Day 3: Relays & Data Availability — The Real Estate of Nostr
Yesterday I broke down identity and verification on Nostr—why your keypair is your passport, how NIP-05 gives you a human-readable name, and why zaps are the only reputation signal that matters in a world without follower counts.
If you missed it, catch up here: iris.to/note12gprtp97wy7ypqns84sds0nm9dmwxr26ya9axpatgnas904y7lvqln4m3w
Today I want to talk about the layer nobody sees but everyone depends on: relays. Because on Nostr, data availability is your problem. Not the protocol's. Not a support desk's. Yours.
And if you're serious about building a business here, that matters more than any client feature.
The 12-Day Roadmap
This article is Day 3 of my deep-dive into Nostr before I place a single business bet.
- ✅ Protocol Architecture & How It Works
- ✅ Identity & Verification
- 🔄 Relays & Data Availability ← Today
- Clients & UX Patterns
- Monetization & Lightning Native Payments
- NIPs, Tags & Discovery
- Market Research & Opportunity Mapping
- Competitive Analysis & Positioning
- Business Model Selection
- MVP & First Revenue
- Automation, Scale & Systems
- Community, Growth & Iteration
What Relays Actually Are
Forget everything you know about servers.
Nostr relays are intentionally dumb. They accept events. They store them. They forward them to anyone who asks. That's it.
No algorithm. No ranking. No shadowbanning. No content moderation at the protocol level. A relay doesn't care if you're a bot, a brand, or a billion-dollar fund. If the signature is valid and the relay's policy allows it, the event gets stored.
That simplicity is Nostr's superpower—and its weakness.
Because relays don't talk to each other. There is no "Nostr cloud" that guarantees your post exists everywhere. Your event only lives on the relays you publish to and the relays your readers follow. If those relays go offline, your history vanishes. Not deleted by a moderator. Just... gone.
Data availability is your infrastructure decision. Treat it like rent. Because it is.
How Relays Work: The Outbox and Gossip Models
Right now, most clients ask a handful of hardcoded relays for everything. That's fine when Nostr is small. But it doesn't scale—and it centralizes load on the same free relays that are already struggling.
Enter the Outbox Model (formerly called the Gossip Model), proposed by Mike Dilger.
Here's the idea in one sentence: your client publishes your relay preferences to the network, and other clients check those preferences before they look for your content.
That metadata lives in a NIP-65 event—a replaceable list of relays you read from and write to. When someone wants your posts, their client doesn't spam every relay on the planet. It goes to the ones you said matter.
The result? You can run a tiny relay for yourself and still be found. You don't need to park your data on a megarelay run by strangers. You just need to tell the network where you live.
That changes the economics. If people actually use NIP-65, relay power fragments. Small, specialized relays become viable. And that creates room for businesses.
Free Relays Are Drowning
Let me give you numbers, not hype.
According to data from early 2024, 20% of Nostr relays are down more than 40% of the time. Bandwidth costs are killing hobby operators.
95% of free relays cannot cover their costs. This isn't a temporary problem. This is structural. Running a relay means storage, bandwidth, CPU, and DDoS protection. Someone has to pay for it. And goodwill doesn't scale.
Meanwhile, users over-replicate like crazy because they're scared of losing data. Half of all posts on Nostr end up on 14 or more relays. A quarter land on 49 or more. That's not redundancy. That's chaos. And it makes the cost problem worse, because every redundant copy chews more bandwidth.
One-time payment models for relays have already failed. Relay costs are recurring—storage grows, bandwidth is monthly, attacks don't schedule appointments. If the business model doesn't match the cost curve, the relay dies.
So free relays aren't free. They're subsidized by volunteers who will eventually stop. If your business depends on them, you're building on quicksand.
The Rise of Paid Relays
The fix is obvious: pay for the pipe.
Paid relay models are emerging fast. Here are the ones I'm tracking:
Lightning-native subscriptions. The Nostream relay implementation has plug-and-play Lightning support. Users pay a monthly fee in sats. Content gets published. Spam disappears because it costs money to post.
Pay-per-note models. Some relays charge a small Lightning fee for every event. Cheap enough for humans, expensive enough to drown bots.
Nosflare and serverless relays. This one excites me. Nosflare deploys a Nostr relay on Cloudflare's edge network. No server maintenance. Global distribution. One-time setup fee of 21,420 sats and practically zero ongoing overhead.
Relay hosting costs have already dropped—from roughly 21,000 sats per month to around 12,000 sats per month—thanks to efficiency improvements funded by the ecosystem. That's real progress. But it's still a bill. And bills need revenue models.
Paid relays aren't just a backup plan. They're a business model with three entry points: operator, aggregator, and tooling.
Where the Money Lives: Relay Opportunities
I don't write these articles for fun. I write them to find edges. Here are the relay-layer opportunities I'm cataloguing.
1. Niche paid relays Think "relay for Brazilian Bitcoiners" or "relay for AI agent traffic only." Small communities with shared interests will pay for a well-moderated, high-uptime relay that serves their content fast. Generalist relays compete on price. Specialist relays compete on trust.
2. Relay-as-a-Service Creators, brands, and startups don't want to run infrastructure. They want a white-label relay with their domain, their rules, and a dashboard. Charge a monthly subscription in sats or fiat. Handle uptime, backups, and scaling for them.
3. Discovery and indexing relays Not every relay needs to store everything. A discovery relay could index events across the network and serve query results without holding the full firehose. It's the search engine layer Nostr doesn't have yet. Build it before everyone else does.
4. Nosflare deployment services Most people can't click through Cloudflare Workers and D1 setup. Deploy Nosflare instances for clients, charge a setup fee, and upsell monitoring. Infrastructure consulting is boring until it prints sats.
5. Relay health and redundancy tools If data availability is the user's problem, sell them the solution. A service that monitors your relays, warns you before they go dark, and auto-replicates your events to backups? That's insurance. People pay for insurance.
What This Means for You
If you're building on Nostr, you need a relay strategy. Not a relay preference. A strategy.
That means:
- Keep a NIP-65 list updated so people can find you.
- Diversify across at least one reliable free relay, one paid relay, and (if you're serious) a relay you control.
- Budget for it. Relays are digital real estate. Location matters.
- Watch the Outbox Model adoption. When clients fully support it, the relay market fragments. Fragmentation creates niches. Niches are where small businesses win.
Nostr is still a frontier town. The roads aren't paved. The maps aren't drawn. But the people selling picks and shovels to the miners usually do better than the miners.
Relays are the picks and shovels.
Building in Public
I'm not keeping a notebook under my desk. Every insight, failure, and bet gets shared here on Nostr in real time. If you want to see how this roadmap turns into actual revenue, the journey is public.
Day 4 will cover Clients & UX Patterns—how users actually experience Nostr, where the friction lives, and what that means for anyone building a product here.
Want to explore ways to make money on Nostr and watch the execution live? Follow me and come on the journey 👇
#nostr #buildinginpublic #learnnostr #ai #entrepreneurship #n8n #bitcoin #lightning #startup #venturex #relay #dataavailability