Platform Documentation Businesses Hosting Blog Company 2084
Get started Sign in

SKY·PHAGENorse ský, cloudLatin -phagous, to devour

Reimagining the internet, where users own the infrastructure.

Enough of the hyperscalers. It's time for the people to run the internet. SkyPhage is harboring the next-generation of cloud platforms, where anyone can be paid to contribute infrastructure to a resilient, self-healing network where everyone benefits. Use your compute, or someone else's.

The town should be owned by the town.

Write inRust · Go · JavaScript · Python · C/C++ · anything that compiles to wasm32-wasip1

Host onRaspberry Pis · old desktops · NAS boxes · hypervisor VMs · rack servers · GPUs

01 / 08The problem

The internet has a centralisation problem

Today's web is dominated by a few hyperscale cloud providers. Your data, your applications and your business all depend on infrastructure you don't control. Outages cascade globally. Costs keep rising. And billions of compute resources sit idle around the world. We're changing that.

3
cloud providers control over 65% of all internet infrastructure
85%
of server capacity worldwide sits idle at any given moment
1
point of failure can take down millions of websites and services
02 / 08Three ways in

Be part of the new internet

Whether you have spare compute, need distributed execution, or just want to build without limits.

The SkyPhage dashboard listing the workloads in the edge-api project, with status, requests, duration and the agent each runs on.
Fig. 1 Workloads in a project, as the dashboard shows them.
03 / 08Orchestration

Compute flows to where it's needed

Every host carries a location and a reliability score it earned. A workload lands on the nearest host that can be trusted with it, is watched while it runs, and moves the moment that stops being true.

scheduler · hello-api · acme/productionlive
  1. schedulerhello-api needs 128 MB · 0.5 vCPU · near Manchester. Scoring 4 candidates.
  2. schedulerphage-7f3a9c Shed, Stockport 12 km rel 0.98 chosen
  3. schedulerphage-9d8e7f Office, Leeds 60 km rel 0.91
  4. schedulerphage-b21e44 Garage, Dublin 265 km rel 0.97
  5. schedulerphage-5e6f7a Shelf, Newcastle 180 km rel 0.72
  6. phage-7f3a9cserving at https://hello-api-acme.phage.cloud · heartbeat 12 ms
  7. monitor1.2k requests/min · p50 14 ms · 0 errors
  8. monitorphage-7f3a9c missed 3 heartbeats
  9. schedulerreassigning hello-api → phage-b21e44 (Garage, Dublin)
  10. phage-b21e44serving at https://hello-api-acme.phage.cloud · heartbeat 48 ms
  11. reputationphage-7f3a9c back online · reliability 0.98 → 0.93
04 / 08How it's different

A fundamentally different architecture

Traditional cloud stacks were designed for a world of walled gardens. SkyPhage was built from the ground up for a distributed, open future.

PlacementGeo-aware orchestrationA modified haversine routing places workloads on the nearest reliable host, weighing hops, latency, distance, reliability and free capacity in real time.distance · hops · latency · reliability · capacity
IsolationWASM sandboxingEvery workload runs in its own WebAssembly sandbox on Wasmtime. Memory-safe, capability-restricted, hardware-agnostic. No containers required.wasmtime · wasm32-wasip1
TrustReputation, not authorityHosts earn scores over time. Higher reliability attracts more workloads; bad actors are deprioritised automatically.+rep per success · −rep per failure
NetworkingZero configurationAutomatic TLS, NAT traversal and a public HTTP gateway. Workloads are reachable the moment they deploy.TLS · STUN · UPnP · <workload>-<org>.phage.cloud
ConfidentialityAttested confidential computeHosts with SEV capability are attested when they are claimed and badged by tier, so a sensitive workload can insist on an enclave it can verify.Platinum SEV-SNP · Gold SEV-ES · Bronze SEV
ServicesComponents built inHost-provided services any workload can call from any SDK. No sidecars, no managed add-ons to provision.http · kv · cache · crypto · state · peers · wasd · phsql
05 / 08Hosting

Turn the hardware you already own into income

GPUs and bandwidth become a revenue stream while strengthening the open network.

  1. 1Boot Phage OSWrite the image to a USB stick or PXE-boot it. Immutable, signed, updated A/B. Your boot disk is never offered to anyone.
  2. 2Type the code into your dashboardThe machine lands in your "Phage OS Hosts" rack as a private, attested host. Move it to any site or rack later.
  3. 3Offer what you chooseDisks and PCI devices are opt-in, one at a time. Share a GPU as MIG slices or SR-IOV functions, or keep it exclusive.
PHAGE OStty1 · before claim
   Add this machine to your account

     code   CDFG-HJKM
     go to  https://skyphage.com

   Sign in, open Hosting, and enter the code above.
   Nothing is hosted here until you do.
PHAGE OStty1 · serving
   Hosting

     address   192.168.1.42
     class     Gold    (SEV-ES)
     depot     enabled
     gateway   disabled

   Change what this machine offers from your
   dashboard, under Hosting.
Fig. 4 The whole local interface of a Phage OS machine.
06 / 08Developers
$ cargo new hello-api && cd hello-api
    Creating binary (application) `hello-api` package
$ cargo add phage
      Adding phage to dependencies
$ cargo build --target wasm32-wasip1 --release
   Compiling hello-api v0.1.0 (/home/you/hello-api)
    Finished `release` profile [optimized] target(s) in 6.18s
$ ls -lh target/wasm32-wasip1/release/*.wasm
-rw-r--r-- 1 you you 1.8M Oct 26 14:02 hello-api.wasm
# upload hello-api.wasm to your project depot, then deploy
→ https://hello-api-acme.phage.cloud

From code to globally distributed execution in three steps

Compile to WASM, push, and let SkyPhage handle the rest. Ship globally in minutes.

  1. 1Compile to WASMBuild your application in any supported language. One artifact runs everywhere.
  2. 2Define requirementsResource needs, geo preferences and duration. Upload the binary to your project depot.
  3. 3SkyPhage orchestratesThe best available host runs it, monitors it and fails it over when it has to.
  4. Quickstart in Go, JavaScript or PythonDeploy a community application
07 / 08For business

Low-cost, reliable global compute without vendor lock-in

The machines already exist: validated hosts in homes, offices and racks, already switched on. SkyPhage runs your workloads on the reliable ones nearest your customers, moves them when a machine fails, and bills only what they use.

Compared onThe hyperscalersColocationSkyPhage
Up frontNothing, then a bill that grows with every service you adoptLand, power, hardware and the people to run itNothing. Top up a wallet and set a budget per workload
Metro datacentersThe regions on offerWherever you built itAnywhere on the planet with an available CPU. On the hosts nearest your users, placed as demand moves
When a machine failsYour architecture has to copeYour on-call engineer has to copeThe scheduler moves the workload to the next reliable host
Under attackProtection you configure and pay forScrubbing you build or rentNearby reliable hosts scrub the attack for you and provide a defence perimeter until the attack stops
Sensitive dataTheir hardware, their attestationYour hardware, your auditAttested enclaves by tier: Bronze, Gold or Platinum
LeavingProprietary services to unpickA building to sellWASM and OCI. Take your artifacts anywhere
More capacityAnother data centreAnother data centreMore hosts and machines onboarded
SaaSAn API answered beside its customersEach request is served by the nearest reliable host serving it, wherever the customer is that day.
AgenciesA render queue that costs nothing while it waitsGPU jobs go to shared MIG slices on Gold hosts and are paid per job from the agency's wallet.
RetailBlack Friday without buying servers for itThe rush spreads across the network, and the extra copies scale back in on Monday.
08 / 08The team
We left Big Tech to decentralise it.

Built by the humans who build the cloud

SkyPhage was founded by engineers who spent years building the infrastructure behind the world's largest hyperscalers. They have seen first-hand how centralised systems fail, and SkyPhage exists to give the internet back to the people who use it.

Registered
England and Wales
Team
UK and Ireland, remote
Lock-in
None, ever
Runtime
WASM + OCI
  1. 1Security above all elseEvery decision, every placement and every line of code is measured against the security and integrity of the network first.
  2. 2An orchestrator and a brokerWe connect demand to independent hosts with spare capacity, and coordinate that exchange rather than owning the infrastructure.
  3. 3We don't run your workloadsThey execute on the independent hosts that make up the network, not on machines we operate.

The future of the internet is community owned