Is execution now more about routing than strategy?

Lately I’ve been spending a lot of time thinking about how fragmented transaction flow has become across high-performance chains.

If you’re building anything latency-sensitive (MEV, arb, liquidation bots, even just aggressive trading infra), you probably already feel this:

• Public RPC is too slow / inconsistent
• Private relays are opaque and fragmented
• Direct validator relationships don’t scale cleanly
• Mempool visibility is partial at best

So everyone ends up duct-taping their own setup:
multiple RPCs, custom routing logic, some relay integrations, maybe a few validator connections if you’re deep enough.

It works… until it doesn’t.

You start seeing weird behavior:

• Transactions landing inconsistently across similar conditions
• Same payload performing differently depending on route
• Latency variance killing otherwise profitable strategies
• No clear attribution of why something failed or got outcompeted

At some point it stops being about strategy and starts being about distribution.

Feels like we’re heading toward a world where:

• Transaction routing becomes a first-class layer (not just infra glue)
• Builders care less about where they send from, more about how intelligently it’s routed
• “Best execution” starts to include path selection across relays/validators, not just price

Curious how others here are approaching this right now:

• Are you running your own routing logic or relying on a single path?
• How are you thinking about redundancy vs latency tradeoffs?
• Anyone experimenting with dynamic routing based on slot/leader conditions?
• Or is everyone just quietly building this in-house and not talking about it 🙂

Feels like an area where a lot is happening, but very little is openly discussed.