When IBM Research contributed Fabric-X last year as an implementation of Hyperledger Fabric for next gen digital assets, the goal was to validate that Hyperledger Fabric’s architecture and design principles can be extended and pushed to 100K+ transactions per second while keeping end-to-end latency low and without giving up decentralization. Based on that success, we are now building out Fabric-X to be a new iteration of Hyperledger Fabric for next gen digital assets
Fabric-X carries on many of the features at the core Hyperledger Fabric. It inherits strong identity management and access control. Every participant operates under an identity issued by an authorized Membership Service Provider (MSP), and that identity gates who can submit transactions, execute smart contracts, participate in consensus, and replicate the ledger.
Fabric-X also follows the execute–order–validate model. A client submits a proposal for endorsement; a set of endorsers execute the smart contract and produce a signed simulation result in the form of a read set (keys observed) and a write set (proposed updates). Those signed results flow through ordering and, during commit, are validated for (1) conflicts (the read set has not gone stale) and (2) access policy compliance (the collected signatures satisfy the write-access policy for the affected state).
This design becomes particularly powerful once you look at governance boundaries. The ledger is partitioned into namespaces, and each namespace is associated with a smart contract. The smart contract’s stakeholders define the namespace’s endorsement policy (i.e., which endorsers must sign for updates to be accepted). In practice, this gives the stakeholders fine-grained control over who can update the state, and the ability to both horizontally and vertically scale execution by provisioning endorsers to match the desired throughput.
But Fabric-X is not just a reimplementation of Hyperledger Fabric. The goal is to address the requirements that show up in regulated financial-market infrastructures: high throughput under strict latency constraints, privacy boundaries across participants, and operational requirements driven by compliance. In our experience, classic Hyperledger Fabric deployments run into three recurring limitations.
Taken together, these changes drove last year’s headline result: Fabric-X sustained 100K+ tps at <2s end-to-end latency in our evaluations.
In 2026, we shift from proving peak throughput to building what’s required to run Fabric-X in production: predictable storage growth, operability at scale, enterprise-grade security, and ease of deployment. The roadmap below translates these priorities into concrete deliverables.
Endorser service (execution + endorsement): simplify setup and operations; harden key protection to industry standards (e.g., FIPS); improve Developer Experience with a dedicated client SDK; broaden adoption via EVM compatibility; and speed up communication with gRPC.
Ordering service (ARMA consensus): support dynamic membership changes with minimal disruption; reduce transaction size with a more efficient MSP/certificate registry; and add storage lifecycle management—snapshotting, pruning, backups, and long-term archiving—to keep storage growth predictable at scale.
Committer service (validation + state): add granular, policy-based access control for all ledger interactions; improve operability with automated consistency checks and divergence detection across committers; and deliver a store service that scales with ledger growth.
Deployment & tools: ship production-ready Fabric-X releases with end-to-end deployment support, plus the documentation and tutorials that developer teams need to adopt it.
The Fabric-X Endorser plays a central role in how transactions are validated and approved before final settlement. With its shift away from traditional chaincode-based endorsement toward a more flexible, application-level model, it opens up new ways to build, scale, and integrate blockchain solutions.
In 2026, our focus is on making the Endorser more powerful, easier to operate, and ready for enterprise-scale deployments.
We’re expanding support for MSP-based endorsement policies, giving organizations more flexibility in how they define approval rules. For additional data protection, we are adopting a robust access control layer provided by the committer service. At the same time, we’re improving cryptographic capabilities with support for Hardware Security Modules (HSMs), ensuring compatibility with enterprise environments that require strong key management and compliance with industry standards such as FIPS.
This new capability will enable EVM‑based chaincode i.e., Solidity smart contracts to run on Fabric-X while preserving permissioned governance and enterprise controls. More details available in this blog.
We’re making the Endorser easier to configure and deploy, with clearer guidelines for developers and operators. Behind the scenes, we’re investing heavily in testing and stability to ensure consistent performance in production environments. This includes expanding our automated test coverage with unit tests, multi-node integration tests, and stress tests for high-throughput scenarios.
We’re packaging core functionality into a dedicated Client SDK, making it easier to build lightweight applications without pulling in the full stack. This will simplify development and reduce complexity for teams adopting Fabric-X.
To support high-throughput use cases, we’re introducing gRPC alongside existing peer-to-peer communication options. Combined with improvements to connection handling and reliability, this will make interactions between components faster and more robust. Moreover, we are integrating the scalable ledger store service into the endorser to support enterprise-grade data loads.
We’re also exploring new directions, such as improved compatibility with existing Fabric chaincode as part of the LFDT Mentorship program. These efforts aim to make Fabric-X more accessible while supporting a broader range of use cases.
The 2026 roadmap for the Fabric‑x ordering service introduces several enhancements aimed at improving flexibility, operational readiness, and long‑term scalability.
The Fabric‑X ordering service is built on Arma, a scalable Byzantine Fault Tolerant protocol designed to achieve high transaction processing rates. Rather than relying on a cluster of monolithic servers, Arma distributes the roles of each ordering node to a structured collection of components known as a party.
A party consists of four server roles:
We have demonstrated that the Fabric-X ordering service can achieve over 400K TPS with a deployment of 4 parties and 4 shards, and over 250K TPS with a deployment of 7 parties and 2 shards. Transaction size was 300B, typical for our asset exchange use case. Those systems were deployed over 4 different AWS data centers in the US. (more details can be found in our recently published paper at SIGMOD conference).
Over the coming year, Fabric‑X will introduce a streamlined dynamic reconfiguration process that will allow parties to update certificates, endpoints, and membership with minimal operational impact. Identity management will be improved through an enhanced MSP design that will reduce transaction sizes by relying on a shared certificate registry rather than embedding full certificates in every payload. The platform will also undergo significant hardening and productization work, with expanded logging, richer metrics, improved integration testing, and broader system‑level robustness to support production‑grade operations. In addition, Fabric‑X will add advanced storage‑management capabilities including snapshotting, pruning, backup options, and long‑term archival storage that will enable sustainable scaling while preserving strong auditability guarantees.
Fabric‑X introduces a streamlined reconfiguration process that allows administrators to update certificates, endpoints, as well as party membership with minimal disruption. Configuration changes are submitted through standard Fabric APIs, validated within the party, and committed through the BFT consensus cluster. Components independently retrieve and apply the new configuration, enabling continuous operation during routine and security‑related updates.
The ordering service is strengthened with an enhanced MSP which helps reduce transaction size. Today, each transaction includes the full submitting & endorsing client(s) certificates, which creates unnecessary overhead at scale. By relying on a “well known” certificate registry instead, parties can validate identities without embedding certificates in every transaction. This results in smaller payloads and more efficient use of network and storage resources.
The first half of 2026 focuses heavily on operational readiness. Enhancements include improved logging and runtime log‑level control, expanded metrics, integration testing for reconfiguration flows, and broader system‑level hardening. These changes aim to ensure robustness, predictable behavior under load, and smoother operations in production environments.
As the ordering service scales toward high transaction processing rates, storage demands grow rapidly. In the second half of 2026, the Fabric‑X ordering service will introduce snapshotting, pruning, backup capabilities, and options for long‑term cold storage. These features support sustainable retention strategies while preserving the auditability required in enterprise deployments.
The 2026 roadmap for the Fabric‑X Committer Service focuses on transforming how organizations validate, store, and query ledger data in high-throughput environments. While the ordering service handles the "how" and "when" of transaction sequencing, the Committer Service is the backbone of "what" the ledger becomes, ensuring that every transaction is accurately reflected in the world state and remains accessible at scale.
Traditionally, the committer has been a monolithic component within a peer, but in Fabric-X, we have decoupled it into a high-performance, distributed architecture. By separating the validation logic from the storage layer and the query interface, we allow each organization to scale its infrastructure independently based on its specific workload. Our current benchmarks show that this decoupled architecture can keep pace with the Arma-based ordering service, processing hundreds of thousands of transactions per second without compromising data integrity.
The key advancements for the Committer Service this year focus on security-first access, automated consensus auditing, and cloud-native storage elasticity. To achieve a zero-trust architecture, we are implementing a granular access layer for all ledger interactions, moving away from broad "all-or-nothing" database permissions to ensure that sensitive state data is only surfaced to authorized services. To maintain high-fidelity data across distributed nodes, we are introducing automated consensus auditing, which uses background state-hash verification to proactively detect and diagnose "silent" ledger divergence caused by non-deterministic chaincode or infrastructure faults. Furthermore, by transitioning to cloud-native storage elasticity, the service will decouple from static disk-bound limitations, allowing the storage mesh to scale horizontally to meet the petabyte-scale demands of enterprise workloads without bottlenecking transaction throughput or query latency.
To strengthen the security posture of the Fabric-X ecosystem, we are implementing a robust access control layer for the query service and its associated sidecar. By leveraging the native Fabric policy language, we are ensuring that every API call whether it’s querying the world state, retrieving blocks, or subscribing to the notification service is governed by a strict ACL. This ensures that only organizations with explicit read permissions can access sensitive ledger data, providing a zero-trust environment for cross-organizational data sharing.
Maintaining a "single source of truth" is critical, especially since committers across different organizations operate independently during the commit process. We are introducing a periodic state-hash consensus mechanism to verify data consistency across the network. If a mismatch is detected, the system won't just flag the error; it will perform a deep-dive analysis to identify the specific divergent keys and affected transactions. This automated "forensic" capability allows operators to quickly pinpoint the exact reason for divergence and maintain ledger integrity.
As data volumes grow toward the petabyte scale, traditional sidecar block storage can become a performance bottleneck. To address this, we are decoupling the ledger into a dedicated Scalable Ledger Store Service. This architecture supports horizontal scaling, allowing the network to handle massive query volumes and notification demands without degradation. By moving to this cloud-native storage model, Fabric-X will be equipped to manage enterprise-grade data loads while ensuring high availability and low-latency access for all participants.
Fabric-X V1.0 — The production‑ready solution for distributed environments will be available by April 17. The toolkit will include all required infrastructure components and SDKs needed to run Fabric‑X and to build applications on top of it. This version will have full ansible support and candidate release for Kubernetes. As part of the release process, we will also conduct comprehensive network‑level penetration testing.
Q2 will focus on delivering comprehensive user documentation, tutorials, and deployment toolkits for Kubernetes and OpenShift. Topics we plan to bring forward for community discussions and collaborations include:
Q3 will focus primarily on finalizing enterprise‑grade features and advancing the topics introduced in Q2. Our preliminary plan includes:
By the end of Q4 2026, we plan to release the long‑term support (LTS) version of Fabric‑X.
This quarter will be dedicated to finalizing the features delivered in Q3, hardening the codebase, storage management, and conducting extensive stability and performance testing.
We welcome everyone who is interested in Fabric-X to join the community and help us with our plans for the year. You are welcome to join any of the upcoming Fabric community calls and also join the discussions on the Fabric channels on LFDT’s Discord.