In the fast-moving world of blockchain, there’s a difference between chasing trends and setting them. StarkWare’s philosophy is simple: build what is mathematically necessary, even if it isn’t easy and if the rest of the industry isn’t ready for it. Historically, many of the innovations we brought to market were initially dismissed. However, today, the concepts we pioneered, from ZK-STARKs to recursive proving, have become the gold standard for scaling.
As we examine our track record, it is obvious that we don’t believe in chasing trends. We set them. StarkWare’s work builds steadily and consistently towards the mission of bringing about the Integrity Web, our vision for a network that connects and empowers people. StarkWare envisions a world where individuals have true ownership over their digital actions, assets, and lives, all with the privacy people have come to expect in the real world.
A History of Firsts
Our journey started with the math of STARKs (2018). While STARKs already existed as a mathematical concept, no one had attempted to use them as a proof system. At a time when many in the industry were focused on using SNARKs for proving (as popularized by Zcash) and debating simple block-size increases, we were laying the groundwork for post-quantum-resistant, transparent, and ultra-scalable proofs.
In 2019, we introduced Validium and the first production-grade prover. While others were skeptical about the time it would take to prepare ZK-proofs for general-purpose scaling, we had already put them into production within a few short years, as demonstrated by StarkEx and, later, Starknet.
By 2020, Ethereum had reached a critical congestion point. Gas fees were skyrocketing, and for high-frequency applications like decentralized exchanges, the network had become virtually unusable for the average trader. While the rest of the industry was debating scaling roadmaps that were years away from reality, we launched StarkEx, proving that ZK Rollups weren’t just a theoretical dream, but a tool capable of processing millions of transactions at a fraction of the cost.
The validation for the massive scaling capabilities of StarkEx came when decentralized trading platform dYdX chose to build on StarkEx. Choosing this untested technology was a risky move, but dYdX recognized that zk-STARKs were the only way to scale their business while maintaining secure, high performance, Web2-like user experience. The risk paid off for dYdX on StarkEx, as it handled more volume than virtually any other platform in the space, all at a fraction of the cost. dYdX and other projects on StarkEx, including Sorare and Immutable, all became leaders in their respective sectors.
From Impossible to Industry Standard
In 2021, Starknet launched as the first general-purpose zk-Rollup. This was a pivotal evolution from StarkEx, an application-specific scaling tool perfect for high-performance use cases like perp trading or NFT minting, but restricted to specific business logics. Unlike StarkEx, Starknet provides a permissionless environment where developers can deploy any application, providing zk-scaling for all use cases.
StarkWare then focused its efforts on creating a zk-friendly programming language. Instead of forcing zk proofs into the constraints of the Ethereum Virtual Machine (EVM), StarkWare invented Cairo, a Turing-complete programming language, and a new way of thinking about computation. Cairo was the logical, secure, and efficient choice for a zk-based Rollup, operating a zkVM. Industry critics argued Cairo would be too difficult for developers and that we were not Ethereum-aligned by building a non-zkEVM. But today, we see the broader ecosystem, including Ethereum, moving toward the same high-performance VM models we championed years ago.
In subsequent years, StarkWare kept shipping milestones:
- 2022 – Recursive Proving & L3s:
- We introduced recursive proving, a scaling breakthrough that allows thousands of proofs to be wrapped into a single “meta-proof.” This creates an exponential scaling effect where the cost of verification on Ethereum stays nearly flat even as the number of transactions increases exponentially.
- We also introduced the concept of an L3 appchain, an application-specific layer built recursively over an L2, demonstrating that it is possible to stack layers of scale. This vision has become a reality with the launch of Privily, the first-ever L3 on Starknet, which creates a dedicated environment for private, user centric applications. By settling directly on Starknet, this L3 achieves hyper-scalability and ultra-low costs. The concept of L3s is now being adopted by nearly every major scaling roadmap.
- 2024 – Circle STARK & S-two: We pushed the boundaries of proving speed again, proving that the “limit” of ZK performance is much higher than anyone anticipated.
- Circle STARK: By utilizing the M31 field (a smaller prime field), S-two performs multiplications 125x faster than the traditional 252-bit fields used in the first generation of zk-Rollups.
- S-two: When benchmarked against top-performing zkVMs and precompiles, S-two is 10-30x faster,depending on the task.
- 2024 – L2 Staking: Starknet became the first major L2 rollup to implement a staking mechanism directly on L2. This was part of Starknet’s gradual transition to a proof-of-stake (PoS) protocol. By launching the first phases of L2 Staking, we empowered the community to secure the network. This move effectively ended the “centralized sequencer” era for Starknet.
2025 into 2026: Starknet Executes
In 2025, Starknet transitioned into a mature, production-grade scaling and privacy engine.
- Stack Optimization: The year was defined by a shift toward stack optimization, most notably with the Grinta (v0.14.0) upgrade and the
integration of the S-two prover. These developments achieved a 100x increase in efficiency over previous models,
reducing transaction latency to just 500ms. - Circuit Recursion: In addition, in 2026 we are moving from Cairo-based recursive proving to circuit-based recursive proving, unlocking efficiency gains vis-à-vis latency, hardware requirements, and costs, and opening the door to use cases including privacy, client-side proving, decentralization, and more.
- Decentralization: Decentralization reached a critical milestone as Starknet became a Stage 1 Rollup by implementing a decentralized sequencer architecture, the first of its kind in production. This was further bolstered by the launch of L2 Staking, which introduced a phased approach to network security and the addition of Bitcoin, alongside STRK, to staking, making Starknet the first, and only, network powered by a dual-token model. This model allows the network to be secured by both STRK and BTC, aligning Starknet’s economic security with the world’s most robust digital asset. In fact, in the first year, Starknet saw over 1 billion STRK staked (then over 23% of the circulating supply) and around 700 BTC staked.
- BTCFi: The emergence of BTCFi became a central pillar of the ecosystem in 2025. Starknet evolved into a premier destination for Bitcoin holders through a number of BTCFi initiatives. Users can engage in native BTC staking, utilize liquid staking derivatives like xyBTC through Endur, trade BTC through AVNU or Ekubo, lend or borrow using Vesu, and access multiple BTC yield strategies through the Earn portal. By bridging Bitcoin’s security with Starknet’s high-performance compute, the network has created a functional Bitcoin economy where users can generate yield and use BTC in DeFi.
- DevX: In February 2026, Starknet introduced Starkzap, a toolkit for builders and devs to build onchain consumer apps. The release of Starkzap v2 saw the integration of confidential transfers and payments into a developer-friendly TypeScript SDK.
Finally, Starknet is strongly focused on privacy as a core standard. The beginning of 2026 has already seen several major StarkWare privacy-related announcements:
- STRK20: A new protocol-level capability that brings confidential balances and private transfers to any ERC-20 token, including ETH, BTC, and stablecoins. It uses a single, unified privacy pool to ensure high anonymity sets while maintaining built-in compliance.
- strkBTC: strkBTC is a privacy-enabled, Bitcoin-backed asset issued on Starknet and redeemable one-to-one for native BTC. It combines “hard money” with yield generation and transaction privacy, allowing Bitcoiners to participate in DeFi without exposing their entire history. No other Bitcoin privacy solution offers this.
STRK20 and strkBTC are enabled by Starknet v0.14.2, an upgrade that provides in-protocol support for proof verification. V0.14.2 provides the native support needed to make our privacy solution seamless. Thanks to v0.14.2’s ability to verify S-two proofs, tokens can now support encrypted balances. Users can swap, stake, and send any ERC-20 token while keeping their financial footprint shielded. For Bitcoin holders, the upgrade facilitates private Bitcoin interactions on Starknet’s DeFi applications.
- Nightfall: Nightfall is a zero-knowledge privacy protocol developed by EY for the public domain and designed as a Layer 3 on top of Starknet. It functions like an appchain environment that enterprises can deploy for privacy-preserving transaction flows, and relies on Starknet for verification and settlement.
Starknet’s goal is to move beyond simple scalability. It aims to become the industry’s primary privacy engine, protecting users from both corporate surveillance and adversarial market behavior. And at its core, Starknet has all the requirements to become the ultimate privacy engine, including speed, succinct verifiability, ZK friendliness, cheap abundant computation, an easy path to post-quantum security, and excellent UX.
Our roadmap is not about chasing Crypto Twitter hype. It is the culmination of the infrastructure we’ve built since 2018.
StarkWare Isn’t Following Trends, We’re Building Them
At StarkWare, we’ve learned that being right early feels a lot like being wrong, until the rest of the world catches up.
When you build on Starknet or partner with StarkWare, you aren’t betting on a project that is figuring things out along the way. You are partnering with a team that invented the primitives the rest of the industry is now adopting. We’ve been thinking about privacy, post-quantum resistance, and proving performance for over six years, long before they became industry buzzwords.
The future of blockchain isn’t being invented today; it’s being scaled by the people who have been building its foundation for years. The math is proven. The track record is clear. And we’re just getting started.
Are you a builder interested in building on Starknet? Start here by checking out all that we have to offer.
