If you have also been following the cross-chain space, the new architecture recently announced by zkLink should catch your attention.
To be honest, I think this might be the most worthy technical route to closely follow in the cross-chain field right now.
Rather than saying zkLink is creating a 'cross-chain protocol', I would prefer to understand it as a foundational reflection and reconstruction of multi-chain DeFi.
What it truly wants to solve are a few structural problems that we have long been unable to bypass:
Information islands between chains, the fragmentation of transaction processes, and that eternally fragmented user experience.
In May, they updated their roadmap, and there was a phrase that left a deep impression on me:
“We aim to build a peer-to-peer cross-chain trading network led by AI agents.”
This sounds very much like a futuristic vision, but looking deeper, we find their design has already been implemented quite concretely:
Completely permissionless
Zero protocol fees
High-frequency low-latency execution mechanism
AI agents responsible for trade matching, negotiation, and cross-chain collaboration execution
The key point is: this is not UI-level optimization, not frontend packaging, and certainly not another cross-chain aggregator.
This is a reconstruction at the protocol layer.
From the timeline perspective, their rhythm arrangement is also quite restrained:
2025: Focus on underlying technology research and development, including white paper release, PoC testing, open-sourcing communication protocols, and integration with mainstream public chains
2026: Launch Alpha/Beta testing, build node verification mechanisms and developer ecosystems
Technically, the Nova mainnet has already launched—this is an L2 aggregation platform that supports asset circulation between multiple Layer 2s and is the infrastructure for future AI cross-chain networks.
In addition, the capital lineup supporting it is also considerable:
Coinbase Ventures, OKX Ventures, Crypto.com, Republic Crypto, SIG, Arrington Capital, BigBrain Holdings, etc.
Their token $ZKL has already launched on platforms like Bybit, Gate, Bitget, Hashkey Global, Coinone, etc., and initial liquidity is also in place.
But for me, what really makes zkLink worth long-term observation is its strategic choices:
It is not stacking UI modules, nor attempting to create a 'better experience bridge', but simply choosing to completely rebuild the protocol layer.
Directly constructing an AI-driven, fully chain-scalable trading execution framework.
If this framework can run smoothly, it may become the public infrastructure for multi-chain asset interactions.
And not just the next slightly faster bridge or a more stable DEX.
To outsiders, it might seem like there's 'no movement';
But for those who have long focused on infrastructure evolution, this is a form of 'structural innovation': slow in pace, but once connected, platform capabilities will emerge.
My mid-term observations will focus on several aspects:
Actual trading execution effects of AI agents
The speed of protocol module implementation (whether it can expand to contracts, Launchpad, etc.)
Depth of integration with various public chains and technical stability
Changes in TVL and real usage data
This path is not fast, but deep enough.
Amidst the various hot market stories, projects like this remind me: truly worthwhile endeavors often require patience to see them bloom.
