Skip to content
Blockworks

The Next Bottleneck: Rethinking Real-time Infrastructure | PMLS 2025 | Day 1 | Main Stage

Avery Ching, Saurabh Sharma

0:000:00

Key takeaways

  • Real-time Storage Gap: Current decentralized storage keeps data safe but locked away; the next wave of high-throughput, low-latency applications needs instant reads, not just durable writes.
  • Shelby Introduction: Jump Crypto and Aptos unveil Shelby, a “hot-storage” layer that combines Web 2 speed with Web3 ownership, offering CDN-grade performance, sub-second global reads, and protocol-level monetization.
  • Embedded Monetization Rail: Every data access can trigger on-chain payments or permissions, enabling pay-per-view video, token-gated AI datasets, and usage-based creator revenue without added middleware.
  • Performance by Pedigree: Shelby’s architecture draws on Aptos engineers’ experience scaling Facebook-class systems and Jump Crypto’s high-frequency trading stack, aiming for parallel reads, low variance, and production-grade reliability.
  • Decentralization without Vendor Risk: By moving hot data off closed clouds, projects remove a single point of failure and regain transparent pricing, access logs, and policy control.
  • Immediate Market Pull: Early pilots with Metaplex, Story, and DoubleZero reflect pent-up demand from NFT, AI, and content platforms that cannot meet user expectations on existing Web3 storage.

On Real-Time Data Bottleneck

Saurabh Sharma frames the sector’s core constraint: storing data is solved, but activating it at interactive speed is not. Slow retrieval cripples AI model hydration, live social feeds, and creator monetization. Closed cloud providers fill the gap yet introduce opaque costs and policy risk, leaving decentralization incomplete. The speakers argue that until the storage layer matches block-times, Web3 cannot support consumer-grade apps.

On Shelby’s Technical and Economic Design

Shelby sits between cold protocols and the application edge. Data is replicated across a performance-optimized network that offers parallel reads and programmable access logic embedded at the protocol layer. Because reads are cryptographically metered, developers can price them in real time, turning every fetch into an economic event. The team claims cost parity with mainstream object stores while eliminating egress penalties and lock-in.

On Partnership Dynamics

Avery Ching highlights how the collaboration began with Aptos engineers flying to Jump’s Chicago headquarters in January. Mutual respect around performance culture drove joint architecture reviews, merging Aptos’s Move-based execution and Jump’s ultra-low-latency networking. Both stress that the venture is incrementally built on prior research rather than a marketing-led launch.

On Use-Case Traction

Pilot users span chains and verticals. Metaplex explores instant NFT media delivery, AI teams prototype pay-per-access training sets, while gaming studios test multiplayer state streaming. All require sub-second reads, global distribution, and enforceable on-chain revenue splits—capabilities the founders assert are native to Shelby’s SDK and command-line tools now in early access.

Key actionable insights

  • Monitor adoption metrics from launch partners to gauge whether hot-storage economics truly outperform centralized clouds.
  • Track integration depth with major AI and content platforms; sustainable demand will validate Shelby’s on-chain metering model.
  • Watch for competing real-time storage primitives, as success here could redefine the minimum feature set for decentralized infrastructure.