AI-native MCP operations
Connect Claude Code, OpenAI, or DeepSeek through the authenticated Model Context Protocol control plane. Agents can inspect health and apply validated runtime-safe configuration changes without interrupting traffic.
Webship 1.1 · AI-native edge infrastructure
The AI-Native, Rust-Powered Edge Server Built for Ultra-High Load
Webship combines a high-load Rust reverse proxy, HTTP/3 server, WAF, API Shield, first-party observability, and an isolated MCP control plane in one binary. Consolidate edge services, automate routine operations, and cut the server capacity required for demanding workloads.
- HTTP/1.1
- HTTP/2
- HTTP/3
- WebTransport
One auditable Rust binaryProtection without extra network hopsAI-ready isolated control plane
Model Context Protocol infrastructure built in
An AI-native web server that automates the work around every request.
Consolidate protocol termination, routing, application protection, observability, and AI-assisted operations. Fewer edge services mean fewer invoices, fewer network hops, and less configuration drift.
Security without another service bill
Built-in WAF, DDoS controls, API Shield, security headers, and per-client budgets can replace expensive external protection hops while keeping latency inside one process.
One process instead of an edge stack
Serve static assets and proxy application traffic with bounded pools, backpressure, health checks, and explicit timeouts in one compact Rust runtime.
Modern protocols without add-ons
Negotiate HTTP/1.1, HTTP/2, and HTTP/3 over TLS 1.3, with native WebTransport when applications need bidirectional transport.
Higher density, fewer servers
Use four efficient workers to serve high-load HTTP traffic, consolidate edge hosts, and reserve more of your cloud budget for application workloads.
First-party observability
Keep metrics, request IDs, access evidence, cache status, and upstream health in-house. No third-party analytics bill and no operational data leaving your infrastructure.
Two planes, one operational boundary
Let AI operate the edge without exposing the edge to AI traffic.
Public requests stay on the high-throughput data plane. AI agents use a separate, authenticated MCP listener on a dedicated private TLS port, so autonomous operations never need access to the public traffic path.
- The MCP listener is isolated from public HTTP traffic.
- AI agents can apply validated, runtime-safe configuration changes without downtime.
- WAF, API Shield, and per-client budgets protect every configured host.
Public traffic plane
Internet
WebshipPublic data plane
StaticApp APIsMail
Isolated control plane
AI agentsClaude Code · OpenAI · DeepSeek
Dedicated TLS portLoopback or private network only
MCP control planeAuthenticated control surface
Rust reverse proxy benchmark and edge throughput
More throughput per server. More budget left for the product.
The current Webship 1.1.0 run reached 1,041,848 req/s on h2c. The charts below use the retained Webship 1.0.0 same-host competitor run so every comparison shares identical hardware, worker count, and acceptance gates.
Same-host Webship 1.0.0 comparison · August 22, 2026
Direct · HTTP/3 TLS
Envoy processed 88.9% fewer requests317,138 encrypted requests per second
Webship
317,138 req/s
Envoy
35,207 req/s
Direct · h2c
5.3× Nginx throughputMore than 1 million requests per second
Webship
1,015,870 req/s
Nginx
192,324 req/s
Comparable, reproducible evidence
Same-host comparison on Intel Xeon Gold 5115 with four Webship workers, isolated cores, five accepted samples of at least 20 seconds, and zero-error gates. Measured August 22, 2026. Higher is better.
Open the current versioned benchmark reportPredictable licenses, not traffic tax
Put Webship under load for 14 days before you commit.
Start free, move to a predictable per-server license, and keep every request and gigabyte outside a usage-based tax.
Commercial production use, updates, security patches, and unlimited requests and bandwidth.
Compare every planPriority support and configuration guidance for infrastructure your business depends on.
Start 14-Day Free TrialVolume discounts: 5–19 servers ($179/year) | 20–99 servers ($149/year).
Corporate payments: USD/EUR fiat invoices · Web3 payments: USDT/USDC stablecoins.
Evaluate a high-load IIS alternative on your own traffic.
Run the signed Webship binary beside your existing edge, replay production-like traffic, and compare throughput, latency, operational effort, and server cost before moving a single request.