Cloud outage
- Signal
- Provider status history, independent monitors
- Who buys
- Online businesses, SaaS, marketplaces
- Status
- Building
Prediction markets already price events in real time. We turn that price into protection a business can buy: pick the event, pick the payout, hold the contract. First up, cloud outages.
Explore a contract that would pay if Amazon EC2 in AWS us-east-1 (N. Virginia) experienced a qualifying outage lasting more than during .
At an assumed 3.5¢ per $1 of payout for EC2-USE1-2H-30D, this contract costs $875 up front.
Prototype contract. Prices are scenario assumptions. First live listing is the next milestone.
Illustrative service and region. AWS isn't involved.
A scenario assumption you control. Live prices will come from the venue, fees included.
Selected specification: EC2-USE1-2H-30D
Each threshold and period pair is its own contract specification and lists separately.
Every business exposed to the same outage buys units of the same contract. Each unit pays $1 on a Yes. Want $25,000? Hold 25,000 units. Nobody needs a bespoke market; the contract is standard and the position size is yours. Change the threshold or the period and you're looking at a different specification, listed separately.
We read the provider, not your website. Yes means the named service was unavailable region-wide for longer than the threshold, without interruption, inside the window. Degraded performance doesn't count. Another region doesn't count. Exactly 2 hours doesn't count. Your own code breaking while AWS runs fine doesn't count either.
A region down for two hours. A port shut for a week. 38°C for three days running. If a public signal can measure it, we can write it as a contract: window, threshold, evidence rules, all fixed before anyone trades.
Everyone exposed to the event buys units of the same Yes/No contract. Your payout sets how many. Whoever thinks the event won't happen takes the other side, and their price is your cost.
The window closes and the contract resolves Yes or No under the venue's rules. The payout per unit was written down on day one.
We took what works in parametric insurance, a written trigger and a payout fixed before anything happens, and put it on rails that already exist: public prediction markets. Prices in the open. Collateral on-chain. Resolution by rule.
Priced in the open.
Every unit has a visible market price, set by people taking either side. No underwriting desk, no quote you can't see.
Collateralised on-chain.
Both sides post collateral into one pool when the position opens. The payout exists before the outage does.
Resolved by rule.
The contract resolves under the venue's approved rules and evidence requirements, the same process the venue runs on every market it lists.
We're building for public prediction markets such as Polymarket. The first listing is the milestone we're working toward.
The agreed amount on a Yes, no matter what your loss turned out to be.
The contract reads public signals about the provider. It never looks at your site, your revenue or your customers.
What the other side charges to carry the risk for the window. Not refunded on a No.
Cloud goes first because the signal is public and the losses are obvious. Freight, power and weather run on the same playbook: find the signal, write the rule, list the contract.
One specification at a time. Cloud outage first.
23 October 2026
Lock the settlement sources. Evidence sources, measurement method and tie-break rules for EC2-USE1-2H-30D, written so a stranger could resolve it.
BuildingList the first contract. Take the specification to venues and line up the other side.
NextWire the execution. Pick a payout on this page, hold the position on the venue. One flow.
NextRuben Hallali, PhD in atmospheric sciences, worked at Météo-France before co-founding HD Rain, and has worked on weather risk data and parametric protection for businesses since. He's launched twice before the machinery was finished and built it in the open both times. Everyrisk is his next one, independent of his other companies.