Blockchain makes it possible for people to control their assets and transact directly with others on a trustless, secure, and censorship-resistant network. That ideal has the potential to transform entire industries, and it can only be achieved on blockchains that are truly decentralized. Blockchains dominated by a single party or conglomerate don’t really have an edge over traditional, centralized networks.
With the goal of bringing blockchain’s promise to everyone, Starknet is on the path to becoming the first fully decentralized Layer 2 (L2) that massively scales Ethereum. The recent launch of the first phase of staking on Starknet was a leap toward greater decentralization. This year, efforts to enhance Starknet’s decentralization will ramp up on top of continued performance optimizations.
Decentralized Starknet in 2025
Ethereum was the first blockchain to enable general-purpose computing. It earned its reputation for staying true to the first principles of blockchain: decentralization and security. As a Layer on top of—and thus an extension of—Ethereum, Starknet must also be decentralized and secure.
The terms “decentralization” and “security” are often used interchangeably. While it’s true that decentralization enhances security, we distinguish between the two in this post. That’s because, unlike other L2 rollups, Starknet has had its ironclad security technology—STARK proofs—in place from the beginning. There has never been a state update approved on Starknet that wasn’t validated by a proof. That means it’s virtually impossible for an invalid transaction to be processed on Starknet. This post will zero in specifically on Starknet’s decentralization journey, and the ways in which it will progress in 2025.
Starknet’s road to decentralization runs on three main lanes:
- Staking: Building economic security on Starknet toward the decentralized operation of its Proof-of-Stake (PoS) consensus mechanism
- Operation: The decentralization of Starknet’s operation
- Governance: The independence of the Starknet Security Council
Let’s dive into Starknet’s progress in each of these lanes, and what’s coming up.
Staking
Crucial to the decentralized operation of Starknet is the acquisition of economic security on the network, a process that is already under way. At a high level, here’s what this means:
Starknet uses Proof-of-Stake (PoS) as its mechanism for preventing Sybil attacks, in which a single entity or group of entities creates multiple validators to gain disproportionate influence over the network. PoS prevents these attacks by requiring validators to stake tokens. This ensures influence over the network is proportional to the economic cost of acquiring and risking tokens, rather than the number of validators created.
On Layer 1 PoS protocols, staking starts from genesis, and the consensus layer’s security generally improves with its economic value. For rollups, such as Starknet, the story is different.
Rollups inherit the security of their settlement layer and accrue economic value even if they don’t decentralize their operation. This dynamic provides an advantage to rollups that are on the path to decentralizing their protocol, such as Starknet. That’s because it opens the door to accumulating high amounts and a wide distribution of stake before the decentralization of the protocol. This sets decentralized, public Starknet up in a way that prevents the disproportionate influence of a select few on the network, fostering its stability and a wide distribution of validators.
As such, Starknet is first gathering economic security through staking, and only afterward allowing for decentralized operation of the protocol.
To this end, Starknet staking began as an application, decoupled from consensus—Phase 1 of Starknet staking (Staking v1) went live on November 26, 2024, and has been accumulating economic security ever since. To date, more than 170 million in STRK is being staked by 63,000 delegators and 106 validators.
Staking v2 will enter mainnet in Q2 of 2025, coupling staking rewards to stakers attesting blocks. Staking v3 will enter mainnet in Q4 of 2025, coupling staking rewards to block validation. Finally, Staking v4 will make validators fully responsible for maintaining and securing the network by producing, attesting, and proving blocks.
Each stage of staking contributes to building economic security and opens the door to the decentralized operation of the Starknet protocol.
Operation
In addition to economic security, the decentralization of Starknet’s operation requires a fully functional open-source stack with end-to-end functionality. Enter the SN Stack, which has just recently been made available and enables anyone to use the parts that power Starknet.
The present architecture of public Starknet relies on the open-source Stone prover and a closed-source sequencer, which are complemented by the open-source Madara and Katana sequencers. Throughout 2025, public Starknet will migrate to a newer stack, founded on the next-gen Apollo sequencer and the Stwo prover, both of which will be open-source. Stwo will make proving on Starknet much more efficient.
The next step toward the decentralized operation of Starknet will be the decentralization of the consensus layer. This means validators will vote on each Starknet block, and only a sufficient quorum of such votes will finalize a new block. Decentralized consensus is planned to go live on mainnet at the end of 2025. Here is the game plan in broad strokes:
- Starknet v0.14.0: Launch of the distributed sequencer architecture comprised of an internal network of 3f+1 nodes running consensus and taking turns building and proposing blocks
- Production release of the next-gen-sequencer for anyone to run
- Deployment on Testnet(s)
- Deployment on Mainnet
We have omitted subtler milestones pertaining to client diversity, which is crucial for a robust network. We’ll just say there are three additional full nodes that are presently integrating consensus and block proposal functionality: Juno, Pathfinder, and Madara (which needs only consensus).
Governance
Given a fixed protocol, there are “meta” matters pertaining to protocol changes. Governance of such matters is a difficult problem without any canonical solutions. For L1 networks, “the protocol” is practically defined by the majority of nodes in the network: If most nodes collaboratively modify their clients, the network remains the same while its protocol changes. Hence, L1 networks can have “informal” governance—nothing is set in stone.
Rollups, on the other hand, have core contracts on their settlement layers. Even if all L2 clients were modified, the core contract itself can only change through action on the settlement layer. Hence, rollup governance requires explicit permissioning for modifying L1 core contracts.
The Starknet Security Council is a governance mechanism that decentralizes control over core contracts (both on L1 and L2) pertaining to state updates, staking, and STRK minting (which is managed by smart contracts, unlike Ethereum, where rewards are implemented at the client level).
Moreover, the Security Council’s capacity to perform L1 state updates allows it to bypass consensus in case of censorship of L1/L2 messages. Thus, the Security Council facilitates censorship resistance in addition to its governance functions.
A Security Council is a first step toward the decentralized governance of a decentralized Starknet. However, the challenge is broader than formal control core contracts. The governance question has social and technical aspects that require much thought. Improving the decentralization of the mechanism is a core goal of decentralized Starknet.
Other features
While big decentralization moves are set to be made on Starknet this year, other features that aren’t decentralization-related, such as additional fee reductions and UX/DevX improvements, will also be launched.
v0.13.4: This version upgrade introduces:
- Stateful compression: Will reduce fees
- L2 gas price: Introduces fixed price, denominated in fri, for a single unit of L2 gas until fee market in v0.14.4; it decouples L2 computation fees from the L1 gas market
- Try/catch for function call failures
- Cairo-native (Sierra → LLVM): Will boost performance, and potentially TPS
v0.14.0: In addition to the distributed sequencer and potentially Stwo integration, this version will introduce:
- 2-second blocks, mempool, & fee market: Better UX, DevX
For a deeper dive into upcoming features, read the Starknet v0.13.4 pre-release notes, with more to come on v0.14.0.
Conclusion
The journey toward decentralized, scalable blockchain is a monumental challenge, but it must be achieved if blockchain is to achieve its full potential. Starknet stands at the forefront of this mission, proving that scalability and decentralization can coexist without compromise. By advancing staking, open-sourcing its stack, and progressing toward decentralized governance, Starknet is not only shaping its own future but also setting a benchmark for what L2s can achieve.
Stay tuned for updates about Starknet’s decentralization journey on X.
