Val Town Alternatives: The Shutdown, the Timeline, and How to Migrate
A small script with a public URL, a database, and a schedule is an easy place to leave a side project. Val Town is winding that place down. Steve Krouse published the decision on 10 November 2026, and the pages that matter next are the Val Town alternatives the company wrote for people who still have apps running there.
The letter, signed Steve Krouse and dated 10 November 2026, says Val Town took a leap of faith from the people who built on it. “We’re sorry we can’t continue to serve you.” The reason he gives for the business is specific. They “couldn’t compete in the cloud computing space.” He calls that bad news for Val Town and useful news for anyone leaving, because other hosts exist and “today’s LLMs are great at these types of migrations.” The letter points to guides for Cloudflare, Vercel, Netlify, and Render, and to help at support@val.town or on Discord.
Billing paused the day of the letter. Existing subscriptions keep being serviced for free. Anyone who prepaid an annual plan gets a partial refund. The letter does not say how large that refund is.
This was written on 11 November 2026 in Sydney. Sydney is on Australian Eastern Daylight Time, eleven hours ahead of UTC. Dates below follow the source. Like counts and Hacker News points are a snapshot from that morning in Sydney. They will have moved.
Why is Val Town shutting down?
The letter’s reason is the cloud-computing line above. Krouse also thanks the team for four years of work, and thanks Dan Levine as “the best investor I could’ve asked for,” plus his other investors for trust, support, and capital. The post prints no amount raised.
The runtime under Val Town has its own end date. Val Town’s migration guide says Val Town uses Deno, that Deno is winding down, and that “the core maintainers won’t support Deno beyond October 2027.” The guide’s recommendation is Node, Bun, or a provider runtime such as Cloudflare Workers or Vercel Functions.
Ryan Dahl’s post, “Deno is joining Cloudflare,” is dated 9 October 2026. It says the entire Deno team is joining Cloudflare, and that future work goes into that shared platform rather than a separate runtime and a separate host. The concrete terms on the page: the Deno runtime gets another year of monthly releases, limited to bug fixes and security updates, and after that year they end their development of the runtime. Deno stays open source. Deno Deploy keeps operating for six months and then shuts down, with migration support for paying customers moving to Cloudflare Workers. The October 2027 line is Val Town’s reading of that year. Dahl’s post says “another year,” and it is dated 9 October 2026. It does not print the words October 2027.
On Hacker News the argument split three ways. Sadness and praise sat next to “this business is hard,” next to “this is ordinary deal-making,” next to “build it small.”
The thread was posted by tosh at 17:31 UTC on 10 November, linking the letter. When checked on the morning of 11 November in Sydney it showed 187 points and 50 comments.
simonw wrote: “I’m really sad about this.” Val Town, in that comment, is “a beautiful collection of ideas,” and “building a sustainable business on top of those ideas” is “extremely hard,” because it is hosting that still asks people to build in Val Town’s way.
cmontella wrote that they knew Steve from Eve, that he had long argued for end-user programming, and that the shutdown was unsurprising. “Nothing to to do with the quality of Val Town, but making a business in this space is really, really hard.” The doubled “to” is theirs. They then named three raises, in parentheses: Eve $2 million, Darklang $3.5 million, Val Town $7 million. Those figures are cmontella’s claim. Krouse’s letter thanks investors and states no round. The rest of the comment is a theory in the same voice: a couple of years of seed money, when they think a project like this needs five, and a Darklang story about laying off the team. That account is theirs. They also asked for a postmortem.
drekipus replied to that comment: “Why bother with investing, why not just build it small and sustainable to begin with?”
plesiv praised Steve’s writing about programming, then described their own product. Steve inspired them to build Gitlip, a Val Town-like platform, in early 2025. They “quietly stopped working on it about a year ago because it was difficult to get paying customers.” The link in the comment is gitlip.com. Read the difficulty as their report about a product they were building.
A second cluster treated the week as a pattern of deals. hardwaregeek wrote: “Feels like there’s an era of devtools companies coming to a close.” They listed Astro, Astral, Deno, Bun, Turso, and Void Zero as acquired, and guessed at strong buyout offers or a hard fundraising climate. That list is their comment. This story checked Dahl’s 9 October post, which says the Deno team is joining Cloudflare. It did not check the other names.
swyx replied to hardwaregeek: “or this is just normal startup M&A cycles and you just happened to pay attention to this one more closely.”
One sourced week of that attention is already on the record. On 4 November 2026 Meta bought Conductor, OpenAI agreed to buy Pierre, and HumanLayer said it was joining Factory. Conductor published a shutdown calendar with the news. The names, the dates, and the Conductor alternatives people posted are a separate story. Val Town’s letter, six days later, gives its own reason: they could not compete in cloud computing, and they set a wind-down.
The public replies on X, the same day, were shorter. Krouse’s post at 17:06 UTC said Val Town is winding down, thanked people for building on it, and said the team is now focused on helping customers migrate. At the check it showed 673 likes. Cassidy (@cassidoo) at 17:24 UTC wrote: “Val Town is awesome and their team did some great work over there.” That post showed 48 likes. Armin Ronacher (@mitsuhiko) at 18:34 UTC wrote: “Val town was lovely! Sad to see :(” That post showed 4 likes.
When does Val Town shut down?
The letter’s table is the calendar. New signups and plan upgrades ended on the day it was posted. The later dates are still ahead.
| What ends | Date |
|---|---|
| New signups and plan upgrades | 10 November 2026 |
| Read-only mode, not including sqlite and blob storage | 10 December 2026 |
| All other services | 10 January 2027 |
| Open-sourcing Val Town | 1 February 2027 |
The rows are the timeline in Steve Krouse's 10 November 2026 letter. The sqlite and blob exception is the letter's own parenthetical on the December row.
The migration guide says the same shape in a sentence. Code becomes read-only on 10 December 2026. Apps on Val Town, which the company calls vals, keep running until 10 January 2027.
Redirects of val.run URLs can be added and edited until 10 December 2026. They stay temporary until that day. On 10 December, redirects already set become permanent, which is also the day code becomes read-only.
Billing paused on 10 November, and existing subscriptions keep being serviced for free. The letter gives 10 January 2027 as the end of all other services. It does not name a later date for that free servicing.
Val Town alternatives
The letter names four written guides: Cloudflare, Vercel, Netlify, and Render. The where-to-migrate page is wider. It compares those hosts with Supabase and Railway across website hosting, SQLite, blob storage, cron, email, access control, AI, and functions. The page says “Cloudflare is the most feature-equivalent platform to Val Town,” and that some people on the team are moving their own vals to Cloudflare, including the Val Town blog and docs.
| Platform | Website | SQLite | Blob | Cron | Access | AI | FaaS | |
|---|---|---|---|---|---|---|---|---|
| Cloudflare | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Render | Yes | — | — | Yes | — | — | — | — |
| Vercel | Yes | — | Yes | Yes | — | Yes | Yes | Yes |
| Netlify | Yes | — | Yes | Yes | — | Yes | Yes | Yes |
| Supabase | — | Yes | Yes | Yes | — | Yes | — | Yes |
| Railway | Yes | — | Yes | Yes | — | — | — | Yes |
Checks follow Val Town's table. A dash is a feature that page left unmarked. The page says a check means the feature exists as a built-in product, and that many gaps can be filled with something else, such as Resend for email on Vercel, or SQLite on Render's persistent disks. FaaS is their column for functions.
SQLite has a named path. The same page says Turso is the underlying provider for Val Town SQLite. It also names Cloudflare D1, Cloudflare’s managed SQLite, and SQLite-backed Durable Object storage. For a host without a built-in database it points at node:sqlite, bun:sqlite, and better-sqlite3.
Blob storage on that page includes Cloudflare R2, Vercel Blob, Netlify Blobs, Supabase Storage, and Railway Buckets, plus S3 and others. Cron names include Cloudflare Cron Triggers, Render Cron Jobs, Vercel Cron Jobs, Netlify Scheduled Functions, Railway Cron Jobs, Supabase Cron, GitHub Actions, and Upstash QStash.
Email is split into receiving and sending. Receiving names include Cloudflare Email Routing, Resend Inbound, and Postmark Inbound Email, and also Mailgun and SendGrid. Sending names include Cloudflare Email Sending, Resend, and Postmark. Access, the old Val Town permissions on a val.run URL, is matched on the page to Cloudflare Access, Vercel Authentication, Supabase Auth, and Tailscale.
The page also lists a different kind of move, for people who want a chat to rebuild the app. It names Lovable, Replit, v0, ChatGPT Sites, and Claude Artifacts. It says Pete tested those five and wrote an opinionated review. The review itself is not quoted here. The host table above is the part with checkmarks.
The thing people said they would miss is the whole bundle. jborichevskiy wrote: “I’m gonna miss it. It was such a solid place for a v0.5 of something where I get hosting, a db, subdomains, logging, and more right off the bat.”
koolala went further: “We still badly need a replacement for Glitch.com. There are no free dynamic server hosts anymore.” That “no free hosts” line is their claim.
alabhyajindal, a self-described early fan, wrote that a cloud for personal applications “makes less sense as time goes on,” because the harness that helps someone build the toy can also deploy it, “like how Claude can deploy your websites.”
rpmisms asked for self-hosted alternatives they like. rubenfiszel answered with Windmill: “Not a 1:1 alternative but lots of overlap.” Their Hacker News profile says they are building Windmill. Read the suggestion as the maker of that product answering a request.
How do I migrate off Val Town?
The migration guide opens with a prompt meant to be pasted to an AI agent. The steps in that prompt are plain. Connect the Val Town plugin or MCP server so the agent can see the account. Decide which vals are worth moving. Pick a platform from the where-to-migrate page. Propose a short plan and wait for approval. The prompt also tells the agent that Deno’s core maintainers will not support Deno beyond October 2027, and to aim at Node, Bun, Cloudflare Workers, or Vercel Functions. It offers an optional HTML tracker for vals, domains, email, and leftover tasks. The full wording sits on that guide. Krouse’s letter is the short version of the same bet: current models are good at this kind of move.
Do the redirect last. The prompt says to verify the migrated app and sync its data, confirm the new URL, and only then point the old val.run address at it. Webhook URLs get updated with the provider that sends them.
The redirects page is the detail under that step. A redirect keeps the path and the query string unless you turn that off. Their example: https://bloggy.val.run/hello-world?ref=home goes to https://example.com/hello-world?ref=home. As soon as the redirect is on, the val stops receiving requests, so they say to add it after the app has actually moved. Many webhooks will not follow a redirect. Change those URLs at the service that calls you. The page covers URLs that end in .val.run. A custom domain on Val Town needs a DNS change at the domain provider.
You can set redirects in three places they name: ask Townie, use the plugin or MCP server, or edit them by hand. By hand, the HTTP section of a val’s settings has the endpoints. Account and org settings put the same list under Custom Domains, in a section called Val.run redirects. Edits stay open until 10 December 2026. That day, the redirects you have set become permanent.
Leaving a host also moves secrets. A webhook URL and a token live at the other company, the one calling the val. Rotating API keys after the Trigger.dev and GitBook breaches is that chore when a disclosure says someone may have read the key. A shutdown asks for the same change at the source: the old address is the one the calendar is retiring, and the redirects page says many webhooks will not follow a redirect.
The guide’s index also lists a GitHub guide beside the host guides. Support, if the guides stall, is the address in the letter: support@val.town, or Discord.
Will Val Town be open source?
The letter’s last row says “Open-sourcing Val Town” on 1 February 2027. It does not name a licence. It does not say which repositories, which history, or which hosted data the release includes.
That date is later than the shutdown of the service. All other services end on 10 January 2027. The code release is listed for 1 February. Until that February date, the path the company has published is the migration guides, and the models Krouse said are good at this kind of move. simonw wrote that they are looking forward to the code in February, when Val Town open-sources it.
A side project on a host you do not own
jborichevskiy’s v0.5 is a real shape. Hosting, a database, subdomains, and logs, on day one, is why a side project ends up on a small platform. Shipping a first version got cheap. The computer it runs on still belongs to a company that can publish a calendar.
A *.val.run address is a hostname Val Town controls. The redirects page says you can point it at a new URL until 10 December 2026, and that saved redirects become permanent that day, while the service itself runs toward 10 January. A custom domain is a record at your registrar. The same page says that record has to be pointed at the new host by you. A product you own is a cleaner position when the domain, the database export, and the cron all have a copy that survives the vendor.
Public stacks already tend to be short. The tools that show up in founder stories treat hosting as a leftover list, and the point of that list is knowing what you still pay for. One public version of keeping the machine is Pieter Levels running sites on a Hetzner VPS and cancelling other subscriptions around it. That ledger is his, with his prices, and it is a poor template for a project that does not already live on a server he controls.
The hard-business comments and the small-and-sustainable comment are about the same gap. simonw and cmontella think a Val-like host is a difficult company. drekipus asked why it was not kept small enough to last. plesiv said paying customers were hard to find for Gitlip, the similar product they stopped. Krouse’s letter answers with the market he says Val Town lost: cloud computing, against hosts that can publish a full row of checkmarks. koolala’s claim that the free dynamic host is gone, and alabhyajindal’s claim that the coding harness will deploy the toy itself, are the arguments for why a new Val Town may not show up to catch the next v0.5.
What do you think? If a side project of yours sits on a small host, is that an acceptable risk for a first version, and which of the Val Town alternatives would you actually move it to?
FAQ
Why is Val Town shutting down?
Steve Krouse wrote on 10 November 2026 that Val Town is winding down because the company could not compete in the cloud computing space. Billing paused that day. Existing subscriptions keep being serviced for free, and a prepaid annual plan gets a partial refund. Val Town's migration guide says vals run on Deno, that Deno is winding down, and that the core maintainers will not support Deno beyond October 2027. On 9 October 2026 Ryan Dahl wrote that the entire Deno team is joining Cloudflare, that the Deno runtime gets another year of monthly bug-fix and security releases before they end development of it, and that Deno Deploy keeps running for six months and then shuts down. On Hacker News, simonw and cmontella called a business in this space extremely hard. cmontella named raises of $2 million for Eve, $3.5 million for Darklang, and $7 million for Val Town. Those amounts are that commenter's claim. The winding-down post thanks investors, including Dan Levine, and states no figure. hardwaregeek wrote that an era of devtools companies was closing and listed acquisitions. That list is their comment. swyx replied that the moment could be ordinary startup mergers and acquisitions. drekipus asked why the project was not built small and sustainable from the start.
When does Val Town shut down?
Steve Krouse's 10 November 2026 letter gives four dates. New signups and plan upgrades ended that day. Read-only mode, which the table says does not include sqlite and blob storage, starts 10 December 2026. All other services end 10 January 2027. Open-sourcing Val Town is listed for 1 February 2027. The migration guide says code becomes read-only on 10 December 2026 and that vals keep running until 10 January 2027. Redirects of val.run URLs can be edited until 10 December 2026. On that day, redirects already set become permanent.
How do I migrate off Val Town?
Val Town's migration guide has a prompt you can paste to an AI agent. The prompt tells the agent to connect the Val Town plugin or MCP server, pick which vals to move, choose a platform from the where-to-migrate page, and propose a plan for you to approve. It says to move to Node, Bun, Cloudflare Workers, or Vercel Functions, because the guide says Deno's core maintainers will not support Deno beyond October 2027. After an app is verified and its data is synced, redirect its val.run URLs. Confirm the new URL first. Update webhook URLs with each provider, because the redirects page says many webhooks will not follow a redirect. Redirects keep the path and query string unless you turn that off. Setting one stops the val from receiving requests, so the guide says to add it after the move. Custom domains need a DNS change at the domain provider. The letter asks people who need help to write to support@val.town or to Discord.
Will Val Town be open source?
The winding-down timeline lists open-sourcing Val Town on 1 February 2027. The letter does not name a licence or which parts of the product the release will include. That date sits after all other services end on 10 January 2027. On Hacker News, simonw wrote that they are looking forward to seeing the code in February, when Val Town open-sources it.
What are the Val Town alternatives?
The 10 November 2026 letter points to migration guides for Cloudflare, Vercel, Netlify, and Render. The where-to-migrate page says Cloudflare is the most feature-equivalent platform, and that some people on the Val Town team are moving their own vals there, including the Val Town blog and docs. Its table marks Cloudflare for website, SQLite, blob, cron, email, access, AI, and functions. Vercel, Netlify, Render, Supabase, and Railway are partial. A check on that page means a built-in product. Turso is named as the underlying provider for Val Town SQLite, and Cloudflare D1 is named as Cloudflare's managed SQLite. Email names on the page include Cloudflare Email Routing, Resend, and Postmark. Cron names include the hosts' own schedulers, GitHub Actions, and Upstash QStash. The same page also names Lovable, Replit, and v0 for people moving to a chat-built app. On Hacker News, rubenfiszel suggested Windmill and called it not a one-to-one alternative. Their Hacker News profile says they are building Windmill.