The Toll Wars Are Ending. The Mint Wars Are Starting.
Base deployed Beryl to testnet on June 19 with mainnet set for June 25. Its centerpiece, B20, puts token issuance inside the node software as a precompile instead of a smart contract, with a compliance toolkit for stablecoin and RWA issuers.
Summary: For most of the last three years, the L2 pitch was a number on a receipt. Cheaper gas, faster confirmation, a sub-cent swap. Everyone was selling the same thing, a cheaper ride, and the only argument was the toll. This week Base changed the subject.
On June 19 it pushed its Beryl upgrade to the Sepolia testnet with mainnet locked for June 25, and the headline feature is not about your transaction at all. It's about the asset. B20 is a token standard built directly into the node software as a precompile, written in Rust, running inside the protocol rather than as EVM bytecode.
It ships with an issuer toolkit aimed squarely at stablecoin and real-world-asset issuers: role-based permissions, mint and burn controls, supply caps, transfer restrictions, and freeze and seizure hooks for regulated tokens. Native execution roughly halves transfer cost, the standard single-proof withdrawal window to Ethereum drops from seven days to five, and Reth V2 comes along to cut node storage and lift throughput.
Read past the spec sheet and the move is plain. Base is optimizing the supply side of onchain assets. It wants to be the cheapest, cleanest place to bring a new asset into existence, and it just put the machinery to do that into the engine itself.
Thesis: The competitive frontier on L2s is moving upstream, from the cost of using an asset to the cost of creating one. Beryl is the clearest version of that shift to date because B20 lives inside the chain rather than on top of it as a contract you deploy.
The token standard is part of the protocol now. Mapping it to the framework, this is Intermediation (Services) × Programmable Logic + Composable Infrastructure doing the heavy lifting, with Settlement (Rails) × Transparent Verification underneath it by way of Base's Azul multiproof system that went live in May. When you fold a token standard, compliance tooling, and the verification path into the execution environment, you collapse the work of issuing a programmable asset from an engineering project into something closer to filling out a form.
The non-obvious part: this matters more than another quarter of tokenized-volume headlines. Volume is demand. Issuance economics are supply. Whoever makes the marginal cost of minting a credible, composable, verifiable asset approach zero gets to host the next decade of them.
The Open Money lens
The framework reads infrastructure across three layers, Settlement, Intermediation, and Coordination, and five dimensions of openness: Permissionless Access, Transparent Verification, Programmable Logic, Composable Infrastructure, and Sovereign Custody. Most weeks the interesting news lands in a cell or two. This week the interesting thing is where in the stack the news lives.
Almost every L2 improvement of the past two years has targeted the usage stage, the moment a person sends, swaps, or settles. That is the demand side of the matrix, and it is largely a Settlement-layer story about making each transaction cheaper and faster.
Beryl reaches into a different part of the matrix. It targets the creation stage, the moment an asset comes into being and acquires its rules. That is an Intermediation-layer story, because issuing and administering an asset is a service, and Base just turned that service into a protocol primitive. The distinction sounds academic until you notice that the entire RWA and stablecoin thesis depends on it.
You cannot tokenize the world if every new instrument requires a bespoke contract, an audit, and a prayer. You can if minting one is native.
What Base actually shipped
Strip the marketing and Beryl is three things stacked together, and the order matters.
The first and largest is B20. A normal token on an EVM chain is a smart contract: you write Solidity, you deploy bytecode, the network runs it like any other application. B20 inverts that.
The token logic is compiled into the node software as a precompile, so a B20 asset executes as part of the protocol rather than as a program sitting on top of it. The practical payoff is mundane and large at once.
Native execution cuts transfer costs by roughly half, and because the standard supports the full ERC-20 spec plus ERC-2612 permit signatures, existing wallets, exchanges, and indexers handle a B20 token with no changes. It looks like an ERC-20 to everything downstream. It just costs less and lives lower in the stack.
The second thing is the part regulators will read first. B20 ships with an issuer toolkit that bakes compliance controls into the standard itself: role-based access, mint and burn authority, optional supply caps, transfer restrictions, and freeze and seizure capabilities. Base offers issuers a general-purpose model and a stablecoin-specific model with six-decimal precision and a customizable currency code.
This is Base saying out loud who it built the standard for: the regulated issuer who needs to freeze a wallet under a court order and prove the supply cap to an auditor, and who would rather inherit those controls from the chain than write and audit them again. The memecoin deployer was never the design target.
The third thing is plumbing, and it is the tell that this was a deliberate sprint rather than a one-off. The single-proof withdrawal window from Base to Ethereum shrinks from seven days to five, and Reth V2 lands as the execution client, cutting node storage and lifting throughput.
Base credits the speed to its February decision to stop sharing Optimism's OP Stack and run its own unified stack. Beryl arrived about four weeks after the previous upgrade hit mainnet. A chain that ships independent hard forks a month apart is a chain that has decided cadence is a weapon.
Why "in the node" is the whole story
The reason B20 is more than a gas optimization is that it changes where trust and capability sit.
When token logic is a contract, every issuer reinvents the same wheel and re-incurs the same risk. They reimplement transfer rules, they reimplement access control, they pay for the same audit of the same patterns, and each new asset is a fresh attack surface.
When that logic is a precompile, the chain carries it once, audits it once, and every issuer inherits it. Base says B20 was audited by its own team and the security firm Spearbit, and a future update will let issuers pay fees in their own B20 token rather than ETH, which quietly removes one of the last frictions in operating an asset, keeping ETH around just to move your own stablecoin.
That is what "optimizing the supply side" means in concrete terms. The marginal asset gets cheaper to mint, cheaper to run, and safer by default, because the safety is the chain's problem now instead of the issuer's.
For a stablecoin desk or a tokenization platform deciding where to put its next product, that is a different kind of argument than "our swaps are a tenth of a cent cheaper." It is the difference between a toll road competing on price and a city offering you the deed to build on.
Verifiability is the part that makes it bankable
None of the issuance story works if moving the asset is sketchy, which is where Azul matters and why Beryl sits on top of it rather than beside it.
Azul reached mainnet in late May as Base's first fully independent upgrade, and its core is a multiproof system that combines a trusted-execution-environment proof with a zero-knowledge proof.
Either can finalize a block on its own, and a permissionless ZK proof can override a permissioned TEE proof in a conflict, which is the design that pushes Base toward Stage 2 decentralization and gives withdrawals a roughly one-day fast path when both proofs agree. The honest footnote is that the fast path has seen limited use because generating ZK proofs is still expensive, which is exactly why Beryl spends its effort shortening the ordinary route instead.
Put the two upgrades together and the strategy reads cleanly. Azul makes the rails verifiable enough to carry serious assets. Beryl makes those assets cheap and native to create. A regulated issuer cares about both halves: cheap to mint is worthless if the bridge home is a trust fall, and a bulletproof bridge is irrelevant if standing up the asset is a six-week engineering slog. Base is building both ends of the same pipe in the same quarter.
The play is credible because Base is not starting from nothing. As of late May it held roughly $4.4 billion in value locked and nearly $5 billion in stablecoins, the leading home for USDC liquidity among optimistic rollups. A chain with that much stablecoin already parked on it building a native standard for stablecoin issuers is aiming the upgrade at demand it can already see, rather than at a market it hopes will show up.
The rest of the stack is pulling the same direction
Base is the sharp edge this week, but the pattern shows up elsewhere, and the throughline is the same: take cost and human overhead out of the parts of the system that used to require effort and attention.
Arbitrum kept its coordination machinery running on schedule, moving a $43.5 million 2027 budget to an onchain vote in early June even with the market soft, which is the governance-as-infrastructure point from the last issue still holding.
NEAR's Dynamic Resharding remains on track in network upgrade 2.13, letting the protocol split a congested shard with no vote and no human in the loop. Those two were the spine of the June 7 issue, so I'll keep them brief here.
Daniel McGlynn
Here's a previous issue that touches more about protocol architecture changes.
The reason they belong in this week's frame at all is that they reduce the ongoing operating drag on anything that settles or scales on those rails, including newly issued assets. Cheap to mint on Base, cheap to keep alive on a chain that reshards itself, governed by a DAO that votes on time. The reductions stack.
The honest case against this read
A careful reader should push here, and there are real soft spots.
Start with lock-in. B20 is ERC-20 compatible at the interface, but it is a Base-specific standard implemented in Base's node software. An asset minted as a B20 inherits Base's roadmap and Base's risk in a way that a portable, chain-agnostic contract does not.
"Native" and "captured" can be the same sentence read in two moods, and an issuer betting its instrument on one chain's precompile is making a bet about that chain's longevity whether it frames it that way or not.
Then the compliance hooks, which cut both ways. Freeze and seizure baked into the standard is genuinely useful for a regulated stablecoin and genuinely uncomfortable as a default property of the rails.
A toolkit that makes it trivial to restrict transfers is a toolkit that makes permissioned money easy to ship, and "easy to ship" is not the same as "good that it exists." The openness dimensions of the framework are not all pulling the same way here. Programmable Logic goes up. Permissionless Access, for assets that use the restrictive features, goes down.
And the whole thing is, for now, a testnet and a press cycle. Beryl hits mainnet June 25. B20 has one audit and zero assets of consequence issued under it. The claim that issuance economics are the new battleground is a thesis about where competition is heading, not a settled fact, and it leans on a standard that has not yet had to survive contact with a real issuer at scale.
The strongest version of the skeptical case is simply this: cheaper minting only matters if the bottleneck on tokenization was ever the cost of minting, rather than law, distribution, and trust. Beryl does almost nothing about those three.
That last point is the one I take most seriously, and it does not dissolve the structural read so much as bound it. The bet is narrower than B20 single-handedly making tokenization happen. Among the things infrastructure can actually move, the cost and safety of creating an asset is one, and Base just moved it hard while everyone else was still advertising the toll.
What to watch
The thesis that L2s are repositioning as issuance platforms is testable, and the next two months give clean reads.
The first is the June 25 mainnet activation and what gets minted in the weeks after. A B20 stablecoin or tokenized fund from a name issuer, rather than a demo, is the single piece of evidence that would turn this from a well-built standard into an adopted one. Watch who issues first and whether they use the restrictive compliance features or leave them off.
The second is whether anyone matches it. If issuance economics are really the new front, a competing L2 should announce its own native-issuance answer before long. Silence from the other major rails would suggest Base read the moment alone. A fast follow would confirm the front is real.
The third is the lock-in question playing out in practice. If B20 assets stay strictly on Base while bridged representations proliferate elsewhere, the portability cost is real and issuers will price it. If tooling emerges to move B20 assets cleanly across chains, the lock-in worry fades and the standard's reach grows.
The fourth is Cobalt, the upgrade Base says is targeted for September and slated to bring native account abstraction and more B20 functionality. A second issuance-focused upgrade three months after this one would settle whether Beryl was a one-time swing or a sustained strategy.
Strategic implications
For builders. The defensible work is moving upstream, from the application that uses an asset to the primitives that create one. If your product is a generic ERC-20 deployer or a contract-based compliance wrapper, a precompile that does the same thing natively and for half the gas is a direct competitor with the chain's balance sheet behind it.
The opportunity is in the layer above the primitive: issuance flows, lifecycle management, cross-chain portability for native standards, tooling that makes launching a B20 asset feel like filling out a form rather than commissioning an audit. Build for the world where minting is cheap and the hard part is everything around it.
For capital allocators. Stop scoring L2s only on TVL and fee revenue, which measure demand, and start scoring them on issuance posture, which measures supply. A chain shipping native token standards, baked-in compliance, and verifiable settlement is positioning to be the default home for the next generation of programmable assets, and that position compounds in a way that a transient fee advantage does not.
The question to ask about any L2 now is how cheap, clean, and safe it is to bring a new asset into existence there, and who is actually choosing to do it, rather than how cheap it is to transact.
For policymakers. Compliance tooling embedded at the protocol level, with verifiability underneath it, is a more monitorable arrangement than the same controls bolted onto an application after the fact, and it is worth understanding rather than reflexively distrusting.
A B20 stablecoin with auditable supply caps and a documented freeze mechanism is, in principle, easier to supervise than an opaque offchain issuer. The design also concentrates a lot of power in whoever controls the standard and its restrictive features, which is the thing to watch. Frameworks that recognize protocol-level compliance as legitimate, while keeping the controls auditable and contestable, get more transparency than rules that force every issuer back through legacy intermediaries.
Daniel McGlynn
Last week's issue, ICYMI.
Sources
[1] crypto.news. "Base rolls out Beryl testnet upgrade with native token standard." June 19, 2026. https://crypto.news/base-rolls-out-beryl-testnet-upgrade-with-native-token-standard/
[2] Base Engineering Blog. "Introducing Base Beryl." June 2026. https://blog.base.dev/introducing-base-beryl
[3] The Block. "Base targets June 25 mainnet launch for Beryl upgrade and new B20 token standard." June 2026. https://www.theblock.co/post/405410/base-targets-june-25-mainnet-launch-for-beryl-upgrade-and-new-b20-token-standard
[4] The Block. "Base launches Azul on mainnet, pushing Coinbase's Ethereum L2 toward full decentralization." May 2026. https://www.theblock.co/post/403003/base-launches-azul-on-mainnet-pushing-coinbases-ethereum-l2-toward-full-decentralization
[5] CoinMarketCap Academy. "Base Activates Azul Upgrade on Ethereum L2 Mainnet." May 2026. https://coinmarketcap.com/academy/article/base-azul-upgrade-mainnet
[6] Bloomingbit. "Arbitrum Foundation Seeks $43.5 Million for 2027 Operations, Vote Starts June 8." June 2026. https://en.bloomingbit.io/feed/news/113152
[7] CoinDesk. "Near Protocol to automate its own growth and its token is skyrocketing." May 22, 2026. https://www.coindesk.com/markets/2026/05/22/near-protocol-to-automate-its-own-growth-and-its-token-is-skyrocketing
[8] RWA.xyz. Analytics on tokenized real-world assets and stablecoin supply. https://app.rwa.xyz/