Space Raises $2.4 Million for an AI-Native Distributed Filesystem

SAN FRANCISCO — August 18, 2026 — Space has raised $2.4 million in pre-seed financing led by a16z Speedrun to develop a distributed filesystem that lets people and AI agents work with large cloud-hosted files as if those files were local. Golden Ventures, Northside Ventures and software-industry angel investors also participated.

The company is addressing a persistent gap between cloud capacity and practical access. Creative teams, architects and data-intensive businesses can store enormous files remotely, yet still wait for complete downloads, duplicate copies or application-specific ingestion before work begins. Space streams only the byte ranges requested by an application, so an existing desktop program can open and edit a file without placing the full object on the local disk.

One live data layer for applications, teams and agents

Space sits above the operating system and presents remote data through the familiar filesystem interface. A video editor can open a project larger than the computer’s remaining disk, while an AI agent can request the portions needed for analysis without first copying an entire library into a separate product. Updates are intended to appear across connected devices through the same live file state.

The company describes its longer-term direction as the “infinite computer”: the physical machine becomes a window into storage and compute rather than the boundary of a workflow. Product information and early-access details are available on the official Space website.

Financing priorities

• Expand engineering for distributed access, caching, synchronization and reliability.
• Support existing creative applications without forcing customers into a new editor.
• Develop controls for human and agent access to the same governed file collection.
• Grow adoption in video, marketing, architecture, engineering and construction.
• Prepare the platform for computer-vision, training-data and enterprise workflows.

Built from a production bottleneck

Space was founded by Matthew Ao, Arihant Bapna and Jason Zhao after their previous company repeatedly moved terabytes of video. The team lost time to transfers, reduced quality through compression and sometimes could not proceed because the required drive was elsewhere. Zhao had encountered the same problem while managing a decade of personal video footage across devices.

The founders built an initial prototype in November 2025. Early demand focused the company on a filesystem rather than another cloud-storage interface. The distinction is important: storage providers solve where bytes reside, while Space is trying to change when an application can use them. Streaming selected ranges can reduce waiting and local duplication, although performance still depends on connectivity, caching and workload behavior.

Security and reliability remain product requirements

A shared layer for humans and agents must enforce permissions consistently. Customers will expect encryption, auditable access, recovery from interrupted sessions and protection against conflicting edits. Space plans to develop the platform while working with early teams whose files are already too large for conventional local workflows.

The financing gives the company resources to turn a promising prototype into repeatable infrastructure. Success will be measured not only by files displayed in a desktop folder, but by startup time, editing responsiveness, synchronization accuracy, recovery behavior and the amount of local storage avoided under real production load.

About Space

Space is building an AI-native distributed filesystem for human creators and software agents. The platform is designed to make cloud-hosted data behave like local files inside existing applications without full-file downloads or duplicated working copies. Space is operated by Space Computer, Inc. in San Francisco.