on-chain, but it is also a feature almost all existing blockchains lack today. For most blockchains, privacy is just an afterthought.
But now, privacy itself is enough to distinguish one blockchain from all others. Moreover, privacy plays an even more important role: it creates on-chain lock-in effects, which we might call privacy network effects. This is especially important in a world where performance alone can no longer make a chain stand out.
With bridging protocols, as long as all information is public, migration between different blockchains is easy. However, once private information is involved, things change completely: bridging tokens is easy, but bridging secrets is very difficult. When entering or leaving a private zone, there is always a risk of identity being exposed by monitored blockchains, mempools, or network traffic. Crossing the boundary between private chains and public chains, or even between two private chains, leaks various metadata, such as the correlation of transaction times and sizes, making it easier to track others.
Compared to the many homogeneous new chains (whose fees will likely be driven to zero by competition, as block space is no longer fundamentally different between chains), privacy blockchains tend to form stronger network effects. The reality is, if a “general-purpose” public chain has neither a thriving ecosystem nor killer apps or distribution advantages, there is almost no reason for users or developers to use it or build on it, let alone remain loyal.
When users use public chains, they can easily transact with users on other chains, so which chain to join doesn’t matter. However, when users use private chains, the choice of chain becomes critical, because once you join a chain, you are less likely to migrate and must bear the risk of privacy exposure—this creates a winner-takes-all scenario. Since privacy protection is crucial for most real-world applications, a handful of privacy chains may dominate the entire crypto market.
——Ali Yahya, a16z crypto General Partner
Future Messaging Needs Not Only Quantum Resistance, But Also Decentralization
As the world prepares for the quantum era, many cryptography-based messaging apps (such as Apple iMessage, Signal, WhatsApp) have led the way and made outstanding contributions. But the problem is, all mainstream messaging software relies on our trust in privately operated servers run by a single organization. These servers are easy targets for government shutdowns, backdoors, or coercion to hand over private data.
If a country can shut down an individual’s server, if a company holds the keys to private servers, or even if a company owns the private servers, what’s the point of quantum encryption? Private servers require people to “trust me,” but no private servers mean “you don’t have to trust me.” Communication doesn’t need a company as an intermediary. Messaging needs open protocols, and we shouldn’t have to trust anyone.
We achieve this through network decentralization: no private servers, no reliance on a single app, all open-source code, and top-level cryptography, including quantum-resistant technology. In an open network, no individual, company, nonprofit, or country can deprive us of our ability to communicate. Even if a country or company shuts down an app, 500 new versions will appear the next day. Even if a node is shut down, economic incentives brought by blockchain and other technologies will immediately replace it with new nodes.
When people can own their information via private keys just as they own money, everything changes. Apps may come and go, but people will always control their information and identity—end users can truly own their information, even if they don’t own the app itself.
This is not just about going beyond quantum defense and cryptography; it’s about ownership and decentralization. If either is missing, what we build is just an apparently unbreakable but still easily shut-down crypto system.
——Shane Mac, XMTP Labs co-founder and CEO
Privacy as a Service
Behind every model, agent, and automated process lies a simple element: data. Yet today, most data pipelines—whether inputting or outputting data to models—are opaque, volatile, and hard to audit. This may be acceptable for some consumer applications, but for many industries and users (such as finance and healthcare), enterprises must protect the privacy of sensitive data. At the same time, this is a major obstacle for many institutions hoping to achieve RWA tokenization.
So how do we promote secure, compliant, autonomous, and globally interoperable innovation while protecting privacy? There are many ways, but I want to focus on data access control: who controls sensitive data? How does it flow? And who (or what) can access it?
In the absence of data access control mechanisms, users who want to ensure data confidentiality currently have to rely on centralized service platforms or build custom systems. This approach is not only time-consuming and costly, but also prevents traditional financial institutions and others from fully leveraging the functional advantages of on-chain data management. As intelligent agent systems begin to autonomously browse, trade, and make decisions, users and institutions across industries need cryptographic verification mechanisms, not just “best effort trust models.”
That’s why I believe we need “privacy as a service”: this new technology can provide programmable native data access rules, client-side encryption, and decentralized key management, precisely controlling who can decrypt what data, under what conditions, and within what time frame—all executed on-chain. Combined with verifiable data systems, data privacy protection will thus be upgraded to a core component of internet infrastructure, rather than just an application-layer patch, making privacy protection truly core infrastructure.
——Adeniyi Abiodun, Mysten Labs co-founder and Chief Product Officer
From “Code Is Law” to “Rules Are Law”
Recently, some battle-tested DeFi protocols have suffered hacker attacks, despite having strong teams, rigorous audits, and years of stable operation. These incidents highlight an unsettling reality: current industry security standards still mainly rely on case-by-case and empirical judgment.
To mature, DeFi security needs to shift from vulnerability patterns to design-level, from “best effort” to “principled” approaches:
In static deployment and pre-deployment stages (testing, auditing, formal verification), this means systematically verifying global invariants, not just manually selected local invariants. Multiple teams are developing AI-assisted proof tools that help write technical specifications, propose invariant hypotheses, and greatly reduce the manual proof engineering that made such verification prohibitively expensive in the past.
In dynamic, post-deployment stages (runtime monitoring, runtime execution, etc.), these invariants can be turned into dynamic guardrails—the last line of defense. These guardrails are directly encoded as runtime assertions that every transaction must satisfy.
In this way, we no longer assume all vulnerabilities can be found, but instead enforce critical security properties in code, and any transaction violating these properties will be automatically rolled back.
This is not just theoretical. In practice, almost all exploit attacks trigger one of these security checks during execution, potentially stopping the hack. Thus, the once-popular “code is law” concept has evolved into “rules are law”: even new attack methods must meet the security property requirements that maintain system integrity, so the remaining attack vectors are either trivial or extremely difficult to execute.
——Daejun Park, a16z crypto engineering team
04 Other Tracks and Applications
Prediction Markets Become Bigger, Broader, and Smarter
Prediction markets have gradually become mainstream. Next year, as they merge with crypto and AI, they will only become bigger, broader, and smarter, but will also bring new challenges for entrepreneurs to solve.
First, there will be more contracts listed. This means we can not only get real-time odds for major elections or geopolitical events, but also for all kinds of niche outcomes and complex cross-events. As these new contracts surface, bringing more information and becoming part of the news ecosystem (which is already happening), they will raise important social questions: how should we weigh the value of this information, and how can we optimize design to make it more transparent, auditable, and full of possibilities—all of which crypto can enable.
To cope with the sharp increase in contracts, we need new consensus methods to verify contract authenticity. Centralized platform adjudication (e.g., did a specific event happen? How to confirm?) is crucial, but controversial cases like the Zelensky lawsuit and the Venezuelan election have exposed its limitations. To address these edge cases and help prediction markets expand to more practical applications, new decentralized governance mechanisms and large language model oracles will help determine the truth in disputed outcomes.
The potential of AI in prophecy is already astonishing. For example, AI agents operating on these platforms can scan global trading signals and gain short-term trading advantages, helping to discover new dimensions of the cognitive world and improve the ability to predict future events. These agents can serve as advanced political analysts for human consultation, and when we study their strategies, they reveal the predictive factors of complex social events.
Can prediction markets replace polls? No, but they can make polls better (and poll data can also be input into prediction markets). As a political scientist, I am most interested in how prediction markets can work synergistically with a rich and vibrant polling ecosystem, but we need new technologies like AI to improve the survey experience, and crypto to provide new ways to prove poll respondents are real people, not bots, etc.
——Andy Hall, a16z crypto research advisor, Stanford University professor of political economy
The Rise of Bet-Based Media
So-called objectivity has long shown cracks in the traditional media model. The internet has given everyone a voice, and more operators, practitioners, and builders are now speaking directly to the public. Their views reflect their stakes in the world, and contrary to intuition, audiences respect them—not in spite of their interests, but because of them.
The innovation here is not the rise of social media, but the arrival of crypto tools that allow people to make publicly verifiable commitments. AI makes it cheap and easy to generate infinite content, claiming anything from any perspective or identity (real or fictional), so relying solely on people’s (or bots’) words may not be enough. Tokenized assets, programmable lockups, prediction markets, and on-chain history provide a firmer foundation for trust: a commentator can publish an argument while proving they are putting real money where their mouth is. A podcast host can lock tokens to show they are not opportunistically flipping or “pump and dumping.” An analyst can tie predictions to publicly settled markets, creating an auditable track record.
I think this is the early form of “bet-based media”: this kind of media not only embraces the idea of “having skin in the game,” but can also provide proof. In this model, credibility comes not from pretending to be neutral or making empty claims, but from the actual stakes you are willing to make publicly verifiable commitments to. Bet-based media will not replace other forms of media; it is a supplement. It provides a new signal: no longer “trust me, I’m neutral,” but “here’s the risk I’m willing to take, and you can verify I mean what I say.”
——Robert Hackett, a16z crypto editorial team
Crypto Provides a New Primitive, with Applications Beyond Blockchain
For years, SNARKs (a cryptographic proof technology that verifies computation results without re-executing the computation) have been mostly limited to blockchain. The overhead was just too high: generating a proof of computation could require 1 million times more work than simply running the computation. When this overhead can be spread across tens of thousands of verification nodes, it’s worth it, but in any other scenario, it’s impractical.
That’s about to change. By 2026, zkVM provers’ overhead will drop to about 10,000 times, with memory usage of only a few hundred megabytes, making them fast enough to run smoothly on a phone and cheap enough to deploy anywhere. The reason 10,000 times may be key is that high-end GPUs have about 10,000 times the parallel throughput of a laptop CPU. By the end of 2026, a single GPU will be able to generate proofs in real time for CPU execution.
This could unlock the vision from old research papers: verifiable cloud computing. If you’re already running CPU workloads in the cloud—whether because your compute isn’t big enough to GPU-ize, you lack the expertise, or for legacy reasons—you’ll be able to get cryptographic proof of correctness at a reasonable price. The prover itself is optimized for GPUs, but your code doesn’t need to be.
——Justin Thaler, a16z crypto research team, Georgetown University associate professor of computer science
Light Trading, Heavy Building
Treating trading as a waypoint, not a destination, is the way to run a crypto business.
Today, aside from stablecoins and some core infrastructure, it seems every well-developed crypto company is pivoting or planning to pivot to trading. But if “every crypto company becomes a trading platform,” what will the industry look like? A glut of companies doing the same thing will only lead to a bloodbath, leaving just a few winners. This means companies that rush to trading miss the chance to build more defensible and lasting business models.
While I sympathize with founders working hard to keep their companies afloat, the pursuit of instant product-market fit comes at a cost. In crypto, this problem is especially acute. The industry’s unique atmosphere around tokens and speculation often lures founders seeking product-market fit down the path of instant gratification. It’s like a marshmallow experiment.
There’s nothing wrong with trading itself—it’s an important market function—but it doesn’t have to be the endgame. Founders who focus on the “product” part of product-market fit may be more likely to win in the end.
——Arianna Simpson, a16z crypto General Partner
How to Unlock Blockchain’s Full Potential After Legal and Technical Alignment
Over the past decade, one of the biggest obstacles to building blockchains in the US has been legal uncertainty. Securities laws have been misapplied and selectively enforced, forcing founders to follow regulatory frameworks designed for ordinary companies, not blockchains. For years, companies have substituted legal risk reduction for product strategy, engineers have taken a back seat, and lawyers have become the main actors.
This situation has led to many odd phenomena: founders are advised to remain opaque. Token distribution becomes arbitrary, driven solely by legal avoidance. Governance mechanisms become performative. Organizational structures are built for compliance, not effectiveness. Token design deliberately avoids economic value, even shunning business models. Worse, crypto projects that skirt the rules often outperform honest builders.
But crypto market structure regulation—government is closer than ever to passing it—could eliminate all these distortions next year. If the bill passes, it will incentivize industry transparency, establish clear standards, and provide a clearer, structured path for fundraising, token issuance, and decentralization, replacing today’s “enforcement roulette” regulatory state. After the GENIUS Act passed, stablecoins saw explosive growth; legislation around crypto market structure will bring even greater change, though this time mainly for network ecosystems.
In other words, this kind of regulation will allow blockchains to truly operate as networks—open, autonomous, composable, credibly neutral, and decentralized.
——Miles Jennings, a16z crypto policy team and General Counsel
Read the original
