Skip navigation
Blacksmith (W24) is building a faster cloud for continuous integration (CI), helping developers test…
0:00 / 9:12

Blacksmith (W24) is building a faster cloud for continuous integration (CI), helping developers test…

Y Combinator

@ycombinator1.7M followers

Visit on X
259 likesDownloadOriginal post

39,651 views

Blacksmith (W24) is building a faster cloud for continuous integration (CI), helping developers test and ship code as AI coding tools generate more of it. As software production accelerates, testing and validation are becoming the new bottleneck. That shift has helped Blacksmith grow 20x over the last year, and the company recently raised a $66 million Series C at a $1 billion valuation. The founders, Aditya Jayaprakash, Aayush Shah, and Aditya Maru, started as outsiders to DevOps and built their first infrastructure using the same high-performance CPUs found in gaming PCs, renting servers originally designed for Minecraft and Call of Duty. Today, they run a much larger fleet with a team of just 30 people, helping companies cut CI waits from minutes to seconds. They share how they found the idea, won their first 200 customers, and why faster testing will become even more important as agents write more of the world’s code. 00:00 — Making CI Faster With Gaming CPUs 03:46 — Winning Over Investors and Developers 06:46 — Scaling CI for the AI Coding Era

24 Comments

Comments from the original post on X (Twitter)

  • chrisreedbates's profile picture

    @chrisreedbates

    Curious if CI will continue to be a thing going forward?

    2 likes

  • Loki's profile picture

    @lokidotdev

    "testing is the new bottleneck" is so real. agents write the code in 2 minutes and then everyone waits 20 for CI

    2 likes

  • Bodi M's profile picture

    @bodimog

    Very cool.

    1 like

  • Bijan | AI for Law Firms | Investor's profile picture

    @Bijan_Automates

    As coding agents spit out more diffs the bottleneck moves to tests Cutting CI from minutes to seconds is the unlock for agent loops Would you pay more for faster green builds than for a smarter model right now

    1 like

  • Jatin Garg's profile picture

    @jatingargiitk

    CI in seconds gives agents a chance to fix a failing change before they build more code on top of it.

  • Jon Cheney's profile picture

    @cheneypiano

    I'm already seeing this. The code shows up faster than anyone can check it, so testing becomes the real bottleneck.

  • Sagiv Ofek's profile picture

    @sagivo

    At what point does CI in seconds stop mattering for me if the agent keeps writing the next change while the last one is still being checked? I want the agent to wait for the result before it builds on top of a failing change.

  • Chen's profile picture

    @chenzhangreal

    testing being the new bottleneck tracks, ai spits out code way faster than we can verify it works

  • Joakim William Hauge's profile picture

    @Monetisedev

    Reviews and validation are the headline issues now. And as agents move from writing code to executing entire workflows, the next question is whether all that autonomous work is actually worth the cost.

  • Mahone's profile picture

    @Mahone_AI

    makes sense. agents write the code in 2 minutes and then everyone sits around for 20 waiting on ci

  • David H. Wever's profile picture

    @DavidWever

    Shorter test runs make it easier for an agent to work in small steps. When something breaks, there's less new code to untangle.

  • Christos Zoukos's profile picture

    @cz4000

    from Minecraft servers to unicorn, love it!

  • Niraj | AmanERP's profile picture

    @nirajkvinit

    Code review used to be the bottleneck, then CI (My biggest Infra spend), and soon it'll be the poor human reading the release notes. 😂😂 What do you think breaks next once tests are basically instant?

  • David Arnal's profile picture

    @davidarngar

    Using gaming-grade CPUs for early infrastructure is a clever starting point. It’s a good reminder that infrastructure innovation doesn’t always require purpose-built hardware.

  • Studio1HQ's profile picture

    @Studio1HQ

    we know a thing or two about helping dev tools find their developers 👀

  • Ed Giansante's profile picture

    @edugiansante

    we saw this exact pattern at persona haha , the faster code gets written, the more testing becomes the thing people actually complain about at dinners

  • Ayush Rawat's profile picture

    @iamayushrawat

    AI made writing code cheap. Now the real bottleneck is trusting what shipped.

  • Jeff Bruchado's profile picture

    @jeffbruchado

    if your suite is mostly single-threaded, a gaming cpu can beat a server with way more cores, because the wait comes down to per-core speed

  • Gaurav Bhowmick's profile picture

    @gaurav0bhowmick

    faster CI helps. the harder problem is tests that actually check behavior, because agents are very good at writing tests that just pass

  • Aapakari's profile picture

    @aapakari

    Were the first 200 customers indie devs or enterprise teams?