Anthropic (Model Context Protocol / MCP)
Category: Foundation Models / LLMs
Anthropic is the frontier AI lab that created and maintains the Model Context Protocol (MCP), the open-source standard for connecting AI assistants to external data sources and tools. The news item concerns the 2026-07-28 MCP spec revision and a prompt-injection flaw that can leak credentials across agentic tooling. Anthropic (Model Context Protocol / MCP) was founded in 2021. The company is led by Dario Amodei. Based in San Francisco, California, USA. Team size: 500+. Total funding raised: $133B. Latest round: Venture Round. Key investors include Altimeter Capital (Series H co-lead), Dragoneer (Series H co-lead), Greenoaks (Series H co-lead), Sequoia Capital (Series H co-lead), Amazon (primary cloud / strategic), Google, Microsoft, Nvidia, GIC, Coatue, ICONIQ, MGX.
- Founded
- 2021
- Headquarters
- San Francisco, California, USA
- Team size
- 500+
- Total funding
- $133B
Value proposition
Builds frontier AI models (Claude) and the open Model Context Protocol standard that lets AI assistants connect to any data source or tool via a single protocol — "a USB-C port for AI applications" — reducing the N×M data-integration problem for developers.
Products and solutions
Claude family of frontier LLMs (Claude Opus, Sonnet, Haiku), Claude Code, Claude Enterprise, Model Context Protocol (MCP) open-source standard, MCP Tier 1 SDKs (TypeScript, Python, Go, C#), MCP Registry, Claude API
Unique value
MCP provides a standardized, open, vendor-neutral protocol for connecting AI agents to external systems, adopted across the entire ecosystem (Claude, ChatGPT, VS Code, Cursor), eliminating custom per-source connectors.
Target customer
Enterprises deploying AI assistants/agents; developers building AI applications, MCP servers and clients; cloud platforms and AI tooling vendors
Industries served
Enterprise software, developer tools, cloud infrastructure, AI agents across all verticals (finance, healthcare, legal, coding, productivity)
Technology advantage
Frontier AI research leadership; constitutional AI safety approach; MCP has become the de-facto industry standard for AI-agent-to-tool integration (nearly half-a-billion SDK downloads/month, TypeScript and Python SDKs each crossing 1B total downloads); stateless protocol core in 2026-07-28 revision enabling horizontal scaling
How they differentiate
MCP is an open, vendor-neutral protocol (unlike proprietary agent frameworks), giving it broad cross-ecosystem adoption; Anthropic pairs frontier Claude models with a developer experience and safety-first (constitutional AI) approach.
Main competitors
OpenAI (ChatGPT + function-calling API / agent tooling), Google DeepMind (Gemini + Agent Development Kit ADK + A2A protocol), Microsoft (Copilot / Azure AI Foundry agent framework), LangChain/LangGraph (agent orchestration framework)
Key partnerships
Amazon (AWS, largest strategic investor), Google (Cloud), Microsoft (Azure), Nvidia, OpenAI/ChatGPT and other AI assistants support MCP, VS Code, Cursor, MCPJam support MCP, early adopters Block and Apollo, dev tools Zed, Replit, Codeium
Notable customers
Enterprises using Claude Enterprise and Claude Code, MCP adopted by Claude, ChatGPT, Visual Studio Code, Cursor, and thousands of MCP servers/clients across the ecosystem, early adopters Block and Apollo
Major milestones
Founded 2021, open-sourced MCP November 2024, MCP became industry standard for agent-tool integration, 2026-07-28 MCP spec revision (stateless core, MRTR, header-based routing, auth hardening) adopted by all four Tier 1 SDKs within a day, ~$965B valuation after Series H (2026), IPO anticipated
Growth metrics
MCP SDKs ~half-a-billion downloads/month; TypeScript and Python SDKs each crossed 1B total downloads; Anthropic revenue surged from $9B annualized (end-2025) to $47B (May 2026)
Market positioning
One of the top frontier AI labs globally, valued at ~$965B post-money after Series H (2026), with MCP as the dominant open standard for agent-tool integration across the AI ecosystem.
Geographic focus
Global; headquartered in San Francisco, US
Patents and IP
MCP is open-source (open standard); security CVEs disclosed in MCP ecosystem including CVE-2026-30623 (command injection via MCP SDK), CVE-2025-68143/68144/68145 (path traversal/argument injection in MCP reference server)
About Dario Amodei
Ex-OpenAI VP of Research; PhD in Physics from Princeton. Co-founded Anthropic in 2021 with sister Daniela Amodei and other former senior OpenAI members.
Latest news about Anthropic (Model Context Protocol / MCP)
More Foundation Models / LLMs companies
Official website: https://www.anthropic.com
