The center of gravity in software development is shifting. For years, GitHub has been the default place where source code is stored, pull requests are reviewed, issues are tracked, and engineering teams coordinate their work. But the rise of AI-native coding tools is challenging that model from a new direction.
Recent developments associated with Zed’s Delta and Cursor Origin point toward a broader industry question: what should the platform around code look like when software is increasingly written, modified, reviewed, and coordinated by humans working alongside AI agents?
GitHub’s position is bigger than code hosting
Replacing GitHub is not simply a matter of building another Git repository service. GitHub’s influence comes from the network of workflows built around repositories: pull requests, code review, issue discussions, CI integrations, permissions, project management, package distribution, and the social graph of open-source development.
Any challenger therefore has to compete on more than Git compatibility. It must offer a compelling way for teams to collaborate, maintain trust, review changes, and manage the lifecycle of software.
Zed’s Delta: moving collaboration beyond the pull request
Zed introduced Delta as a multiplayer environment for coding with agents and reviewing what they build. According to Zed, Delta keeps code and conversations connected, allowing developers, teammates, and agents to work with the context behind a change rather than reconstructing that context later from a commit or diff.
Its underlying DeltaDB replicates conversations and worktrees together in real time. It works with existing Git repositories, so users can continue to commit and push normally while using Delta as a collaborative workspace.
The most significant change is the proposed replacement of the pull request as the primary unit of collaboration. In Zed’s public-beta announcement, the company said it had disabled pull requests on Delta’s own repository and was building the product inside Delta itself. Zed describes the new model as continuous engineering: the idea, implementation, review, and landing of changes happen in one evolving thread.
Delta is not yet a wholesale replacement for GitHub. Zed says its own main repository will remain on GitHub for now because that is where the community finds issues and submits contributions. That detail is important: Delta can challenge the GitHub workflow while still depending on GitHub for public collaboration and distribution.
Cursor Origin: an AI company enters Git hosting
Cursor’s Origin takes a more direct route into GitHub’s territory. Cursor’s official documentation describes Origin as its Git forge for storing and sharing code. In early beta, users can create repositories, clone, push, and pull with standard Git, mirror repositories from GitHub, browse and search code, and open, review, and merge pull requests.
Origin also connects repositories to Cursor’s broader AI environment. The documentation lists integrations with Cursor agents, cloud agents, automations, repository apps, and a command-line workflow. This is more than an editor feature: it gives Cursor a place where code can live, be shared, and become the input for AI-driven work.
There is also a gradual migration path. For repositories synced from GitHub, GitHub remains the source of truth and Origin acts as a mirror. For Origin-hosted repositories, Origin becomes the primary home. That distinction lowers the barrier to experimentation because teams do not have to abandon their existing GitHub setup immediately.
The real race is not “GitHub versus an editor”
It may be tempting to describe this as a race in which Zed or Cursor will directly replace GitHub. That framing is too narrow. The more realistic contest is over the developer’s primary working context.
- Code storage and version history: Repositories, branches, commits, and reliable recovery remain foundational.
- Change production: AI agents can generate code, refactor systems, and implement tasks at increasing speed.
- Change verification: Tests, review, security checks, and human judgment become more important as the volume of generated code rises.
- Collaboration: Teams need shared context around why a change was made, what it affects, and whether it is safe to merge.
- Execution and deployment: The platform that connects code to CI, infrastructure, environments, and production has significant leverage.
A product can weaken GitHub’s centrality without replacing every GitHub feature. If developers spend most of their time in an AI-native environment and treat code hosting as an underlying service, the balance of power can shift even while GitHub remains widely used.
Why AI changes the economics of developer platforms
Traditional developer platforms were designed around human-scale actions: a developer writes code, commits it, opens a pull request, waits for review, and merges it. AI agents change the speed and volume of those actions.
When an agent can create multiple implementation paths, update a large number of files, or work on tasks in parallel, the bottleneck moves. The challenge is no longer only producing code. It is managing the flow of proposed changes and deciding which changes deserve trust.
This creates demand for better diff experiences, richer context, traceability, sandboxed execution, automated verification, and review interfaces that explain more than a line-by-line patch. Tools that solve these problems may become central to the next generation of development infrastructure.
What challengers still have to prove
Despite the momentum behind AI-native development, replacing an established platform is difficult. A credible GitHub challenger would need to address several practical concerns:
- Reliability: Teams need dependable repository hosting, backups, access controls, and uptime.
- Interoperability: Existing Git repositories, CI systems, integrations, and developer habits cannot be discarded overnight.
- Security and governance: Enterprises need auditability, identity management, policy controls, and clear handling of proprietary code.
- Open-source network effects: A large user base, public repositories, maintainers, and contribution workflows are difficult to reproduce.
- Human trust: AI-generated changes must be understandable, testable, and attributable.
A platform shift may happen before a platform replacement
The most important outcome to watch is not whether a new product announces that it will replace GitHub. It is whether developers begin to think of code hosting as a background service rather than the center of their workflow.
Zed’s Delta and Cursor Origin represent two different approaches to that shift. Delta starts with conversations, evolving worktrees, and collaborative review. Origin starts with Git hosting and connects it to Cursor’s agent ecosystem. Both approaches recognize that AI changes the relationship between code, context, and collaboration.
GitHub still benefits from deep integration, familiarity, and a massive ecosystem. But the next generation of developer platforms may be defined less by where code is stored and more by how software work is coordinated between people and intelligent agents.
The race, in other words, may not be to replace GitHub in one dramatic move. It may be to make GitHub—or any traditional code host—less central to the daily act of building software.
Sources and further reading
