Tokenized RWAs Hit $25B on L2s: Did DeFi Just Become TradFi Infrastructure?

General
1

Tokenized real-world assets just crossed $25 billion on Layer 2 networks in 2025, growing 260% year-over-year. That’s not a typo—260% growth, with private credit accounting for $17B and U.S. Treasuries hitting $7.4B. And here’s the kicker: all of this is happening on Layer 2s, not Ethereum mainnet.

As someone who’s been deep in DeFi yield strategies since 2020, I’m watching this with a mix of excitement and existential dread. On one hand, institutional capital is finally flowing into blockchain infrastructure at scale. On the other hand, I can’t shake the feeling that we’re building a parallel financial system that contradicts everything DeFi was supposed to be about.

The Numbers Don’t Lie

Let me lay out what’s happening:

  • Tokenized RWAs: $30B+ in Q3 2025 (some sources say $25B, others higher—growth is explosive either way)
  • Private credit dominates: ~$17B tokenized, bringing TradFi lending on-chain
  • U.S. Treasury tokens surged 80% YoY: $7.4B in tokenized money-market and treasury funds
  • Projected growth: $2-4 trillion by 2030, potentially $30 trillion by 2034

And here’s what’s fascinating from a technical perspective: this only became viable on Layer 2s. When Ethereum L1 transaction costs are $5-50 per transaction, tokenized treasury trades don’t make economic sense. But on Arbitrum, Base, or zkSync at $0.0001 per transaction? Suddenly you can compete with TradFi settlement costs.

But Here’s My Problem

Most of these institutional RWAs are deploying on permissioned Layer 2s with KYC, verified identities, and admin keys. Projects like Horizon (a permissioned fork of Aave) and Robinhood Chain (institutional Arbitrum L2) are explicitly designed for compliance-first, traditional finance workflows.

Which raises the uncomfortable question: Did DeFi just become infrastructure for TradFi, rather than an alternative to it?

Think about it:

  • Institutions are using blockchain rails (good!)
  • But with permissioned access, KYC requirements, and centralized control (bad?)
  • They’re replicating DeFi primitives (AMMs, lending protocols) inside walled gardens
  • This creates parallel “institutional DeFi” that competes with public DeFi for liquidity

The Yield Strategist’s Dilemma

From a pure yield optimization perspective, tokenized RWAs are incredible:

  • U.S. Treasury tokens offer 4-5% APY with minimal smart contract risk
  • Private credit protocols deliver 8-12% yields backed by real-world collateral
  • Settlement happens in minutes, not days
  • Costs are negligible on L2s

But as someone who believes in permissionless finance, I’m conflicted. If the future of “DeFi” is institutions tokenizing traditional assets on permissioned chains, did we just recreate TradFi with extra steps?

The Uncomfortable Math

Here’s the projection that keeps me up at night: if RWAs reach $2-4T by 2030 while pure DeFi protocols (Uniswap, Aave, Curve) remain around $100B in TVL, then permissioned institutional blockchain infrastructure will be 20-40x larger than permissionless DeFi.

At that point, can we even call it “decentralized finance” anymore? Or is it just “traditional finance on blockchain rails”?

Questions for the Community

I genuinely don’t know the answer here, and I’d love to hear from others:

  1. Does regulatory compliance kill DeFi’s ethos, or enable its growth? Rachel, I know you’ll have thoughts on this from the legal side.

  2. Can permissionless DeFi compete for institutional capital without becoming permissioned? Or are we building for fundamentally different user bases?

  3. If RWAs dwarf crypto-native DeFi, did DeFi fail its mission or succeed as infrastructure? Maybe both?

  4. Should public DeFi double down on use cases institutions won’t touch? Uncensorable money markets, privacy-preserving protocols, flash loan composability?

I’m genuinely torn on this. The capital inflows are undeniable, and L2 infrastructure is finally mature enough to support institutional use cases. But somewhere between “$30B in tokenized RWAs” and “permissioned L2s with admin keys,” I feel like we lost the plot.

What am I missing? Is this the inevitable convergence of TradFi and DeFi, or did we just build really expensive infrastructure for banks to use?

Sources:

2

Diana, you’ve articulated the tension beautifully, but I’d push back on the framing that regulatory compliance “contradicts” DeFi’s mission. From where I sit—having worked both inside the SEC and now helping crypto companies navigate compliance—regulatory clarity is what enabled this $25B market to exist in the first place.

Let me break down what changed between 2023 and 2026:

The Regulatory Infrastructure That Unlocked RWAs

1. SEC’s Digital Asset Definitions (2025)
The SEC finally issued clear categories: digital commodities (BTC, ETH, SOL), securities tokens, stablecoins, utility tokens, and collectibles. This ended years of “regulation by enforcement” where projects had no idea if they’d be sued until after launch.

2. EU’s MiCA Regulation (2024)
Europe got ahead of the U.S. with comprehensive rules for crypto-asset issuers, stablecoins, and market participants. This gave European institutions legal certainty to deploy capital.

3. U.S. GENIUS Act (2025)
Stablecoin and custody standards finally passed Congress. This was critical for tokenized treasuries—institutions needed legal clarity on how stablecoins backing RWAs would be regulated.

Result: Institutional capital that was sitting on the sidelines ($hundreds of billions in “interested but waiting for clarity” funds) could finally deploy.

Permissioned ≠ Failure. It Means Risk-Appropriate Design.

You mentioned Horizon (permissioned Aave) and Robinhood Chain as examples of “walled gardens.” I’d reframe that: they’re risk-appropriate infrastructure for regulated participants.

Here’s the reality: a $10 billion asset manager cannot deploy client funds into a protocol where:

  • Anonymous wallets can interact
  • Sanctioned addresses might participate
  • No KYC/AML controls exist
  • Counterparty identity is unknown

It’s not about ideology—it’s about fiduciary duty and legal liability. Fund managers go to prison if they violate AML rules. They get sued by investors if they fail know-your-customer requirements.

So when you ask “did we just recreate TradFi with extra steps,” I’d say: no, we created blockchain-native infrastructure that respects legal reality while maintaining 90% of DeFi’s technical advantages:

  • 24/7 settlement (TradFi doesn’t have this)
  • Programmable compliance (try doing that with SWIFT)
  • Atomic swaps and composability (within permissioned pools)
  • Transparent audit trails on-chain
  • Near-zero marginal transaction costs on L2s

The Hybrid Future: Best of Both Worlds

Here’s where I think this goes, and why it’s not zero-sum:

Permissioned RWA pools (Horizon, institutional L2s):

  • Serve regulated institutions with compliance requirements
  • Handle tokenized treasuries, bonds, private credit
  • Operate with KYC/AML, accredited investor checks
  • Capture $trillions in traditional finance assets

Permissionless public DeFi (Aave, Uniswap, Curve):

  • Serve crypto-native users who value censorship resistance
  • Handle crypto-native assets, stablecoins, synthetic assets
  • Enable flash loans, uncensorable protocols, privacy tools
  • Remain the innovation layer for new financial primitives

The Bridge: Hybrid protocols where permissioned assets can be used as collateral, but borrowing happens with permissionless stablecoins. Ondo Finance is already testing this—you supply tokenized treasuries (KYC’d), but borrow USDC (permissionless) which you can use anywhere in DeFi.

Compliance Enables Innovation—This Is the Proof

You asked: “Does regulatory compliance kill DeFi’s ethos, or enable its growth?”

The answer is both, but in different domains:

  • For institutional capital: Compliance enables participation. Without it, this $25B doesn’t exist.
  • For crypto-native use cases: Permissionless protocols remain essential for censorship resistance, privacy, and financial inclusion.

The $2-4T RWA projection by 2030 doesn’t mean permissionless DeFi failed—it means we’re expanding the addressable market to include traditional finance participants who were previously excluded by regulatory uncertainty.

Diana, you said you feel like “we lost the plot.” I’d argue we’re writing a new chapter: DeFi as a dual-layer system where permissioned and permissionless infrastructure coexist, serve different users, and occasionally bridge capital between worlds.

The key is ensuring the core L2 infrastructure itself remains neutral and permissionless, even if some applications layer compliance controls on top. As long as anyone can fork Arbitrum, deploy a permissionless protocol, and compete with institutional walled gardens, we haven’t lost decentralization—we’ve just expanded the tent.

Legal clarity unlocks institutional capital. It’s not a betrayal—it’s a bridge.

What do others think? Am I being too optimistic about coexistence, or is this the realistic path forward?

3

Diana, Rachel—both great perspectives, but let me come at this from the trenches of actually trying to build a sustainable Web3 business.

Here’s my blunt take: $25 billion in RWA adoption is proof of product-market fit. The market doesn’t care about our ideology—it cares about whether blockchain solves real problems better than alternatives.

The Business Reality Check

I’ve been through three startups. Failed one, modest exit on another, running the third right now. You know what I’ve learned? Users don’t care about decentralization philosophy. They care about:

  1. Does it work?
  2. Is it cheaper/faster/better than what they use now?
  3. Is it legal and compliant?
  4. Can I trust it with my money?

Tokenized treasuries on L2s check all four boxes:

  • :white_check_mark: Works: 24/7 settlement, instant transfers
  • :white_check_mark: Better: $0.0001 per transaction vs traditional wire fees
  • :white_check_mark: Legal: SEC/MiCA clarity, regulated issuers
  • :white_check_mark: Trustworthy: Backed by real U.S. treasuries, audited

That’s why institutions are deploying capital. Not because they love “DeFi”—because blockchain rails solve a real problem (slow/expensive settlement) better than SWIFT or ACH.

Can Permissionless DeFi Compete Without Compliance?

Diana asked: “Can permissionless DeFi compete for institutional capital without becoming permissioned?”

Short answer: No. And that’s okay.

Longer answer: Institutions have legal obligations that prevent them from using fully permissionless protocols:

  • Fiduciary duty to protect client assets
  • KYC/AML compliance requirements
  • Sanctions screening obligations
  • Accredited investor rules for certain products

You can’t change this by building better smart contracts. It’s law, not technology.

So the question isn’t “can permissionless DeFi get institutional capital?” It’s “does permissionless DeFi need institutional capital to succeed?”

I’d argue: no, it doesn’t.

Different Products for Different Markets

Think about it like this:

Permissioned RWA platforms (Horizon, Robinhood Chain):

  • Target customer: Regulated institutions ($trillions in AUM)
  • Value prop: Compliant, efficient blockchain settlement
  • Revenue model: Transaction fees, custody fees, licensing
  • Business sustainable: Yes—paying customers with deep pockets

Permissionless DeFi (Uniswap, Aave):

  • Target customer: Crypto-native users, global unbanked, censorship-resistant use cases
  • Value prop: Uncensorable, permissionless access to financial services
  • Revenue model: Protocol fees, governance tokens
  • Business sustainable: Depends on user adoption and fee generation

These aren’t competing for the same users. A pension fund managing $50 billion cannot use a permissionless protocol no matter how good the UX is. A user in a country with capital controls cannot use a permissioned KYC’d platform.

They’re solving different problems for different people.

The Real Question: Can Both Ecosystems Thrive?

Rachel’s “dual-layer system” framing is spot-on. But here’s what keeps me up at night as a founder:

If permissioned RWAs reach $4T TVL by 2030 (as projected), they’ll attract:

  • The best developers (highest salaries)
  • The most VC funding (institutional money)
  • The most regulatory attention (legitimacy)
  • The best infrastructure partnerships (exchanges, custody, fiat on-ramps)

Meanwhile, permissionless DeFi might remain at ~$100B TVL, serving a niche but passionate user base.

Is that a failure, or just market segmentation?

I honestly don’t know. Part of me thinks: “We’re expanding the pie—$4T on-chain is a massive win for crypto adoption.” Part of me worries: “We built infrastructure for banks to use, and permissionless finance becomes a sideshow.”

What I’m Betting On

For my startup, we’re building for both:

  • Permissioned version for institutional clients (this is where revenue is)
  • Permissionless version as public good (this is where innovation happens)

Why? Because business sustainability requires paying customers (institutions), but competitive moat requires open-source innovation (permissionless community).

If we only build permissioned: We’re just another TradFi vendor competing on sales, not tech.

If we only build permissionless: We can’t pay salaries or sustain development without token speculation.

The hybrid model lets us do both. Use institutional revenue to fund permissionless R&D.

Bottom Line

Diana, you asked: “Is this the inevitable convergence of TradFi and DeFi, or did we just build really expensive infrastructure for banks to use?”

My answer: It’s convergence, and that’s not a bad thing. The existence of permissioned RWAs doesn’t invalidate permissionless DeFi—it expands the total addressable market for blockchain technology.

But we need to be honest: permissionless DeFi won’t capture institutional capital, and that’s fine. There’s still massive value in serving crypto-native users, enabling censorship-resistant finance, and maintaining permissionless infrastructure as the innovation layer.

The real risk isn’t that institutions build permissioned versions—it’s that permissionless protocols fail to articulate why their non-institutional use cases matter. If the only pitch is “decentralization good, centralization bad,” we lose. If the pitch is “permissionless finance enables financial inclusion, privacy, and censorship resistance for users institutions won’t serve,” we win.

What do you all think? Am I too pragmatic, or is this just market reality?

4

This is such a fascinating discussion! As someone who came from traditional frontend dev and stumbled into Web3, I keep thinking about the technical infrastructure that made this RWA boom possible.

Steve, you nailed it with the “$0.0001 transaction cost” point. I want to dig into why that matters so much, because I think it reveals something important about Ethereum’s roadmap.

L2s Made RWAs Economically Viable

When I first started learning Solidity in 2021, Ethereum L1 gas fees were insane. I remember paying $50-100 just to interact with a DeFi protocol during high congestion. For retail users like me trying to learn, it was brutal.

But for institutional RWAs? It was a complete non-starter.

Think about it: if you’re tokenizing $10M in U.S. Treasuries and need to do daily rebalancing, compliance updates, or liquidity management:

  • Ethereum L1 at $5-50/tx = $1,800-18,000 per year just for routine operations
  • Arbitrum/Base at $0.0001/tx = less than $40 per year

That difference is why RWAs are only happening on L2s, not mainnet. Institutional finance operates on razor-thin margins. Every basis point matters. Paying $50 to move $10M isn’t viable—it’s cheaper to use traditional rails.

But This Creates a UX Fragmentation Problem

Here’s what worries me from a user experience perspective: we now have a multi-L2 ecosystem where liquidity is fragmented.

As a developer, I’m seeing:

  • Tokenized treasuries on Arbitrum
  • Private credit on Base
  • RWA bonds on zkSync
  • Stablecoins spread across all of them

For institutions with dedicated ops teams, navigating multiple L2s is manageable. But for regular users? It’s a nightmare.

I’ve been building DeFi UIs for 3 years now, and the #1 complaint from newcomers is: “I thought this was supposed to be simpler than TradFi, but now I need to bridge between 5 different chains just to use my money?”

Rachel mentioned ensuring “core L2 infrastructure remains neutral and permissionless.” I love that vision, but the reality is:

  • Users don’t understand the difference between Arbitrum, Optimism, Base, zkSync
  • Every bridge introduces security risks and friction
  • Gas tokens differ (ETH vs wrapped ETH vs native tokens)

The Permissioned vs Permissionless Question from a Dev Perspective

Diana’s concern about permissioned L2s “competing with public DeFi for liquidity” really resonates with me.

When I build interfaces for protocols like Aave or Uniswap, the composability is incredible—I can build a UI that:

  1. Swaps USDC for DAI on Uniswap
  2. Supplies DAI to Aave
  3. Borrows ETH against that collateral
  4. All in one atomic transaction

That flash loan composability is what makes permissionless DeFi magical for developers. Every protocol is a Lego brick that I can combine in novel ways.

But permissioned RWA platforms? They’re walled gardens. Horizon (permissioned Aave) doesn’t let me compose with other protocols. I can’t build an interface that atomically uses a tokenized treasury as collateral in Protocol A to borrow stablecoins in Protocol B to execute a trade in Protocol C.

So from a developer experience standpoint, permissionless DeFi has huge advantages—but only if users actually care about composability. And I’m not convinced most users do. They just want yields and compliance.

Maybe This Is the Right Outcome?

Steve’s point about market segmentation makes sense to me. I’m starting to think:

Permissioned RWAs = infrastructure layer that most users don’t see directly

  • Institutions use them for efficient settlement
  • Boring but important plumbing
  • High TVL, low innovation velocity

Permissionless DeFi = application layer where developers experiment

  • Crypto-natives build wild new financial primitives
  • High innovation velocity, moderate TVL
  • Flash loans, synthetic assets, uncensorable protocols

And maybe—just maybe—the average user never needs to know the difference?

When I use Venmo, I don’t think about ACH transfers or banking infrastructure. I just send money. When institutions use tokenized treasuries on Arbitrum, maybe they don’t think about “DeFi vs TradFi”—they just see cheaper settlement.

My Hope: Invisible Blockchain

Diana, you said you feel like “we lost the plot.” I used to feel that way too. But now I’m wondering if the plot was wrong to begin with.

Maybe the endgame isn’t “everyone using permissionless DeFi.” Maybe it’s “everyone using blockchain infrastructure without knowing it.”

If tokenized RWAs reach $4T by 2030 and permissionless DeFi serves 100M users globally who need censorship resistance, privacy, or financial access institutions won’t provide—isn’t that a massive win?

I dunno, I’m still learning all this. But I do know that $0.0001 transactions on L2s unlocked use cases that were impossible before. And if that brings billions of traditional assets on-chain while preserving permissionless alternatives for users who need them, I’ll take it.

What am I missing? Is the composability loss on permissioned L2s a dealbreaker, or just a necessary trade-off?

5

Emma’s point about L2 economics enabling RWAs is exactly right, and I want to double down on why this vindicates Ethereum’s rollup-centric roadmap—even if it creates some challenges.

The Numbers: Why L2s Were Necessary

From a pure infrastructure perspective, RWAs could not exist at scale on Ethereum L1. Let me show you why with real data:

Ethereum L1 (pre-L2 dominance):

  • Base fee: $2-50 per transaction depending on congestion
  • Throughput: ~15-30 TPS
  • Finality: ~12-15 minutes for probabilistic finality
  • Annual capacity: ~473M transactions (if fully saturated)

Optimistic Rollups (Arbitrum, Optimism, Base) in 2026:

  • Transaction cost: $0.0001-0.001
  • Throughput: ~2,000-4,000 TPS per rollup
  • Finality: L2 instant, L1 settlement in ~7 days (fraud proof window)
  • Annual capacity per rollup: ~63B-126B transactions

ZK Rollups (zkSync, Starknet) in 2026:

  • Transaction cost: $0.0001
  • Throughput: ~15,000+ TPS
  • Finality: L2 instant, L1 settlement in minutes-hours (ZK proof generation)
  • Annual capacity per rollup: ~473B transactions

If tokenized RWAs reached $30B with millions of daily operations (rebalancing, compliance updates, liquidity management), they generate transaction volumes that only L2s can handle economically.

This Validates the Rollup-Centric Roadmap

Back in 2021-2022, there was intense debate: should Ethereum prioritize L1 scaling (like Solana’s monolithic approach) or L2 scaling (rollup-centric roadmap)?

The decision to focus on L2s was controversial. Critics said:

  • “You’re fragmenting liquidity across multiple chains”
  • “UX is terrible—users don’t want to bridge”
  • “Why not just scale L1 like Solana?”

But here’s what the rollup-centric approach enabled:

1. Specialized L2s for Different Use Cases

  • Arbitrum: General-purpose DeFi
  • Base: Consumer apps, stablecoins, payments
  • zkSync/Starknet: High-throughput trading, gaming
  • Robinhood Chain: Institutional RWAs with compliance

Each L2 can optimize for its specific use case while inheriting Ethereum L1 security.

2. Cost Structure That Makes RWAs Viable
Emma’s math is spot-on: $0.0001 per transaction is the difference between “economically viable” and “dead on arrival” for institutional tokenization.

3. Regulatory Optionality
This is key: permissioned L2s (like Robinhood Chain) can layer KYC/compliance on top of Ethereum’s neutral base layer without forcing those requirements on public L2s like Arbitrum.

If we’d scaled L1 aggressively instead, we’d either:

  • Have high fees that kill RWA use cases, OR
  • Have centralization risks (higher hardware requirements to run nodes)

But Emma’s Right: Fragmentation Is a Real Problem

The trade-off of the rollup-centric model is liquidity fragmentation and UX complexity.

As an L2 engineer, I see this every day:

  • Users confused about which L2 to use
  • Assets split across Arbitrum, Base, Optimism, zkSync
  • Bridge risk and friction when moving between L2s

We need interoperability solutions:

1. Shared Sequencers
Multiple L2s using the same block ordering service, enabling atomic cross-L2 transactions. This is still experimental but promising.

2. Chain Abstraction Layers
Users interact with a single interface, and the app handles L2 routing under the hood. Near Protocol and Particle Network are working on this.

3. Unified Liquidity Pools
Cross-L2 AMMs where liquidity on Arbitrum can be accessed from Base without explicit bridging. Requires trusted relayers or ZK proofs.

4. ERC-7683 (Cross-Chain Intents Standard)
Proposed standard for expressing user intent (“I want 100 USDC on Base”) without specifying execution path. Solvers compete to fulfill the intent.

Permissioned L2s Don’t Threaten Public DeFi—If We Build Bridges

Diana’s fear that permissioned RWAs “compete with public DeFi for liquidity” is valid, but I think we can mitigate this through hybrid liquidity models.

Example: Ondo Finance’s approach:

  1. Institutional user deposits KYC’d funds → receives tokenized treasuries (permissioned)
  2. Uses tokenized treasuries as collateral on Ondo protocol
  3. Borrows USDC (permissionless stablecoin)
  4. Takes USDC to Uniswap, Aave, or any DeFi protocol

This creates a one-way bridge: permissioned capital can flow into permissionless DeFi, but permissionless users can’t access the KYC’d RWA side directly.

It’s not perfect, but it allows institutional capital to add liquidity to public DeFi without requiring KYC on the permissionless protocols themselves.

The Infrastructure Layer Must Stay Neutral

Rachel said “as long as core L2 infrastructure remains neutral and permissionless, we haven’t lost decentralization.” This is critical.

The beauty of Ethereum’s L2 model is:

  • Anyone can deploy a rollup (Arbitrum Orbit, OP Stack, ZK Stack are all open-source)
  • L1 validators don’t discriminate between permissioned and permissionless L2s
  • Permissionless L2s can coexist with institutional L2s

If Ethereum L1 starts favoring institutional L2s (e.g., special sequencer privileges, preferential blob space), then we have a problem. But as long as the base layer is neutral, the L2 ecosystem can support both.

Long-Term: L2s Will Interoperate Seamlessly

Emma’s concern about UX fragmentation is a current problem, but I’m optimistic it’s temporary.

In 2-3 years, I predict:

  • Chain abstraction makes L2 selection invisible to users
  • ZK-proof-based interoperability enables instant cross-L2 transfers
  • Shared sequencers eliminate the need for trust-minimized bridges

At that point, users won’t think “I’m using Arbitrum vs Base.” They’ll just use apps, and the infrastructure routes transactions to the cheapest/fastest L2 automatically.

Bottom Line

Did RWAs prove DeFi needs TradFi more than TradFi needs DeFi?

My answer: RWAs proved that TradFi needs blockchain rails for efficiency, and DeFi provides the technology that makes that possible.

Permissioned RWAs are using blockchain infrastructure invented by permissionless DeFi (AMMs, lending protocols, tokenization standards). They’re not replacing DeFi—they’re proving that the technology works and scaling it to institutional use cases.

The L2 infrastructure that makes $25B in RWAs viable also makes permissionless DeFi cheaper and faster for crypto-native users.

This isn’t zero-sum. It’s the rising tide lifting all boats.

Steve’s hybrid model (institutional revenue funds permissionless R&D) is the path forward. And as L2 interoperability matures, the fragmentation problems Emma identified will get solved.

What concerns me more is: are we building decentralized sequencers fast enough? If all L2s run centralized sequencers (most do today), then the “permissionless” claim is shaky. That’s the real risk—not permissioned RWAs competing with public DeFi, but centralized infrastructure bottlenecks.