Space raises $2.4M for AI-native filesystem to replace hard drives
▼ Summary
– Space, a San Francisco startup with $2.4m in pre-seed funding led by a16z’s Speedrun accelerator, is launching a distributed filesystem that streams only the exact byte ranges needed, eliminating local disk usage for files.
– The system mounts like a local drive, allowing native apps and AI agents to access live data without full downloads, targeting tasks like editing large videos or agents scanning files without ingesting entire documents.
– Founders argue existing tools like LucidLink or Dropbox are built for human syncing, while Space is rebuilt natively for both humans and AI agents, positioning the filesystem as the natural access layer for AI workloads.
– Space is in private beta with about 100 users and an organic audience of over 80,000, stemming from a previous company where transferring terabytes of video monthly became a bottleneck.
– The long-term vision is a “Space Computer” with unlimited storage and compute, but the viability of delivering this on $2.4m is uncertain, though the core problem—data access as the AI bottleneck—is significant.
Every few years, a startup emerges with the same bold promise: kill the hard drive. The latest contender is Space, a San Francisco company exiting stealth mode with $2.4 million in pre-seed funding and a strikingly direct approach. Their pitch is a distributed filesystem that lets both humans and AI agents work across petabytes of live data while reportedly using zero local disk space.
The funding round was spearheaded by a16z’s Speedrun accelerator, with additional backing from Golden Ventures, Northside Ventures, and roughly a dozen angel investors connected to companies like Parsec, Sentry, Superwhisper, and Modem.
At its core, the concept merges cloud storage and the local drive into one unified layer that operates just above the operating system. Files show up in the Finder and open in familiar native applications, yet nothing is ever fully downloaded. When an app or an agent requests data, Space streams only the exact byte ranges necessary to complete the task. That means a video editor could theoretically open a project larger than their laptop’s capacity, and an agent could scan an entire organization’s file structure while reading only the fragments relevant to its objective.
That final point is where the founders are concentrating their efforts. Most agentic tools currently require users to upload whole files to a chatbot, even when the task involves a single page or frame. It is slow, inefficient, and reminiscent of how giving agents their own native primitives has become a niche industry in itself.
“Humans and agents are working across more data than any local disk can hold,” said co-founder Jason Zhao. “They should not have to wait for files to be copied or ingested before work can begin.”
The timing aligns with a wider push to build the infrastructure beneath the AI surge, from VAST Data’s massive storage empire to a wave of startups selling the unglamorous data layers that agents depend on. Space’s argument is that the humble filesystem, the oldest abstraction in computing, is the natural home for that access layer, precisely because every app, device, person, and agent already understands it.
The challenge, however, is that a cloud drive that streams only what you need and mounts like a local disk is not a novel idea. LucidLink has given creative teams on-demand access to massive media projects for years, while Dropbox, Box, and Google Drive have all added selective sync features. Space’s bet is that those tools were built for humans syncing files one device at a time, whereas the upcoming wave of demand involves agents crawling data at machine speed. A filesystem rebuilt natively for both scenarios, they argue, outperforms a sync tool with AI bolted on afterward.
Space is currently in private beta with roughly 100 users and teams onboarded, and claims an organic following of more than 80,000 across various platforms. The founders trace that audience to a previous venture where shifting terabytes of video each month became a company-wide bottleneck.
“Space is disrupting one of the oldest assumptions in computing: a file has to exist on your device before you work with it,” said Jonathan Lai, a general partner at a16z. He framed the product as a single mechanism that serves human creators and AI agents simultaneously, a direction visible across the firm’s broader investments in agent infrastructure.
The founders’ longer-term vision is even more ambitious: a “Space Computer” where the physical machine becomes a window onto effectively limitless storage and compute. Whether a small team can deliver anything close to infinite computing on $2.4 million is, obviously, the open question, and the history of “the disk is dead” promises is littered with failures.
Still, the observation beneath the hype is solid: AI agents have quietly shifted the bottleneck from storage to data access, and the filesystem is a smart place to compete. Space has identified a genuinely large problem. The harder part, as always, is building the dependable, unglamorous machinery that makes petabytes feel local.
(Source: The Next Web)