Caldera is a Rollup-as-a-Service (RaaS) provider that lets teams deploy production-grade Layer 2 and Layer 3 chains across the OP Stack, Arbitrum Orbit, ZK Stack, Polygon CDK, and Eclipse SVM frameworks. The platform handles sequencer operation, batcher posting, prover infrastructure, settlement wiring, data availability, and chain monitoring so applications get dedicated blockspace without rebuilding the rollup stack from scratch. By Q1 2026, Caldera-deployed chains include Manta Pacific, Kinto, B3, Plume, RARI Chain, ApeChain, and Treasure, with cumulative TVL across Caldera rollups tracked on L2BEAT's rollup summary. This guide covers what Caldera does, how rollups deployed on Caldera actually work, the stacks and settlement layers it supports, the Metalayer interoperability product, the RaaS competitive landscape against Conduit, AltLayer, and Alchemy Rollups, and the trade-offs teams take on when they choose dedicated blockspace over a shared L2.
​

Caldera is a rollup deployment and operations platform founded in 2023 and headquartered in San Francisco. Teams use Caldera to launch a custom Layer 2 or Layer 3 chain, configure parameters like gas token, block time, and data availability layer, and outsource the day-to-day operation of sequencer, batcher, and prover infrastructure to Caldera's engineering team. Documentation lives at docs.caldera.xyz, the marketing surface at caldera.xyz.
​

The pitch is straightforward. Running a rollup is closer to running a small bank than running a smart contract. Sequencers need 24/7 uptime, batchers need to post calldata or blobs to settlement on schedule, provers need to generate validity proofs (for ZK chains) or fault-proof games (for optimistic chains), and monitoring needs to catch reorgs, missed batches, and gas-price spikes before users do. Caldera does all of that as a managed service, charging a per-chain operating fee and revenue share on sequencer fees.
​

By April 2026, the company had raised approximately $25M in Series A funding led by Founders Fund and Dragonfly, with Sequoia Capital, Lattice, Neo, and 1confirmation participating. The Caldera blog at blog.caldera.xyz publishes chain launches, the Metalayer roadmap, and partnership announcements. Production rollups operated by Caldera are visible on L2BEAT, which tracks TVL, security model, and stage classification for every public rollup.
​

Caldera is not the only RaaS provider, and the market has consolidated around four serious operators (Caldera, Conduit, Alchemy, AltLayer), each with different stack coverage and pricing. The "Caldera vs Conduit vs Alchemy Rollups vs AltLayer" section below covers that landscape in detail.
​

To understand what Caldera operates, it helps to walk through the four moving parts of a rollup: sequencer, batcher, prover (or fault-proof system), and settlement.
​

The sequencer is the node that orders user transactions and produces blocks for the rollup. It runs continuously, accepting transactions through a JSON-RPC endpoint and emitting an L2 block stream that wallets and indexers consume. Caldera operates the sequencer as a hosted service. Block times are configurable per chain, with B3 (gaming, sub-second blocks) and Plume (RWA, two-second blocks) representing typical extremes.
​

The batcher compresses sequencer output and posts it to the settlement layer. On an Ethereum-settled OP Stack chain, the batcher writes EIP-4844 blob data to L1 once every several blocks, which is where most of the chain's operating cost shows up. Optimism's stack documentation describes the batcher contract pattern. Caldera tunes batch frequency to balance L1 cost against finality latency.
​

The prover is required for ZK rollups. It generates validity proofs that compress the entire L2 state transition into a SNARK or STARK that the L1 verifier contract can check. ZK Stack chains use Boojum (Matter Labs' STARK + SNARK prover); Polygon CDK chains use the Polygon Zero prover. Optimistic rollups skip this step but require an interactive fault-proof system, which the OP Stack added with the Cannon fault-proof program in 2024 and Arbitrum has had through Nitro since 2022.
​

Settlement is the final L1 contract that holds the rollup's state root and adjudicates withdrawals. For most Caldera chains the settlement layer is Ethereum L1. Some chains use alternative data availability (Celestia, EigenDA, NEAR DA, Avail) while keeping settlement on Ethereum, which lowers blob cost without giving up Ethereum-grade settlement security. The Avail project and EigenLayer publish their respective DA throughput numbers in their docs.
​

Caldera also maintains an internal control plane for chain operators. The dashboard exposes sequencer health, batch posting status, gas price across L1 and DA, and per-chain telemetry. Chain operators (the team that owns the rollup) get logs and incident response without running their own SRE function.
​

Caldera is stack-agnostic by design. A team picks the rollup framework that matches the application, then Caldera handles the deployment. Five stacks are in production by Q1 2026.
​

The OP Stack is Optimism's open-source rollup framework, used by Optimism Mainnet, Base, Mode, Zora, and many others. It runs as an optimistic rollup with EVM equivalence, meaning Solidity contracts deploy without changes. Caldera deploys OP Stack chains as part of the Optimism Superchain when the operator opts in, which gives shared bridging and a unified governance surface. B3 (gaming-focused, built by Layer3) and Manta Pacific (zkEVM-mode OP Stack via Polygon zkEVM Validium) both run on Caldera using the OP Stack.
​

Arbitrum Orbit lets teams launch a custom chain that settles to Arbitrum One or Nova rather than directly to Ethereum, lowering settlement cost and inheriting Arbitrum's fraud-proof system through Nitro. Orbit chains can pick their own gas token (any ERC-20) and run as either an L2 (settling directly to Ethereum) or an L3 (settling to Arbitrum). XAI, ApeChain (Yuga Labs), and several gaming chains use Orbit on Caldera.
​

The ZK Stack from Matter Labs powers ZKsync Era and lets third parties launch dedicated ZK chains called Hyperchains. Validity proofs are generated by the Boojum prover, and chains can share liquidity and messaging through ZKsync's Elastic Network. Caldera supports ZK Stack deployments where teams want validity-proof finality and EVM compatibility through ZKsync's modified zkEVM.
​

The Polygon CDK packages Polygon's zkEVM into a chain development kit. CDK chains can settle to Ethereum directly or aggregate through the AggLayer, Polygon's settlement aggregation contract that batches proofs from multiple CDK chains into a single L1 verification. Manta Pacific shifted to a CDK-based architecture in 2024 for improved proof costs.
​

Eclipse runs the Solana Virtual Machine as an Ethereum L2. The chain executes Solana programs, settles to Ethereum, and uses Celestia for data availability. Caldera supports Eclipse-compatible deployments for teams that want SVM throughput (parallel execution, higher TPS) with Ethereum settlement. The Eclipse documentation describes the SVM-EVM bridge and developer tooling.
​

Every rollup makes two architectural decisions independent of the execution stack: where state roots settle, and where transaction data goes. The settlement layer determines the security model. The DA layer determines the operating cost.
​

Ethereum L1. The default settlement target for most Caldera chains. EIP-4844 blobs (live since March 2024) cut the cost of posting calldata roughly 10x relative to the prior calldata pricing, which is why most rollups remained on Ethereum DA rather than migrating to alt-DA. ApeChain, Plume, and several Orbit chains use Ethereum settlement and Ethereum DA.
​

Celestia. A purpose-built DA chain that lets rollups post transaction data at a lower cost than Ethereum blobs while still settling to Ethereum. Manta Pacific runs on Celestia DA, and Eclipse uses Celestia by default. Celestia's namespace model lets a single rollup pull only its own data shards rather than the entire DA stream.
​

EigenDA. A restaked DA service built on EigenLayer. Restakers commit ETH (or LSTs) to back DA correctness, and operators run nodes that store and serve blob data. EigenDA targets higher throughput than Ethereum blobs at lower cost, and Caldera supports it as a DA option for teams that want to inherit Ethereum's security through restaking.
​

NEAR DA. NEAR Protocol exposes its sharded state as a data availability target through the Nightshade architecture. Caldera supports NEAR DA for teams that want sub-cent per-transaction DA cost.
​

Avail. Polygon-spinoff Avail provides modular DA with KZG commitments and light-client verification. The Avail project publishes throughput and finality numbers; several Caldera-deployed chains run Avail in production.
​

The decision matrix is concrete. A high-volume gaming chain like B3, with thousands of transactions per second, picks a cheaper DA layer because Ethereum blobs would dominate operating cost. A treasury-heavy chain like Plume, with low transaction volume but high value per transaction, stays on Ethereum DA for the strongest security model. Caldera's role is to make the settlement and DA layers configurable per-chain rather than locking operators into one combination.
​

Caldera operates dozens of chains in production. The list below covers the chains with the largest activity or notable use cases as of Q1 2026.
​

Manta Pacific. A modular L2 originally built on the OP Stack with Celestia DA, later migrated to Polygon CDK for tighter proof economics. Manta is one of Caldera's longest-running chains and supports DeFi protocols and ZK-native applications. Manta Pacific TVL is tracked on L2BEAT.
​

Kinto. A KYC-permissioned L2 designed for institutional DeFi, with onchain identity primitives and modular smart-account support. Kinto users complete KYC at the chain level rather than per-application, which addresses one of the recurring blockers for institutional onchain deployment.
​

B3. A gaming-focused L2 from Layer3 that runs sub-second blocks and uses an OP Stack base. B3 hosts game studios and game-related applications that need fast finality and low gas costs. B3 launched mainnet in 2024 and ranks among the most active gaming-specific rollups in 2026.
​

Plume. A real-world-asset (RWA) chain that runs onchain compliance primitives at the chain level, including KYC, accreditation checks, and transfer restrictions. Plume's documentation at plumenetwork.xyz describes the asset onboarding pipeline. The chain hosts tokenized treasuries, private credit, and other RWA primitives.
​

RARI Chain. A NFT-focused L2 launched by Rarible. The chain enforces creator royalties at the protocol level, which is a deliberate response to royalty-bypass marketplaces on Ethereum and Solana. RARI Chain settles to Ethereum and uses Arbitrum Orbit.
​

ApeChain. Yuga Labs' Orbit chain that uses APE as the native gas token. ApeChain hosts BAYC-related applications, NFT trading, and ApeCoin staking. The chain launched in late 2024 on Caldera and ranks among the largest gas-token-customized Orbit chains.
​

Treasure. A gaming chain that previously ran as an Arbitrum-native sidechain and now runs as an Orbit chain on Caldera. Treasure hosts MAGIC-denominated games and the Treasure marketplace.
​

Other notable Caldera chains include XAI (gaming on Orbit), Lyra (derivatives), Injective EVM (EVM extension on Injective via Caldera), and several validator-set or appchain experiments that are smaller in TVL but specific to particular teams.
​

The Caldera Metalayer is an interoperability product that connects Caldera-deployed chains. The launch was announced at blog.caldera.xyz and the product page sits at caldera.xyz/metalayer. Rollouts proceed through 2026, with new Caldera chains shipping Metalayer compatibility by default.
​

The problem Metalayer solves is fragmentation. Every new rollup creates a new island. A user holding USDC on ApeChain can't trivially spend it on B3 because the two chains have separate state, separate liquidity, and separate (or no) bridging. Across the broader L2 ecosystem this fragmentation is solved by general-purpose bridges and intent networks; Metalayer is Caldera's attempt to solve it specifically for chains they operate, where Caldera has direct access to sequencer infrastructure on both ends.
​

Mechanically, Metalayer combines a messaging primitive (built on existing protocols rather than reinventing one) with a sequencer-level coordination service that lets transactions on chain A trigger state on chain B without waiting for both to finalize on Ethereum. For users this looks like a single click crosses two Caldera chains in seconds. For developers it exposes a contract surface for cross-chain calls scoped to the Caldera rollup network.
​

Metalayer is an additive product, not a replacement for general bridges. Cross-chain movement off the Caldera network (to Ethereum, Solana, Polygon, Arbitrum One, Base, etc.) still uses third-party bridges or the underlying stack's canonical bridge. Metalayer's coverage is the Caldera-operated chain set.
​

Four RaaS providers serve serious production traffic in 2026. The choice between them comes down to stack coverage, pricing model, and the operator's risk appetite.
​

Caldera. Stack coverage is the broadest of the four, supporting OP Stack, Arbitrum Orbit, ZK Stack, Polygon CDK, and Eclipse SVM. Caldera's pitch leans on engineering depth and the Metalayer interoperability product. Notable customers: Manta Pacific, Kinto, B3, Plume, RARI Chain, ApeChain, Treasure.
​

Conduit. The Conduit platform is OP Stack and Arbitrum Orbit-focused, with a self-serve console that lets teams deploy testnets in minutes. Conduit operates Proof of Play's Pirate Nation, Aevo, Lyra, Public Goods Network, and several Optimism Superchain members. Pricing is published as a flat monthly fee plus revenue share, which contrasts with Caldera's negotiated model.
​

Alchemy Rollups.Alchemy Rollups launched in 2023 and integrates with Alchemy's existing developer platform (Embedded Wallets, Smart Wallets, indexing infrastructure, RPC). The pitch is a fully integrated stack rather than rollup ops in isolation. Coverage is OP Stack and Arbitrum Orbit primarily, with newer Alchemy chains targeting AA-native and gas-sponsorship-native designs.
​

AltLayer.AltLayer's documentation covers their RaaS platform and the Restaked Rollups concept, which uses EigenLayer-style restaking to back fast finality. AltLayer supports OP Stack, Arbitrum Orbit, ZK Stack, and Polygon CDK, and operates ephemeral rollups that live for the duration of a single application or event. Customers include Cyber, Xterio, and several gaming chains.
​

The differentiation is real but narrower than the marketing suggests. All four can run an OP Stack or Orbit chain. The choices that distinguish them are non-EVM coverage (Eclipse SVM is Caldera-specific in production), pricing structure (Conduit's published fees vs Caldera's negotiated rates), bundled tooling (Alchemy's wallet and RPC infrastructure), and interop products (Metalayer, AltLayer's restaking-secured fast bridge). Operators tend to evaluate two providers and pick the one that aligns better on stack and pricing.
​

The recurring question for any team considering RaaS is whether the application needs its own chain. Most don't. Deploying on Base, Arbitrum One, or Optimism gets a team to production in weeks instead of months and inherits liquidity that a fresh rollup has to bootstrap.
​

The conditions that actually justify a dedicated rollup are narrow.
​

Dedicated blockspace. A high-throughput application (gaming with sub-second blocks, derivatives with deterministic gas pricing) gets squeezed during peak demand on a shared L2. ApeChain, B3, and Treasure all picked dedicated chains because their applications generate enough volume that competing for L2 blockspace would degrade UX.
​

Custom gas tokens. Orbit and OP Stack with custom-gas-token modules let chains use any ERC-20 as gas. ApeChain (APE), Treasure (MAGIC), and several other Orbit chains use this pattern. The economic argument is that the chain's native token captures gas revenue rather than the holder having to keep ETH in their wallet.
​

Application-tailored economics. Some applications need policies that aren't possible on a shared L2: protocol-enforced royalties (RARI Chain), KYC at the chain level (Kinto), or per-asset transfer restrictions (Plume). A shared L2 doesn't let an application enforce these primitives outside its own contract; a dedicated chain can write them into the rollup specification.
​

Faster product cycles. A dedicated chain lets a team ship a feature (new precompile, modified EVM behavior, custom MEV policy) without coordinating with the entire L2 ecosystem. The cost is taking on that maintenance burden permanently.
​

Dedicated rollups have non-trivial costs that the RaaS pitch sometimes glosses over.
​

Liquidity bootstrapping. A new chain starts with zero liquidity. Bridging in user funds, attracting market makers, and seeding lending protocols all cost time or capital. Manta Pacific spent its first year building incentive programs to draw liquidity from Ethereum mainnet; ApeChain bootstrapped through APE-denominated incentives. The dedicated-chain TVL number on L2BEAT is consistently smaller than the team's pre-launch projections.
​

Fragmentation. Every new rollup adds another chain users have to manage. Wallets, bridges, and indexers all need to update. From the user's perspective, rollup proliferation is a tax on attention. This is why interop products (Metalayer, the Optimism Superchain, Polygon AggLayer, Arbitrum Cross-Chain Messaging) keep getting funded.
​

Sequencer centralization. Most Caldera chains run a single sequencer operated by Caldera. That sequencer can censor, reorder, or front-run transactions, and a sequencer outage halts the chain. Decentralization roadmaps exist (the Espresso Sequencer integration, shared sequencing experiments through Astria) but production rollups in 2026 still mostly run centralized sequencers.
​

Settlement-layer dependency. A rollup's security ultimately reduces to its settlement layer. An Ethereum-settled OP Stack chain inherits Ethereum's reorg risk and finality time (12 minutes for hard finality through epoch finalization). A chain settling to a smaller L1 takes on whatever risk that L1 carries. Caldera publishes the settlement target for each operated chain; treasury teams should verify it before holding meaningful balances.
​

Data availability trust. Posting blobs to Ethereum gives the strongest security; alt-DA layers (Celestia, EigenDA, NEAR DA, Avail) introduce additional trust assumptions. The L2BEAT classification system distinguishes "rollups" (DA on L1) from "validiums" (DA off L1) for this reason.
​

The RaaS market in 2026 is post-novelty. Caldera and the other major providers compete for serious teams rather than crypto-curious experimentation. Three trends define the year.
​

Stack consolidation. The OP Stack and Arbitrum Orbit dominate new launches, with ZK Stack and Polygon CDK splitting the validity-proof segment. Polygon's CDK documentation and Optimism's Stack documentation both point to ongoing convergence on EVM equivalence and EIP-4844 blob usage as table stakes.
​

Interop primitives. Metalayer is Caldera's answer; the Optimism Superchain, Polygon AggLayer, and the Arbitrum Cross-Chain Messaging contract are the broader-ecosystem answers. The expectation in 2026 is that any new rollup needs an interop story on day one, not as a roadmap item.
​

Application-specific value capture. The chains that worked in 2025–2026 are the ones tied to a specific application or company (ApeChain, B3, Plume, Kinto). Generic "L2 for everything" launches without a built-in user base have struggled. RaaS providers now spend serious energy qualifying which teams should run a dedicated chain at all, since a chain that doesn't fill its blockspace ends up costing operators more in fees than the chain captures.
​

Caldera's investor update at the Series A round in 2024 framed RaaS as a 10-year cycle. The 2026 reading is that the cycle is real but compressed: most of the chains that should exist already do, and the next phase is depth (interop, decentralization, application-specific tuning) rather than width.
​

Eco is a stablecoin orchestration platform that routes USDC, USDT, USDS, and other regulated stablecoins across 15+ chains. Eco Routes integrates with the bridges native to each Caldera chain's underlying stack: Optimism's canonical bridge for OP Stack chains, Arbitrum's native bridge for Orbit chains, and Polygon's CDK bridge for CDK chains. For a stablecoin team building on or around a Caldera-deployed chain, Routes abstracts the bridge layer so the application receives a single intent (move X USDC to chain Y) and Eco's solver network handles the per-stack settlement. The integration matters because every Caldera chain represents a new bridge surface for any payments or treasury app, and orchestrators like Eco close that gap without each app team integrating against each chain's bridge directly.
​

Rollup-as-a-Service (RaaS) is a managed-infrastructure model where a provider operates the sequencer, batcher, prover, and monitoring stack for a Layer 2 or Layer 3 chain on the customer's behalf. The customer owns the chain configuration and economics; the provider handles uptime, batch posting, and operational SRE. Caldera, Conduit, Alchemy Rollups, and AltLayer are the four leading providers in 2026.
​

Caldera supports five rollup frameworks: OP Stack (Optimism), Arbitrum Orbit, ZK Stack (Matter Labs), Polygon CDK, and Eclipse SVM. Settlement layers can be Ethereum L1 or alt-DA layers like Celestia, EigenDA, NEAR DA, and Avail. The combination is configurable per chain, which is part of why operators with specific economics (gas token, throughput, DA cost) pick Caldera over more constrained providers.
​

Production Caldera chains include Manta Pacific (modular DeFi L2), Kinto (KYC-permissioned institutional L2), B3 (gaming on OP Stack), Plume (real-world assets), RARI Chain (NFTs with protocol royalties), ApeChain (Yuga Labs Orbit chain with APE gas), and Treasure (gaming with MAGIC). TVL across Caldera-operated chains is tracked on L2BEAT.
​

The Caldera Metalayer is an interoperability product that connects Caldera-deployed chains through a sequencer-level coordination service plus a messaging primitive. It lets a transaction on one Caldera chain trigger state changes on another in seconds rather than waiting for both to finalize on Ethereum. Metalayer covers the Caldera chain network specifically; cross-chain movement off Caldera still uses third-party bridges.
​

Both Caldera and Conduit operate OP Stack and Arbitrum Orbit chains. Caldera supports more stacks (ZK Stack, Polygon CDK, Eclipse SVM) and ships the Metalayer interop product. Conduit publishes a self-serve console with flat-fee pricing and operates large Optimism Superchain members. The choice often comes down to stack coverage, pricing model, and which provider's reference customers match the operator's application class.
​

Dedicated rollups give teams custom blockspace, custom gas tokens, and application-tailored economics, but cost teams in liquidity bootstrapping, user attention (one more chain to manage), centralized sequencer trust, and DA-layer trust if the chain uses alt-DA. The decision depends on whether the application generates enough volume or specificity to justify those costs, which most applications don't.
​