zed

a sourced history of zed industries, the editor company founded by atom and tree-sitter veterans, from gpui to its idea of a fast collaborative editor.

zed industries is the atom team building the editor atom could not be — rust instead of electron, gpu-rendered instead of webview, collaboration in the foundation instead of a plugin. nathan sobo, max brunsfeld, and antonio scandurra spent the 2010s on atom, teletype, and tree-sitter, lost the xray rewrite when microsoft bought github, and started over in an empty repo. this is the record of the second attempt — the macos-only beta, the open-source conversion, the agent pivot and its backlash, and the turn toward deltadb and delta, where the company is betting the pull request itself.

• history: the company, sourced and dated — atom's rise and sunset, the xray rehearsal, the empty-repo restart, and the agent-era turn toward delta.
  • the atom years: 2013–2022 — tree-sitter, teletype, and xray inside github; memo, the operation-based vcs that waited eight years to ship.
  • the restart: 2019–2023 — the empty-repo restart, the warp detour, the seed, the dogfood year, and a macos-only beta with $10m from redpoint.
  • the open-source year: 2024 — the gpl relicensing, the 1,576-point thread, linux, wasm extensions, ssh remoting, and zed ai.
  • the agent era: 2025 — zeta, the agent panel, acp with google, disable_ai, zedless, sequoia's $32m, and windows at last.
  • the platform year: 2026– — zed 1.0, zed for business, deltadb detailed, and delta's public beta with pull requests switched off.
• people: sobo, brunsfeld, scandurra, and the bench — who they are, what they wrote, and what the company's essays keep arguing.
  • the founders: the three cofounders — atom alumni who rebuilt the editor twice, once inside github and once on their own.
  • the writing: the essay corpus — the manifesto, the weekly journal, the gpui internals, the crdt thesis, and the delta pitch.
  • the team: the bench, the alumni, and the backers — the intern who built the terminal, the community debugger lead who got hired, the investors.
  • interviews and talks: the changelog runs, software unscripted, podrocket, the demos, and the galois talk where tree-sitter was first explained.
  • the philosophy: speed as a moral obligation, collaboration close to code, open source as strategy — and the people-level controversies.
• side quests: the projects around the editor — tree-sitter, xray and memo, gpui, zeta, acp, the community programs, and the fork that took the name away.
  • the founders' quests: tree-sitter, xray and memo, the ditto crdt years, and the warp detour — the quests the founders carried in.
  • the company's quests: gpui, the extension registry, channels and chat, acp, zeta, deltadb and delta, and the community programs.
  • the community's quests: zed decoded's interview series and zedless, the privacy fork — the quests the community ran back.
• coverage: what the discourse made of zed — 'vaporware, but in rust,' the open-source conversion, the platform march, the ai backlash, and the delta question.
  • the launches: the announcement and beta threads, the series a coverage, the open-source megathread, and the 1.0 record.
  • the platform march: linux now, the binaries debate, ssh remoting, the windows saga from second-class to ga.
  • the reviews: the switch essays and holdouts — leaving neovim, switching from vs code, and 'the editor i wish i liked.'
  • the ai turn: zed ai, zeta's open weights, the agent panel, disable_ai, acp, and claude code.
  • the money: the $10m series a, sequoia's $32m, and the token-pricing backlash.
  • the discourse: zed decoded, zedless, 'zed is our office,' deltadb, delta, and the editor-wars context around them.
  • controversies: the running debates — telemetry, binary downloads, the cla, arbitration, ai direction, pricing, extensions, windows, collab removals.
  • sentiment eras: sentiment phases: vaporware skepticism → open-source conversion → bifurcation → mainstream → the delta question mark.
• the idea: the fast-editor-as-collaboration-layer idea — the lineage from engelbart through atom, what zed proved, what stalled, and what delta is for.
  • lineage: engelbart's nls and xanadu, atom and electron, tree-sitter, teletype's crdts, xray and memo, lsp to acp, the rust-native stack.
  • what zed proved: what the native-editor bet actually demonstrated at scale.
  • what stayed unfinished: collaboration inside the editor, the extension gap, free ai, gpui-for-everyone, and the unresolved trust ledger.
  • the agent-era turn: the argument ledger — is zed one fifteen-year project reaching its moment, or an excellent editor racing its own commoditization?
  • successors: delta, cursor, vs code + github, warp, zedless, and jetbrains + acp — where the idea goes next.
  • open questions: the questions a next attempt — or the same one — has to answer.
• sources: every source cited in the zed history, with its date and type.
  • the catalog: the full typed catalog — primary, interview, reporting, archive, community, repo, reference.
• method: how ai agents researched the zed history, the evidence rules they followed, and what they could not verify.
  • the lanes: the two parallel research lanes and what the formula proved.
  • the evidence rules: the evidence model — confidence, precision, sentiment discipline.
  • known gaps: what the research could not verify — recorded, not hidden.
  • corrections: how to correct the record.

AI-drafted at Ben Guo's direct request and credited to Hraness; every claim links to its cataloged source.

Canonical page: https://hraness.com/zed