As Solana continues to thrive—demonstrating remarkable growth in user activity and on-chain economic value—it remains evident that the base layer, while currently sufficient, may face limitations as demand continues to scale. The question is not whether Solana can handle today’s bandwidth, but whether it can meet the demands of a future where internet-native capital markets rival the scale of institutions like NASDAQ. In anticipation of this future, a range of solutions have emerged—from Firedancer and network extensions, to Layer 2s and Solana forks—all exploring different pathways to unlock the next frontier of performance and scalability. Today we are proud to announce our investment into Magic Block as we feel it represents the most elegant and Solana-native solution to scale without compromising base layer composability. Unlike traditional L2s that require asset bridging and suffer from state fragmentation, MagicBlock introduces Ephemeral Rollups (ERs)—just-in-time, SVM-based runtimes that offer horizontal scaling on demand while maintaining seamless interoperability with Solana’s L1. Why is this important? Base layer composability ensures seamless access to the rich liquidity and application ecosystem that defines Solana’s user experience. It means developers and users can tap into protocols like Jupiter for routing, Kamino for automated yield strategies, Meteora for liquidity provisioning, Raydium for trading, and Jito for liquid staking—all without leaving the native environment. This uninterrupted connectivity is what gives Solana its signature fluidity, enabling applications to interoperate with deep liquidity and shared infrastructure in real time.
ERs in MagicBlock are specialized, transient execution environments designed to enhance the scalability and performance of decentralized applications on the Solana blockchain. Leveraging Solana's account-based architecture and parallel execution capabilities, ERs allow developers to temporarily delegate specific state accounts to a dedicated auxiliary layer. This delegation enables a configurable runtime that operates independently, facilitating ultra-low latency (as low as 10 milliseconds) and high-throughput transaction processing without compromising the composability and integrity of the base layer.
The process begins with state delegation, where developers lock one or multiple accounts to an ER via the Delegation Program, specifying parameters such as lifetime and update frequency. A sequencer then temporarily modifies the state within this rollup. Importantly, even as accounts are delegated, transactions on the base layer can still read their state, ensuring compatibility, while non-delegated accounts remain fully modifiable and unaffected.
ERs offer several customizable features, including gasless transactions, faster block times, and integrated scheduling mechanisms akin to Clockwork for automated transaction execution. These capabilities make ERs particularly suited for real-time applications like gaming, finance, AI agents, and DePIN.
Throughout their operation, ERs maintain seamless integration with the Solana ecosystem. Programs and assets reside directly on the base layer, ensuring that any improvements or advancements at the base layer are immediately available without the need for modifications or redeployments. Transactions are dynamically routed to the appropriate execution layer—base or ephemeral—via specialized RPC providers, optimizing execution and preserving a seamless user experience.
Upon completion of the specified operations, the ephemeral state is committed back to the base layer, including new state and relevant pointers. This state is finalized using a fraud-proof mechanism to ensure integrity. Finally, accounts are undelegated via the Delegation Program, returning control to the original owner. This approach ensures that applications can scale horizontally and handle millions of transactions per second while maintaining the composability and security inherent to Solana's architecture.
MagicBlock’s technology is profoundly extensible. While the team initially targeted fully on-chain gaming—one of the few domains demanding low latency, high throughput, and composability—the implications stretch far beyond. We see MagicBlock as a foundational primitive with the potential to transform execution across DeFi (orderbooks, MEV auctions), AI agents, payments, DePIN, NFT mints, airdrops, and more. The total addressable market is massive: these verticals already command billions in cumulative volume, and all share a common constraint—scalability without sacrificing composability. MagicBlock’s model unlocks that.
We foresee near-term adoption in:
DeFi: CLOB acceleration, MEV protection via batch auctions, flash-mint-style primitives.
Payments and Agent Systems: AI agents executing tasks within gasless, isolated but composable runtime environments.
DePIN and Attribution Markets: Campaigns and sensors operating in ERs with L1-settled rewards.
The unifying theme is custom performance without giving up access to Solana’s ecosystem.
The founding team behind MagicBlock, Andrea and Gabriele, began their journey initially focused on building fully on-chain game experiences. However, it was through this work that they uncovered a far deeper insight: the same execution engine designed to support immersive, real-time gaming could serve as a foundational primitive for an entirely new class of high-performance, composable applications across a variety of verticals. Their partnership with Flash Trade—a real-time perpetual DEX offering instant execution—is a compelling example of this broader vision in action, demonstrating how Magic Block's architecture can unlock sub-10ms latency for DeFi protocols without sacrificing base-layer composability. Andrea and Gabriele are representative of founders we want to back - high-velocity team that combines deep technical expertise with relentless execution. We are proud to work together with them on this journey.
We invested in MagicBlock because it offers the best path to scale Solana without breaking its most powerful feature: composability. In a world where every L2 isolates execution, MagicBlock preserves the economic density, developer synergy, and network effects of the base layer—while extending it with lightning-fast, gasless, modular environments. It’s not just a new scaling layer; it’s a new primitive for programmable blockspace.
MagicBlock isn’t just scaling Solana—it’s extending what’s possible on Solana.
