a16z Annual Report: The 17 Most Exciting Web3 Ideas for 2026

Original Title: 17 things we’re excited about for crypto in 2026

Translated by: Jiahua, Chaincatcher

Editor's Note: This week, a16z released its annual “Big Ideas” from partners across teams (Apps, American Dynamism, Bio, Crypto, Growth, Infra, Speedrun). Below are observations about the future from a16z crypto partners and guest contributors—covering topics such as agents and AI; stablecoins, tokenization, and finance; privacy and security; prediction markets, SNARKs, and other applications, as well as how we will build.

1. Higher-quality, smarter stablecoin on/off ramps

Last year, stablecoin trading volume was estimated to have reached $46 trillion, continuously hitting all-time highs. To put this number in perspective: that's more than 20 times PayPal's transaction volume; nearly 3 times that of Visa (one of the world’s largest payment networks); and rapidly approaching the transaction volume of ACH (the U.S. electronic network used for direct deposits and other financial transactions).

Today, you can send stablecoins for less than a cent, in less than a second. However, the unresolved issue is how to connect these “digital dollars” to the financial rails people use every day—in other words, stablecoin on/off ramps.

A new generation of startups is filling this gap, connecting stablecoins with more familiar payment systems and local currencies. Some companies use cryptographic proofs to allow people to privately exchange local balances for digital dollars. Others integrate regional networks, using QR codes, real-time payment rails, and other features to enable interbank payments; while others are building truly interoperable global wallet layers and card-issuing platforms, allowing users to spend stablecoins at everyday merchants. These approaches collectively expand the participants in the digital dollar economy and could accelerate the direct use of stablecoins as mainstream payment methods.

As these on/off ramps mature, and digital dollars directly connect to local payment systems and merchant tools, new behaviors will emerge. Workers can receive cross-border wages in real time; merchants can accept global dollars without a bank account; apps can instantly settle value with any user worldwide. Stablecoins will fundamentally shift from niche financial tools to the foundational settlement layer of the internet.

—— Jeremy Zhang, a16z crypto engineering team

2. Thinking about RWA tokenization and stablecoins in a more “crypto-native” way

We’re seeing strong interest from banks, fintechs, and asset managers in bringing U.S. stocks, commodities, indices, and other traditional assets on-chain. As more traditional assets go on-chain, current tokenization is often “skeuomorphic”—rooted in the concepts of real-world assets, without leveraging crypto-native features.

However, synthetic representations like perpetual futures (Perps) allow for deeper liquidity and are often easier to implement. Perps also provide easy-to-understand leverage, so I believe they have the strongest product-market fit (PMF) among crypto-native derivatives. I also think emerging market equities are one of the most interesting “perpification” asset classes. (For some stocks, the liquidity of “zero-day-to-expiry” or 0DTE options markets is often deeper than spot markets, which will be a fascinating experiment for perpification.)

It all comes down to the question of “perpification vs. tokenization”; but in any case, we should see more crypto-native RWA tokenization in the coming year.

Along similar lines, in 2026, as stablecoins go mainstream in 2025, we’ll see more “origination, not just tokenization”; outstanding stablecoin issuance will continue to grow.

However, stablecoins without strong credit infrastructure look like “narrow banks,” holding specific liquid assets considered especially safe. While narrow banks are a valid product, I don’t think they’ll be the long-term backbone of the on-chain economy.

We’ve already seen many new asset managers, curators, and protocols begin to facilitate on-chain, asset-backed lending based on off-chain collateral. These loans are often originated off-chain and then tokenized. I think tokenization offers little benefit here, except perhaps for distribution to users already on-chain. That’s why debt assets should be natively originated on-chain, not issued off-chain and then tokenized. On-chain native origination reduces loan servicing costs, back-office structure costs, and increases accessibility. The challenging part here will be compliance and standardization, but builders are already working to solve these problems.

—— Guy Wuollet, General Partner, a16z crypto

3. Stablecoins unlock a bank ledger upgrade cycle—and new payment scenarios

The software that runs ordinary banks is unrecognizable to modern developers: in the 1960s and 1970s, banks were early adopters of large software systems. Second-generation core banking software began in the 1980s and 1990s (e.g., Temenos’s GLOBUS and Infosys’s Finacle). But all this software is aging and upgrades are too slow. As a result, the banking industry—especially the critical core ledgers (the key databases tracking deposits, collateral, and other obligations)—still often runs on mainframes, programmed in COBOL, and uses batch file interfaces instead of APIs.

The majority of global assets reside on these same decades-old core ledgers. While these systems are battle-tested, trusted by regulators, and deeply integrated into complex banking scenarios, they also stifle innovation. Adding key features like real-time payments (RTP) can take months, or more likely years, and requires dealing with layers of technical debt and regulatory complexity.

This is where stablecoins come in. Not only have stablecoins found product-market fit and gone mainstream in recent years, but this year, traditional finance (TradFi) institutions have embraced them at unprecedented levels. Stablecoins, tokenized deposits, tokenized treasuries, and on-chain bonds allow banks, fintechs, and financial institutions to build new products and serve new customers. More importantly, they can do so without forcing these organizations to rewrite their legacy systems—which, while aging, have reliably run for decades. Stablecoins thus provide institutions with new ways to innovate.

—— Sam Broner, Investment Partner

4. The Internet as the bank

With the mass arrival of agents and more commerce happening automatically in the background rather than through user clicks, the way money (i.e., value!) moves needs to change.

In a world where systems act based on “intent” rather than step-by-step instructions—because AI agents identify needs, fulfill obligations, or trigger outcomes and move funds—value must flow as quickly and freely as information does today. This is where blockchains, smart contracts, and new protocols come in.

Smart contracts can already settle a dollar payment globally in seconds. But in 2026, emerging primitives like x402 will make that settlement programmable and responsive: agents pay each other instantly and permissionlessly for data, GPU time, or API calls—no invoices, reconciliation, or batch processing needed. Developers release software updates bundled with built-in payment rules, limits, and audit trails—no fiat integration, merchant onboarding, or banks required. Prediction markets self-settle in real time as events unfold—odds update, agents trade, and global clearing happens in seconds… no custodians or exchanges needed.

Once value can move this way, “payment rails” are no longer a separate operational layer but become network behaviors: banks become part of the internet’s basic plumbing, and assets become infrastructure. If money becomes data packets that the internet can route, then the internet doesn’t just support the financial system… it becomes the financial system itself.

—— Christian Crowley and Pyrs Carvolth, a16z crypto go-to-market team

5. Wealth management for everyone

Personalized wealth management services have traditionally been limited to banks’ high-net-worth clients: providing customized advice and personalized portfolios across asset classes is expensive and operationally complex. But as more asset classes are tokenized, crypto-enabled strategies—using AI recommendations and co-pilots for personalization—can be executed and rebalanced instantly at extremely low cost.

This is not just robo-advisors; everyone can have active portfolio management, not just passive. In 2025, TradFi increased its portfolio allocation to crypto (directly or via ETPs), but that’s just the beginning; in 2026, we’ll see platforms built for “wealth accumulation”—not just “wealth preservation”—as fintechs (like Revolut and Robinhood) and centralized exchanges (like Coinbase) leverage their tech stack advantages to capture more market share.

Meanwhile, DeFi tools like Morpho Vaults automatically allocate assets to lending markets with the best risk-adjusted yields—providing core yield allocation for portfolios. Holding residual liquid balances as stablecoins instead of fiat, and as tokenized money market funds instead of traditional money market funds, further expands yield possibilities.

Finally, retail investors now have easier access to more illiquid private market assets, such as private credit, pre-IPO companies, and private equity, as tokenization helps unlock these markets while still maintaining compliance and reporting requirements. As various components of a balanced portfolio are tokenized (along the risk spectrum from bonds to equities to private assets and alternatives), they can be automatically rebalanced without cumbersome wire transfers.

—— Maggie Hsu, a16z crypto go-to-market team

6. From “Know Your Customer” (KYC) to “Know Your Agent” (KYA)

The bottleneck of the AI agent economy is shifting from intelligence to identity.

In financial services, “non-human identities” now outnumber human employees by 96 to 1—yet these identities are still unbanked “ghosts.” The missing key primitive here is KYA: Know Your Agent.

Just as humans need credit scores to get loans, agents will need cryptographically signed credentials to transact—linking agents to their principals, constraints, and responsibilities. Until this exists, merchants will continue to block agents at the firewall. The industry that spent decades building KYC infrastructure now has only months to figure out KYA.

—— Sean Neville, Circle co-founder and USDC architect; Catena Labs CEO

7. We will use AI for substantive research tasks

As a mathematical economist, it was hard for consumer AI models to understand my workflow in January this year; but by November, I could give models abstract instructions as I would to a PhD student… and sometimes they would return novel and correctly executed answers. Beyond my own experience, we’re starting to see AI used more broadly for research—especially in reasoning, where models now directly assist discovery and even autonomously solve Putnam problems (perhaps the world’s hardest undergraduate math exam).

This remains an open question: which fields will benefit most from this research assistance, and how. But I expect AI research will enable and reward a new polymath research style: one that favors the ability to speculate about relationships between ideas and quickly infer from even more speculative answers. Those answers may not be accurate, 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 to collide may still produce nonsense—but sometimes opens the door to discovery, just as people are often most creative when not working in a linear, explicit direction.

Reasoning this way will require a new AI workflow style—not just agent-to-agent, but more agent-wrapping-agent—where model layers help researchers evaluate early model approaches and gradually separate the true from the false. I’ve been using this approach to write papers, while others do patent searches, invent new art forms, or (unfortunately) discover new smart contract attacks.

However: operating such research-oriented, wrapper reasoning agent sets will require better interoperability between models, and a way to identify and appropriately compensate each model’s contribution—both of which crypto can help solve.

—— Scott Kominers, a16z crypto research team and Harvard Business School professor

8. The “invisible tax” on open networks

The rise of AI agents is imposing an invisible tax on open networks, fundamentally undermining their economic foundation. This disruption stems from a growing mismatch between the internet’s context layer and execution layer: currently, AI agents extract data from ad-supported websites (the context layer) to provide convenience to users, while systematically bypassing the revenue streams (like ads and subscriptions) that fund content.

To prevent the erosion of open networks (and preserve the diverse content that powers AI itself), we need to deploy technical and economic solutions at scale. This may include next-generation sponsored content models, micro-attribution systems, or other new funding models. Existing AI licensing agreements have also proven to be financially unsustainable “band-aids,” often compensating content providers with only a fraction of the revenue lost to AI traffic cannibalization.

The web needs a new techno-economic model, where value can flow automatically. The key shift in the coming year will be from static licensing to real-time, usage-based compensation. This means testing and scaling systems—possibly using blockchains to enable nano-payments and complex attribution standards—to automatically reward every entity that contributes information to a successful agent task.

—— Elizabeth Harkavy, a16z crypto investment team

9. Privacy will be the most important moat in crypto

Privacy is a key feature for the world’s financial shift to on-chain. It’s also a feature almost all existing blockchains lack. For most chains, privacy is an afterthought.

But now, privacy itself is compelling enough to distinguish one chain from all others. Privacy does something even more important: it creates lock-in effects for a chain; if you will, a privacy network effect. Especially in a world where competing on performance alone is no longer enough.

Thanks to cross-chain bridge protocols, as long as everything is public, moving from one chain to another is trivial. But once you make things private, that’s no longer the case: bridging tokens is easy, bridging secrets is hard. When moving in and out of private zones, there’s always a risk that those monitoring the chain, mempool, or network traffic can figure out who you are. The boundaries from private to public chains—even between two private chains—leak all kinds of metadata, such as correlations in transaction time and size, making it easier to track someone.

Compared to many undifferentiated new chains (where fees may drop to zero due to competition—block space has become commoditized), blockchains with privacy can have stronger network effects. The reality is, if a “general-purpose” chain doesn’t already have a thriving ecosystem, killer apps, or an unfair distribution advantage, there’s almost no reason for anyone to use it or build on it—let alone be loyal to it.

When users are on public blockchains, it’s easy for them to transact with users on other chains—it doesn’t matter which chain they join. When users are on private blockchains, on the other hand, their choice of chain matters more, because once they join, they’re less likely to leave and risk exposure. This creates a winner-take-all dynamic. And because privacy is essential for most real-world use cases, a small number of privacy chains may capture most of the crypto market.

—— Ali Yahya, General Partner, a16z crypto

10. The (near) future of messaging is not just quantum-resistant, it’s decentralized

As the world prepares for quantum computing, many crypto-based messaging apps (Apple, Signal, WhatsApp) are ahead of the curve and have done excellent work. The problem is, every major messaging app relies on private servers run by a single organization we must trust. These servers are easy targets for governments to shut down, backdoor, or coerce into handing over private data.

What’s the use of quantum encryption if a country can shut down your server; if a company has the keys to private servers; or even if a company owns a private server? Private servers require “trust me”—but no private servers means “you don’t have to trust me.” Communication doesn’t need a single company in the middle. Messaging needs open protocols, so we don’t have to trust anyone.

The way to achieve this is through decentralized networks: no private servers. No single app. All open-source code. Best-in-class encryption—including for quantum threats. With open networks, no single person, company, nonprofit, or country can take away our ability to communicate. Even if a country or company shuts down one app, 500 new versions will pop up the next day. Shut down one node, and blockchains and other economic incentives will immediately bring a new node online.

When people own their messages the way they own their money—that is, with private keys—everything changes. Apps may come and go, but people will always retain control over their messages and identities; end users can now own their messages, even if not the app.

This is not just about quantum resistance and encryption; it’s about ownership and decentralization. Without these two, all we’re building is unbreakable crypto that can still be shut down.

—— Shane Mac, XMTP Labs co-founder and CEO

11. Secrets as a service

Behind every model, agent, and automation lies a simple dependency: data. But today, most data pipelines—inputting or outputting data to models—are opaque, mutable, and unauditable. That’s fine for some consumer apps, but many industries and users (like finance and healthcare) require companies to keep sensitive data private. This is also the main obstacle currently preventing institutions from tokenizing real-world assets.

So, how do we enable secure, compliant, autonomous, and globally interoperable innovation while maintaining privacy? There are many approaches, but I’ll focus on data access control: who controls sensitive data? How does it move? Who (or what) can access it?

Without data access control, anyone wanting to keep data confidential must use centralized services or build custom setups—which is not only time-consuming and expensive, but also prevents traditional financial institutions from fully unlocking the features and benefits of on-chain data management. And as agent systems begin to autonomously browse, trade, and make decisions, industry users and institutions need cryptographic guarantees, not “best-effort trust.”

This is why I believe we need “secrets as a service”: a new technology that provides programmable, native data access rules; client-side encryption; and decentralized key management, enforcing who can decrypt what, under what conditions, and for how long… all enforced on-chain. Combined with verifiable data systems, “secrets” can then become part of the internet’s basic public infrastructure—not just an application-level patch (where privacy is often an afterthought)—making privacy core infrastructure.

—— Adeniyi Abiodun, Chief Product Officer and co-founder, Mysten Labs

12. From “code is law” to “spec is law”

Recent DeFi hacks have hit battle-tested protocols with strong teams, diligent audits, and years of production experience. These events highlight an uncomfortable reality: today’s standard security practices are still mostly heuristic and case-by-case.

For DeFi security to mature, it needs to shift from a bug mode to design-level properties, and from “best effort” to “principled” approaches:

• On the static/pre-deployment side (testing, audits, formal verification), this means systematically proving global invariants, not just checking manually selected local invariants. Several teams are building AI-assisted proof tools that can now help write specifications, propose invariants, and take on much of the manual proof engineering that was previously too expensive to implement.

• On the dynamic/post-deployment side (runtime monitoring, runtime enforcement, etc.), those invariants can become real-time “guardrails”: the last line of defense. These guardrails will be directly encoded as runtime assertions that every transaction must satisfy.

So now, we no longer assume every bug is caught, but instead enforce key security properties of the code itself, automatically rolling back any transaction that violates them.

This is not just theoretical. In practice, nearly every exploit to date would have triggered one of these checks during execution, potentially stopping the hack. So the once-popular “code is law” evolves into “spec is law”: even new types of attacks must satisfy the same security properties that keep the system intact, so the remaining attacks are either trivial or extremely difficult to execute.

—— Daejun Park, a16z crypto engineering team

13. Prediction markets get bigger, broader, and smarter

Prediction markets have gone mainstream, and in the coming year, as they intersect with crypto and AI, they’ll only get bigger, broader, and smarter—while also posing new important challenges for builders.

First, more contracts will be listed. This means we’ll be able to access real-time odds not just for major elections or geopolitical events, but for all sorts of granular outcomes and complex, intersecting events. As these new contracts surface more information and become part of the news ecosystem, they’ll raise important social questions: about how we balance the value of this information, and how to better design them for transparency and auditability—which crypto enables.

To handle larger contract volumes, we need new ways to reach consensus on truth to resolve contracts. Centralized platform resolutions are important, but controversial cases like the “Zelensky suit market” and “Venezuela election market” show their limitations. To address these edge cases and help prediction markets scale to more useful applications, new decentralized governance and LLM oracles can help determine the truth of disputed outcomes.

AI opens up new possibilities for prediction markets—including agents automatically betting based on real-time data, synthesizing new contracts, and mechanisms that dynamically adjust markets based on agent behavior. This will make prediction markets smarter, more responsive, and could unlock new use cases like real-time risk assessment, automated hedging, and AI-driven forecasting.

However, as scale grows, builders will face new challenges: ensuring markets are manipulation-resistant, handling the complexity of dispute resolution, and balancing information transparency with privacy. These challenges will drive innovations like advanced cryptographic proofs and decentralized arbitration systems.

—— Andy Hall, a16z crypto research advisor and Stanford University professor of political economy

14. The rise of “Staked Media”

The cracks in the traditional media model—and its so-called objectivity—have been showing for some time. The internet has given everyone a voice, and more operators, practitioners, and builders now speak directly to the public. Their perspectives reflect their stakes in the world, and counterintuitively, audiences often respect them not despite these stakes, but because of them.

The new thing here isn’t the rise of social media, but the arrival of crypto tools that allow people to make publicly verifiable commitments. As AI makes it cheap and easy to generate infinite content (real or fictional, and able to claim any viewpoint or persona), relying solely on what people (or bots) say feels insufficient.

Tokenized assets, programmable locks, prediction markets, and on-chain histories provide a firmer foundation for trust: commentators can make arguments and prove they put their money where their mouth is. Podcasters can lock tokens to prove they won’t rug-pull or “pump and dump.” Analysts can tie predictions to publicly settled markets, creating auditable records.

This is what I think of as the early form of “Staked Media”: a media form that not only embraces the idea of “skin in the game,” but provides proof. In this model, credibility comes not from pretending to be detached, nor from baseless claims; instead, it comes from having stakes you can make transparent and verifiable commitments to. Staked Media won’t replace other forms of media; it complements what we already have. It provides a new signal: not just “trust me, I’m neutral,” but “here’s the risk I’m willing to take, and how you can check if I’m telling the truth.”

—— Robert Hackett, a16z crypto editorial team

15. Crypto offers a new primitive beyond blockchains

For years, SNARKs—which allow you to cryptographically verify computation without re-executing it—were mainly just a blockchain technology. The overhead was simply too high: proving a computation could take 1,000,000 times more work than just running it. That’s worth it when you spread it across thousands of validators, but impractical anywhere else.

That’s about to change. In 2026, zkVM prover overhead will drop to about 10,000 times, with memory usage in the hundreds of megabytes—fast enough to run on a phone, cheap enough to be everywhere. Here’s one reason why 10,000x might be a magic number: 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 of CPU execution in real time.

This can unlock a vision from old research papers: Verifiable Cloud Computing. If you’re running CPU workloads in the cloud anyway—because your computation isn’t heavy enough to need GPU-ization, or you lack expertise, or for legacy reasons—you’ll be able to get cryptographic proofs of correctness at a reasonable price. Provers are already GPU-optimized; your code doesn’t need to be.

—— Justin Thaler, a16z crypto research team, Associate Professor of Computer Science, Georgetown University

16. Trading is just a waypoint, not the endpoint for crypto businesses

It seems that today every well-run crypto company (except for stablecoins and some core infrastructure) has pivoted or is pivoting to trading. But if “every crypto company becomes a trading platform,” where does that leave everyone? So many participants doing the same thing cannibalizes mindshare, leaving only a few big winners. This means companies that pivot to trading too quickly miss the chance to build more defensible, enduring businesses.

While I deeply sympathize with all the founders trying to get their financials in order, chasing instant product-market fit (PMF) comes at a cost. This issue is particularly acute in crypto, where the unique dynamics around tokens and speculation can lead founders down the path of instant gratification in their search for PMF… if you will, a “marshmallow test” (delayed gratification test).

There’s nothing wrong with trading itself—it’s an important market function—but it doesn’t have to be the final destination. Founders who focus on the “product” part of product-market fit may ultimately become bigger winners.

—— Arianna Simpson, General Partner, a16z crypto

17. Unlocking the full potential of blockchains

For the past decade, one of the biggest obstacles to building blockchain networks in the U.S. has been legal uncertainty. Securities laws have been overextended and selectively enforced, forcing founders into a regulatory framework built for “companies” rather than “networks.” For years, mitigating legal risk replaced product strategy; engineers were forced to defer to lawyers.

This dynamic led to many strange distortions: founders were told to avoid transparency; token distribution became arbitrary from a legal perspective; governance became performative; organizational structures were optimized for legal cover. Tokens were designed to avoid economic value/no business model. Worse, crypto projects that ignored the rules often outperformed sincere builders.

However, crypto market structure regulation—the likelihood of this bill passing is higher than ever—has the potential to eliminate all these distortions in the coming year. If passed, this legislation will incentivize transparency, establish clear standards, and replace “enforcement roulette” with clearer, structured paths for financing, token issuance, and decentralization. Following the GENIUS Act, stablecoin proliferation has already exploded; legislation around crypto market structure will be an even bigger shift, but this time for networks.

In other words, such regulation will allow blockchain networks to operate as networks—open, autonomous, composable, credibly neutral, and decentralized.

—— Miles Jennings, a16z crypto policy team and General Counsel