We like the buyer's side of this: a verifiable dataset ID gives you a concrete reference to check later, even after the provider's feed has moved on. Perceptron putting that into its enterprise data products is a practical use of Walrus. Especially when decisions need an audit trail. Great work by both teams 👏 As Walrus operators, we're glad to help run the infrastructure behind use cases like this 🦭

As one of Espresso’s validators, we really like where this is going. One detail in the benchmarks deserves more attention: 50 MB blocks deliver ~98% of peak throughput with roughly 25% lower finality latency than 70 MB blocks. That’s a useful operating point. There’s very little extra throughput to gain by pushing block sizes further, while the latency cost grows considerably. The test uses 100 validators with emulated global network delays, running production consensus code. At peak load, it reaches 181 MB/s with 778 ms median finality. Execution stays with the chains submitting the data, so the “1 million TPS” figure represents aggregate capacity across those chains. For us, the interesting result is how much throughput is available while keeping consensus latency low. That leaves room to optimize for faster finality without giving up much capacity. Good engineering ☕