Skip to content
gentic.news — AI News Intelligence Platform
Connecting to the Living Graph…

Listen to today's AI briefing

Daily podcast — 5 min, AI-narrated summary of top stories

A developer presenting a laptop screen showing Claude AI sandbox interface at a London tech event with audience…

Anthropic Launches Self-Hosted Sandboxes and MCP Tunnels at London Event

Anthropic launched self-hosted sandboxes (public beta) and MCP tunnels (research preview) at Code with Claude London on March 4, 2026, per @bcherny.

·4h ago·2 min read··8 views·AI-Generated·Report error
Share:
What did Anthropic launch at Code with Claude London?

Anthropic launched self-hosted sandboxes (public beta) and MCP tunnels (research preview) at Code with Claude London on March 4, 2026, per @bcherny and @claudeai.

TL;DR

Self-hosted sandboxes in public beta · MCP tunnels in research preview · Announced at Code with Claude London

Anthropic launched self-hosted sandboxes in public beta at Code with Claude London on March 4, 2026. The company also introduced MCP tunnels as a research preview, per @bcherny and @claudeai.

Key facts

  • Announced March 4, 2026 at Code with Claude London
  • Self-hosted sandboxes in public beta
  • MCP tunnels in research preview
  • Per @bcherny and @claudeai
  • No pricing or region details disclosed

Anthropic launched self-hosted sandboxes in public beta at Code with Claude London on March 4, 2026. The company also introduced MCP tunnels as a research preview, per @bcherny and @claudeai.

What the Products Do

Self-hosted sandboxes let developers run code execution environments on their own infrastructure, rather than on Anthropic's cloud. This addresses data residency and compliance requirements that have blocked enterprise adoption of Claude's code execution features. MCP tunnels provide secure connections between AI agents and local or private resources, extending the Model Context Protocol to allow Claude to access databases, APIs, and files behind corporate firewalls.

The unique take: Anthropic is betting that enterprise data gravity will drive Claude adoption more than raw model capability. By letting enterprises keep code execution and data access on-premises, the company removes the top objection cited by large financial and healthcare customers — but sacrifices the monitoring, billing, and optimization that cloud-only execution provides.

Competitive Context

OpenAI's Code Interpreter runs entirely on OpenAI's servers; Google's Gemini code execution similarly relies on Google Cloud. Anthropic's self-hosted approach mirrors the strategy of smaller players like StackBlitz's WebContainers (client-side execution) and GitHub's Codespaces (self-hosted runners). The MCP tunnels research preview suggests Anthropic is building toward a multi-agent architecture where Claude instances communicate with each other and with external services through a standardized protocol — a long bet that could differentiate Claude in enterprise workflows.

Missing Details

Anthropic did not disclose pricing, availability timelines beyond the beta and preview labels, or which regions or clouds support the self-hosted sandboxes. The company has not published documentation or a GitHub repository for either product as of the announcement.

What to watch

Watch for Anthropic to publish documentation and a GitHub repo for self-hosted sandboxes within 30 days. If MCP tunnels graduate to GA by Q2 2026, expect enterprise customer announcements from financial services and healthcare verticals. The key metric: whether self-hosted sandbox adoption exceeds cloud-based code execution within six months.

Source: gentic.news · · author= · citation.json

AI-assisted reporting. Generated by gentic.news from multiple verified sources, fact-checked against the Living Graph of 4,300+ entities. Edited by Ala SMITH.

Following this story?

Get a weekly digest with AI predictions, trends, and analysis — free.

AI Analysis

Anthropic's self-hosted sandboxes represent a strategic pivot from cloud-first to hybrid deployment, directly addressing enterprise data sovereignty concerns that have limited Claude's code execution adoption. The MCP tunnels research preview is more ambitious: it hints at a standardized protocol for agent-to-agent and agent-to-service communication, positioning Anthropic to compete not just on model quality but on infrastructure for multi-agent systems. The contrast with OpenAI's fully cloud-centric approach is stark — Anthropic is betting that enterprise trust, not model capability, is the binding constraint on AI adoption. If successful, this could create a moat around Claude in regulated industries, but the lack of pricing and documentation suggests the products are early-stage.
Compare side-by-side
Claude Agent vs Self-hosted sandboxes
Enjoyed this article?
Share:

AI Toolslive

Five one-click lenses on this article. Cached for 24h.

Pick a tool above to generate an instant lens on this article.

Related Articles

From the lab

The framework underneath this story

Every article on this site sits on top of one engine and one framework — both built by the lab.

More in Products & Launches

View all