By ByteBot
Protocol Labs announced today — August 24, 2026 — that it will not renew Interplanetary Shipyard’s funding. Shipyard, the team maintaining Kubo, Helia, Boxo, Rainbow, IPFS Desktop, IPFS Companion, and public gateways including ipfs.io and dweb.link, will wind down all IPFS operations by September 30, 2026. Developers with applications or infrastructure relying on any of these projects have five weeks to act.
What Is Actually Shutting Down
Shipyard maintained more of the IPFS stack than most developers realize. The shutdown covers nine implementations losing dedicated maintainers: Kubo, Helia, Boxo, Rainbow, IPFS Desktop, IPFS Companion, Someguy, Service Worker Gateway, and IPFS Check. That is the Go reference implementation, the JavaScript implementation, production gateway software, and the browser tooling — effectively the entire maintained IPFS toolchain in one announcement.
Public infrastructure goes too. By September 30, the following will have no committed operator: ipfs.io, dweb.link, check.ipfs.network, delegated-ipfs.dev, IPFS bootstrap nodes, and the Wikipedia-on-IPFS collaborative cluster. The official Shipyard announcement leaves no ambiguity about Kubo specifically: “Kubo will no longer have dedicated maintainers responsible for new features, bug fixes, releases, or long-term stewardship.”
If you fetch content through ipfs.io or dweb.link — whether for NFT metadata, decentralized frontend delivery, or any other reason — those URLs will either go dark or fall under uncertain new ownership after September 30. If you run Kubo and depend on security patches, plan accordingly now.
This Didn’t Come Out of Nowhere
The Shipyard announcement is the last in a clear sequence. In August 2024, Cloudflare shut down its cloudflare-ipfs.com gateway, citing unsustainable traffic-to-adoption ratios. A week later, Brave deprecated native IPFS support in version 1.69.153 — their own data showed that fewer than 0.1% of Brave users had ever used the feature. Then, on August 15, 2026, Infura shut down its IPFS service. Now Shipyard.
Each exit cited the same numbers: high maintenance cost, low real-world adoption. The pattern suggests this was not a funding problem specific to Shipyard — it is a market signal. IPFS was built for a decentralized web that the market has not adopted at scale. The infrastructure providers that bet on ambient IPFS support have been withdrawing those bets one by one since 2024.
What Survives the IPFS Shutdown
The IPFS content addressing specification is open and does not die with Shipyard. CIDs — the hash-based identifiers at IPFS’s core — are permanent and deterministic. Data you have already pinned stays wherever you have pinned it. Pinata, which runs its own independent infrastructure with no dependency on Shipyard’s services, remains operational. Protocol Labs retains ownership of the ipfs.io domain and infrastructure, though no commitments have been made about future use.
The distinction matters: losing Shipyard means losing maintained implementations and public gateways, not losing the content-addressing protocol itself. Developers who depend on managed services like Pinata are in a fundamentally different position than those running self-hosted Kubo daemons or linking directly to ipfs.io endpoints in production code.
What To Do Before September 30
Five weeks is tight but workable. The first step is a dependency audit: search your codebase for references to ipfs.io, dweb.link, check.ipfs.network, and any other Shipyard-maintained endpoints. NFT projects should check whether token metadata links to an ipfs.io or dweb.link gateway — those URLs need to point to Pinata’s gateway or a self-hosted Rainbow or Kubo instance instead.
For Kubo users, the binary is MIT-licensed and you can self-host indefinitely. However, understand what you are accepting: no bug fixes, no security patches, and no new features from any organized team. Shipyard remains available through September 30 for transition questions — their announcement explicitly invites contact for handoff support. Reach out now rather than in week five.
The longer-term alternative worth watching is Iroh, built by ex-Protocol Labs developers at n0.computer. Iroh is not a drop-in replacement — it makes breaking changes to the IPFS spec in pursuit of better reliability and performance. Nevertheless, it is the most credible path toward sustainable content-addressed storage for developers who genuinely need this category of technology.
Key Takeaways
- Shipyard — the primary maintainer of Kubo, Helia, and IPFS public gateways — shuts down September 30, 2026, after Protocol Labs pulled funding
- Any production dependency on ipfs.io, dweb.link, or Shipyard-maintained tools needs to be remediated in the next five weeks
- Pinata and the IPFS specification itself survive; the risk is concentrated in Shipyard’s specific implementations and public infrastructure
- The exit follows Cloudflare (2024), Brave (2024), and Infura (2026) — low adoption and high maintenance cost is the consistent pattern
- Iroh (n0.computer) is the most credible longer-term alternative, but migration is not trivial and requires planning
