Shelby is the product of two converging engineering lineages. Aptos Labs engineers draw on their time at Meta on projects like helping to scale Instagram, hyperscale distributed data infrastructure built for consumer-grade reliability. Jump Crypto brings the complementary half: petabyte-scale real-time systems from high-frequency trading.
The timeline at a glance
| Date | Event | What happened |
|---|---|---|
| 2011, 2021 | Meta infrastructure DNA | Future Aptos engineers ran hyperscale distributed systems at Meta, Spark, Giraph, Hive/Hadoop, powering Instagram and Facebook at billions-of-users scale |
| 2019, Dec 2021 | Libra / Diem | Meta's 300+ person blockchain team built Move, BFT consensus, and parallel execution before regulators killed the project, clearing the codebase to go external |
| Mar, Oct 2022 | Aptos Labs formed with Jump already in | The $200M seed round was co-led by FTX Ventures and Jump Crypto, placing Jump on the cap table from day zero |
| 2023, 2024 | Jump builds the other half | Jump shipped Firedancer and DoubleZero in parallel, converging on a Network + Data + Compute stack that identified storage as the next gap |
| 2024 | Storage gap becomes undeniable | As Aptos pushed into AI, DePIN, and streaming, existing cold decentralized storage (Arweave, Filecoin) proved unable to serve CDN-grade workloads |
| Jun 24, 2025 | Shelby announced | Aptos Labs and Jump Crypto unveiled Shelby, a chain-agnostic hot-storage network with fiber-optic backbone and sub-second reads |
2011, 2021, Meta infrastructure DNA
Avery Ching spent over a decade at Meta as the overall tech lead for batch processing, Spark, Apache Giraph graph processing, Hive/Hadoop/Corona, distributed scheduling, systems that scaled to hundreds of thousands of machines and served billions of consumers. Sital Kedia was Tech Lead at Facebook from 2015-2022 on Spark at 60TB+ production workloads before becoming a founding engineer at Aptos. This is the cohort that learned how to keep feeds online under load, not in a supercomputing lab, but in the "don't take Instagram offline on Super Bowl Sunday" sense.
2019, Dec 2021, Libra / Diem
The 300+ person Meta blockchain team built Move, BFT consensus, and parallel execution primitives. Regulators squashed it. Pre-Thanksgiving 2021, Ching and Shaikh walked out of a Menlo Park meeting with implicit green light to take the open-source Diem codebase external.
Mar, Oct 2022, Aptos Labs formed with Jump already in
The $200M seed round was co-led by FTX Ventures and Jump Crypto, followed by a16z's $150M Series A. Jump was on the cap table from day zero, this is the structural reason the eventual Shelby collaboration was possible, not a cold outreach story.
2023, 2024, Jump's parallel stack builds the other half
Jump didn't just invest, they were building Firedancer (high-performance Solana validator client) and then DoubleZero in parallel. Through their work on Firedancer, Jump became convinced that a high-performance network was also required for current blockchains, which led to the launch of DoubleZero, a high-performance decentralized network for high-bandwidth applications. Shelby was designed to leverage that foundation to deliver high-performance storage. The mental model they converged on: Network + Data + Compute = Decentralized Cloud.
2024, The storage gap becomes undeniable
As Aptos pushed into AI, DePIN, and consumer content use cases, the team ran into the same wall every serious Web3 consumer app hits: 82.5% of global internet traffic is video and live feeds, and traditional decentralized storage was cold, slow, and static, great for archiving but not for streaming. You cannot run an Instagram-like product on Arweave or Filecoin. The Aptos team's Meta background made this viscerally obvious, they'd already built the thing that needed to exist, just inside Menlo Park.
Jun 24, 2025, Shelby announced
Jump processes hundreds of petabytes of market data that feeds their tiered archive, and Shelby applies that high-performance storage expertise to decentralized storage. Shelby operates on a global mesh of high-performance nodes connected by a dedicated fiber-optic backbone, with edge caching enabling sub-second data access at scale. Chain-agnostic from day one, Aptos, Ethereum, Solana. Devnet Q4 2025, public testnet after.
Hot data, live monetization, and the path to the largest internet market
Hot streaming data changes the economic unit of the internet. Cold storage forces a pricing model where you charge once for access and serve the file rarely, the economics of an archive, not a marketplace. Every file is a liability on the balance sheet until someone pays to retrieve it, and the retrieval itself is slow enough that nothing real-time can live there. Hot storage inverts this. When content is served at CDN speed with sub-second reads and programmable access logic, every file becomes a live instrument. It can be priced by the second it's watched, the paragraph it's read, the query it answers, or the trade it influences. The asset isn't the file, the asset is the moment of consumption, and the moment is what gets billed.
That's the argument for Whop as the largest internet market rather than the largest creator platform. The largest internet markets in history, Amazon, Alibaba, Shopify, all scaled by owning the rail that everything else runs on, then letting sellers and buyers discover each other on top of it. The rail here is hot, verifiable, micropayment-gated content delivery, and it works across every vertical Whop already serves. Trading is the beachhead because it's the vertical where verification has the highest dollar-per-proof value, but once the rail exists, the same primitives monetize everything else. The ex-Meta Instagram cohort built the distribution infrastructure for billions of users at the last internet-scale platform. Plugging that infrastructure into onchain monetization rails is how you build the next one.
What Whop builds on top
Shelby is the pipe. Whop is the plug. Content Rewards, Whop Terminal, Whop Explorer, and Feed are four monetization shapes running on the same underlying stack, Shelby serving verifiable content at CDN speed, PropWhop resolving attribution, and CA/ACE primitives handling payment and privacy. Content Rewards is the labor market, Terminal is the institutional surface, Explorer is the trust graph, and Feed is the publishing rail. Together they're what a consumer-grade Meta stack looks like rebuilt with onchain monetization primitives from the start.