BlockDeliver — Decentralized CDN Marketplace

The problem

Every CDN provider covers a patch of the world well and the rest poorly — so whichever provider a website picks, some of its users are stuck fetching content from another continent. Meanwhile, enormous amounts of bandwidth sit idle: providers' off-peak capacity, ISPs' spare links, even individual users' unused connections. The two problems are each other's solution — if you can build the market that connects them.

The product — a marketplace, not another CDN

BlockDeliver didn't build servers; it stitched existing ones into a single network. CDN providers, ISPs and users listed spare bandwidth; websites bought coverage from the combined pool; token-based incentives (the blockchain part, used for the payments-and-trust problem it actually solves) rewarded whoever served the traffic. One provider covers a few countries well — a marketplace of many covers the world.

2The catch — whichever CDN a website picks, someone's left unhappy

CDN provider A · 10 servershappyunhappy — server too far⦿user A⦿user B

A website on CDN A: user A is served nearby — user B in Australia waits on every click.

CDN provider B · 8 serversunhappy — server too farhappy⦿user A⦿user B

On CDN B it flips: user B is happy — now user A in Brazil is the one waiting.

3BlockDeliver — combine A + B + everyone's spare bandwidth, and both users are happy

happy⦿user Bhappy⦿user A

CDN provider ACDN provider BPeer nodes — anyone's spare bandwidth

18 servers + 9 peers → coverage everywhere · 30 countries

providers, ISPs and users rent out spare bandwidth for token incentives · server counts & positions illustrative

What founding taught me

• The whole stack of building — product strategy, technical architecture, hiring and execution, from concept to MVP, with no safety net and no playbook.
• Partnerships are product work — the Kingsoft Cloud and ArvanCloud deals shaped the architecture as much as any spec; a marketplace is only as good as its supply side.
• Recruiting against the odds — 15 people across five countries, from IIT Kharagpur, IIT Madras and Microsoft, signed up for an unfunded idea. Selling a vision to talent is the same muscle as selling a product to customers.

Scope, honestly

BlockDeliver reached MVP, not scale — the partnerships and 30-country coverage were real; a mature, self-sustaining marketplace was not. I wound it down and went to learn product craft properly inside companies solving hard data problems. The server counts and node positions in the maps above are illustrative; the model they explain is exactly what we built.