Datum is critical for the AI agent ecosystem because it solves the "last mile" of secure connectivity. For an agent to be truly useful, it needs to reach into private databases, interact with APIs, and move across different cloud environments. Datum provides the tunnels and machine-account infrastructure that allow these agents to travel securely without exposing sensitive internal networks to the public internet.
They are active at the infrastructure and connectivity layer of the agent stack. By prioritizing "Agentic Experience" (AX) and explicitly supporting protocols like MCP, they are building the plumbing that allows agents to be first-class citizens of the network. This makes them a fundamental enabler for developers building autonomous systems that require multi-tool and multi-cloud access.
Datum is built on the premise that the internet was designed for humans browsing websites, but the next billion connections will be agents talking to services. Founded in 2024 by Zac Smith and Jacob Smith, the company focuses on the connectivity gap between specialized AI platforms and traditional data infrastructure. Zac Smith previously founded Packet, a bare-metal cloud provider that sold to Equinix, bringing a high-pedigree infrastructure background to the problem of AI networking. Based in New York, the team consists of engineers and operators who have spent decades building the physical and logical backbones of the cloud.
The company is vocal about a shift in priority from Developer Experience (DX) to Agentic Experience (AX). In their internal hierarchy of values, AX sits at the top. This means the platform is designed to be automated at scale through tools like the Model Context Protocol (MCP), machine accounts, and automated skills rather than manual dashboards. If a platform cannot be managed by an autonomous agent, Datum considers it legacy technology. This focus manifests in their product suite, which centers on programmability and high-speed tunneling.
At the core of the offering is the Datum Network Cloud, which provides "connectors" based on QUIC tunnels and an "AI Edge" stack utilizing Envoy Proxy and the Coraza Web Application Firewall (WAF). These tools allow developers to create secure, private connections between disparate environments without the complexity of traditional BGP routing or site-to-site VPNs. For smaller teams and alternative cloud providers, this provides a "Galactic VPC"—a virtual private cloud that can span across multiple providers as if they were a single region.
Datum occupies a specific niche as a neutral intermediary. While the major cloud providers (AWS, Google, and Azure) build networking tools designed to keep data within their own ecosystems, Datum is designed to bridge them. This is specifically relevant for AI companies that might train models on specialized GPU clouds like CoreWeave or Lambda but store data on S3 or run inference at the edge. By providing a unified networking layer that is cloud-agnostic, Datum enables a modular approach to infrastructure.
The business model reflects this infrastructure-as-a-utility approach. They offer a free tier for builders and experimenters, moving to a usage-based scaler plan for production workloads. Their commitment to open-source software and transparent pricing is a direct reaction to the "sales gates" and vendor lock-in common in enterprise networking. By making infrastructure ergonomic and automated, they aim to lower the barrier for the next generation of cloud platforms and autonomous agents to operate at global scale.
Programmable network infrastructure for AI agents and alternative cloud providers.
Datum is hiring.