StartupNews · Breaking News
BLACKSMITH RAISES $45 MILLION AS THE AGE OF AI-WRITTEN CODE CREATES A NEW BOTTLENECK: TESTING IT
San Francisco Infrastructure Startup's Valuation Rockets Nearly Tenfold in Under a Year, as Peak XV Partners Bets That Validating Code Will Matter More Than Writing It
By Startup news · Published · Updated
For the past two years, the dominant story in software development has been how fast artificial intelligence can help engineers write code. Tools like Claude Code, GitHub Copilot and OpenAI's Codex have compressed tasks that once took days into minutes, and adoption has scaled at a pace few enterprise software categories have ever matched. But a quieter, less glamorous question has been building in the background of that boom: once code is being generated faster than any human team could ever produce it by hand, who — or what — checks that it actually works?
Blacksmith, a three-year-old San Francisco startup, is betting its entire business on the idea that this second question is about to become just as important, commercially, as the first. Today the company announced it has raised $45 million in a Series B funding round led by Peak XV Partners, with existing backers Y Combinator and GV — Google's venture arm — also participating. The round values Blacksmith at $550 million, a figure that represents a nearly tenfold jump from the roughly $60 million valuation the company carried less than a year ago, when it closed a $10 million Series A round.
That kind of valuation trajectory — a 9x increase in well under twelve months — is rare even by the frothy standards of the current AI infrastructure funding cycle, and it says as much about where investors currently believe the next major bottleneck in software engineering lies as it does about Blacksmith's own execution.
QUICK-LOOK SNAPSHOT
Company : Blacksmith Headquarters : San Francisco, California Founded : 2024 (Y Combinator Winter 2024 batch) Founders : Aditya "JP" Jayaprakash (CEO), Aayush Shah, Aditya Maru Round : Series B Amount Raised : $45 million Lead Investor : Peak XV Partners Participating Investors : Y Combinator, GV (Google Ventures) New Valuation : $550 million Prior Valuation (Series A, ~11 months earlier) : ~$60 million Total Funding to Date : $58.5 million (seed $3.5M + Series A $10M + Series B $45M) Round Closed : March 2026 Round Announced : August 12, 2026 Customers : 6,000+ (up from ~800 a year earlier) Notable Customers : Supabase, Clerk, Ashby, Mercury, Expensify
WHAT BLACKSMITH ACTUALLY DOES
Blacksmith's core product is, on its face, a fairly narrow piece of developer infrastructure: a faster, cheaper drop-in replacement for GitHub Actions, the automation service most software teams already use to build, test and deploy their code every time someone proposes a change. This category — known in the industry as continuous integration, or CI — is the automated checkpoint that runs every time a developer submits new code, compiling it, running test suites against it, and flagging problems before that code ever reaches production.
Rather than running these jobs on shared, general-purpose cloud infrastructure the way GitHub's own hosted runners do, Blacksmith built dedicated, purpose-specific compute for CI workloads, paired with caching and storage systems tuned specifically to make repetitive build-and-test cycles faster. The company says the switch requires only a one-line change to an existing GitHub Actions workflow file, and it claims — based on its own internal comparisons — that its infrastructure runs roughly twice as fast as GitHub-hosted runners while costing significantly less, with the company's own marketing citing cost reductions of 60% to 75% depending on workload type. Those performance and pricing figures are Blacksmith's own claims rather than independently audited benchmarks, a caveat worth keeping in mind, though the company's rapid customer growth suggests the value proposition is resonating in the market regardless.
Beyond the core CI-acceleration product, Blacksmith has also built codesmith, a cloud-based coding agent that goes a step further than simply running tests faster — it is designed to automatically diagnose and fix CI failures, essentially closing the loop between "a test broke" and "the problem is fixed," with less direct human intervention required at each step.
THE BOTTLENECK THESIS: WHY THIS MOMENT, WHY THIS COMPANY
The single most important number in Blacksmith's pitch to investors is not its revenue or its valuation — it's the growth rate of activity on its own platform. According to the company, the number of CI jobs running on Blacksmith's infrastructure has grown between 5% and 10% every single week since the start of 2026, a compounding growth rate that, sustained over a full year, would represent an increase measured in multiples rather than percentage points.
WEEKLY CI JOB GROWTH ON BLACKSMITH'S PLATFORM (illustrative trend, Jan–Aug 2026)
Jan 2026 ████░░░░░░░░░░░░░░░░░░░░░░░░░░ Mar 2026 ██████░░░░░░░░░░░░░░░░░░░░░░░░ May 2026 ██████████░░░░░░░░░░░░░░░░░░░░ Jul 2026 ███████████████░░░░░░░░░░░░░░░ Aug 2026 ██████████████████░░░░░░░░░░░░
(Illustrative compounding curve based on the company's disclosed 5–10% weekly growth rate since January 2026; exact weekly figures are not publicly disclosed.)
Blacksmith's own explanation for this surge is straightforward: as AI coding assistants and autonomous coding agents have become genuinely capable — not just autocomplete tools suggesting the next line, but systems that can plan, write, and revise entire pull requests with limited human supervision — the volume of code being produced by any given engineering team has increased dramatically. Every one of those AI-generated changes still needs to be built, tested and validated before it can ship, which means the CI layer of the software development pipeline is absorbing far more traffic than it was even eighteen months ago, even though the number of human engineers writing that code has not grown anywhere near as fast.
Company co-founder and CEO Aditya Jayaprakash has framed this dynamic bluntly in the company's own public statements: as AI code-generation tools make writing code easier, validating that code becomes the new bottleneck. It's a thesis that reframes CI infrastructure — historically viewed as unglamorous, commoditized plumbing — as one of the more strategically interesting layers of the software stack precisely because of how AI has reshaped the economics of writing code in the first place.
FROM 800 CUSTOMERS TO 6,000 IN UNDER A YEAR
Perhaps the clearest evidence that Blacksmith's bet is paying off, at least so far, is the sheer speed of its customer growth. When the company announced its Series A round in September 2025, it counted roughly 800 to 900 paying customers. Less than a year later, that number has grown to more than 6,000 — a roughly seven-fold increase in the customer base within a single year.
BLACKSMITH CUSTOMER GROWTH
Series A (Sept 2025) ████░░░░░░░░░░░░░░░░░░░░░░░░░░ ~800 customers Series B (Aug 2026) ████████████████████████████░░ 6,000+ customers
That customer list now includes a number of well-regarded, fast-growing technology companies — Supabase, the open-source database platform; Clerk, a developer-authentication provider; Ashby, a recruiting software company; Mercury, the business banking platform; and Expensify, the expense-management company — a roster that reads less like early-adopter hobbyists and more like a genuine cross-section of mid-market and growth-stage software companies making a deliberate infrastructure choice.
Revenue has scaled alongside the customer base, though Blacksmith has been characteristically sparse with hard numbers. The company has said it crossed $10 million in annualised revenue with a team of just ten employees at an earlier stage, and that revenue has since grown into "tens of millions of dollars" annualised, achieved with a team that has grown to roughly thirty people. That is an extraordinarily lean revenue-per-employee ratio even by the standards of modern, AI-era infrastructure startups, and it is likely a meaningful part of why Peak XV Partners was willing to underwrite such a steep valuation increase in a single funding round.
WHERE THE MONEY GOES
Unlike many funding announcements that lean on vague language about "accelerating growth," Blacksmith has been unusually specific about what its new $45 million is actually earmarked for: compute. In the company's own words, posted publicly by its founders, the money is going toward building out cloud infrastructure — expanding the dedicated compute capacity that underpins its entire CI-acceleration product — because, as the company put it plainly, "that ain't cheap."
USE OF SERIES B PROCEEDS (BY STATED PRIORITY)
Compute capacity expansion ████████████████████████░░░░ Primary use Engineering hiring (NYC & SF) ██████████████░░░░░░░░░░░░░░ Secondary use Coding tool suite expansion ████████████░░░░░░░░░░░░░░░░ Product growth
Alongside expanding raw infrastructure, the company has said it intends to grow its engineering headcount, specifically hiring in both New York and San Francisco, and to broaden its product suite beyond pure CI acceleration into a wider set of coding tools intended to help developers write and test software faster — an extension of the codesmith agent concept into a more complete platform for the AI-assisted software development lifecycle, rather than a single-purpose CI runner.
THE VALUATION JUMP, IN CONTEXT
A near-tenfold valuation increase within eleven months is the kind of number that invites scrutiny, and it is worth putting Blacksmith's trajectory in the context of the broader AI infrastructure funding environment rather than treating it as an isolated data point.
BLACKSMITH'S FUNDING HISTORY
Round Date Amount Post-Money Valuation ───────────────────────────────────────────────────────────────── Seed 2024 $3.5M Undisclosed Series A September 2025 $10M ~$60M Series B March 2026 (closed)/ $550M August 2026 (announced) $45M
Total raised to date: $58.5 million
Several factors help explain the scale of the jump. First, Blacksmith's own operating metrics — customer count, CI job volume, and revenue — all appear to have grown at a similarly explosive pace over the same period, meaning the valuation increase is not purely a function of investor enthusiasm outrunning business fundamentals; the underlying usage numbers moved dramatically too. Second, the round arrives at a moment when venture investors across Silicon Valley have been actively searching for the "picks and shovels" layer of the AI coding boom — infrastructure and tooling companies that benefit from surging AI code generation regardless of which specific AI coding assistant or foundation model ultimately wins the largest market share. A CI acceleration platform that gets faster and more valuable the more AI-generated code exists in the world is, in that sense, a relatively model-agnostic bet within an AI market where picking the eventual winner among competing coding agents is notoriously difficult.
Notably, all three of Blacksmith's institutional backers — Peak XV Partners, Y Combinator, and GV — chose to participate again in this round rather than a wholly new set of investors stepping in, with the company describing its earlier backers as "tripling down." That kind of continuity, where existing investors increase their exposure rather than simply holding their position, is often read within the venture industry as a stronger signal of conviction than a valuation jump driven primarily by new, less-informed capital.
THE COMPETITIVE LANDSCAPE: A CROWDED, HIGH-STAKES CORNER OF DEVELOPER TOOLING
Blacksmith is not operating in a vacuum, and the company's own public statements acknowledge a competitive field that spans both specialised startups and the cloud computing giants. Its most direct point of comparison is GitHub Actions itself — the very product Blacksmith's core offering is designed to be a faster, cheaper alternative to — along with newer entrants like Cursor's automation features. Beyond that, Blacksmith also competes, at least partially, with the testing and CI/CD services built directly into Amazon Web Services, Microsoft Azure and Google Cloud, each of which offers its own native tools for automating software builds and tests within their respective ecosystems.
WHO BLACKSMITH COMPETES WITH
Competitor Positioning ───────────────────────────────────────────────────────────────────── GitHub Actions (native) The default, ubiquitous CI/CD product most teams already use; Blacksmith positions itself as a faster, cheaper drop-in replacement rather than a rival ecosystem Cursor Automations Automation features built into a leading AI-native code editor, competing for some of the same "validate AI-written code" workflows AWS / Azure / Google Cloud Hyperscale cloud providers offering their own native build, test and CI/CD tooling as part of broader cloud platform bundles
What differentiates Blacksmith from all of these, at least in its own framing, is specialisation: rather than being one feature within a much larger, general-purpose cloud platform, Blacksmith's entire infrastructure stack — compute provisioning, caching layers, storage systems — has been purpose-built around the specific, repetitive access patterns of CI workloads, which the company argues allows it to meaningfully outperform general-purpose alternatives on both speed and cost for that narrow but increasingly critical task.
The bigger competitive question hanging over Blacksmith, however, is less about any single rival product and more about the broader AI coding tools market it now sits adjacent to. The market for AI coding agents themselves — the tools actually generating the code that then needs to be validated — has become intensely competitive and fast-moving through 2026, with Anthropic's Claude Code, OpenAI's Codex, Cursor, GitHub Copilot and a growing list of open-source alternatives all racing to capture developer workflows. That competitive intensity, somewhat counterintuitively, benefits Blacksmith regardless of which specific agent wins: more AI-generated code from any of these tools means more CI volume flowing through validation infrastructure, and Blacksmith does not need to bet on any single coding assistant's long-term dominance to benefit from the category's overall growth.
THE BROADER TREND: TESTING AND VALIDATION AS THE NEXT AI INFRASTRUCTURE LAYER
Blacksmith's fundraise arrives as part of a wider pattern investors have been tracking across the software development tooling landscape throughout 2026: as AI coding agents have matured from simple autocomplete suggestions into genuinely autonomous systems capable of planning multi-step changes, running their own tests, and submitting complete pull requests with minimal human oversight, the software industry's traditional review and validation processes have come under real strain. Engineering teams that once had a small number of human developers generating a manageable, human-paced stream of pull requests now often have both human engineers and multiple AI agents producing code simultaneously, at a volume that can easily overwhelm manual code review and traditional testing cycles alike.
This shift has created what several developer-tooling investors have begun describing as a new, distinct layer of the AI infrastructure stack — sitting downstream of the coding agents themselves and upstream of production deployment — dedicated entirely to verifying, testing and validating machine-generated code at scale before it ships. Blacksmith's rapid ascent is, in that sense, one of the clearest early proof points that this validation layer is attracting serious institutional capital in its own right, rather than being treated as a minor feature bolted onto existing CI/CD products.
Whether Blacksmith can maintain its current growth trajectory as larger, better-resourced competitors — including the hyperscale cloud providers whose own CI/CD offerings it currently competes against — respond to the same underlying trend remains an open question. But for now, the company's numbers tell a compelling story: a sevenfold increase in customers, compounding weekly growth in platform usage, and a valuation that investors were willing to increase nearly tenfold within less than a year, all built on the wager that in an era where AI can write software almost as fast as anyone can imagine it, the ability to trust that software before it ships becomes the far more valuable — and far more defensible — business to be in.
WHO ARE THE FOUNDERS: FROM A NARROW IDEA TO A BROADER PLATFORM
Blacksmith was founded in 2024 by Aditya "JP" Jayaprakash, Aayush Shah and Aditya Maru as part of Y Combinator's Winter 2024 accelerator batch — the same start-of-life pipeline that has produced a long list of now-prominent technology companies. The founding team's original insight was narrow and specific rather than sweeping: GitHub Actions, despite being the default choice for millions of software teams, runs on general-purpose, shared cloud infrastructure that was never purpose-built for the highly repetitive, bursty, cache-friendly workload pattern that CI jobs actually represent. That gap between "good enough, ubiquitous default" and "infrastructure genuinely optimised for this one job" is a classic pattern in enterprise software — the same logic that has powered countless successful infrastructure startups before it, from specialised databases to purpose-built networking hardware.
What has changed since that founding thesis, and what the Series B round is explicitly designed to fund, is the scope of the company's ambition. Blacksmith no longer describes itself purely as a faster CI runner; its own language now positions it as building "the infrastructure to validate" the surge of AI-generated code — a noticeably broader mission than "make GitHub Actions faster," and one that opens the door to a wider suite of products beyond the original CI-acceleration offering. The codesmith agent, which automates the diagnosis and repair of CI failures rather than simply reporting them to a human, is the clearest existing example of that broader ambition already shipping in product form, and the company has signalled that further tools in this vein are part of what the new capital will fund.
A GLOSSARY FOR READERS NEW TO DEVELOPER INFRASTRUCTURE
For readers less familiar with the terminology that comes up repeatedly in stories like this one, a few definitions are worth laying out plainly:
CONTINUOUS INTEGRATION (CI) — the practice, now standard across most professional software teams, of automatically building and testing every proposed change to a codebase as soon as it is submitted, rather than waiting to test large batches of changes all at once. The goal is to catch problems early, before they compound or reach production.
GITHUB ACTIONS — GitHub's own built-in automation service, used by a large share of the world's software teams to run CI jobs, deployments and other automated workflows directly from their code repositories. It is the default choice for most teams simply because it is bundled with GitHub itself.
CI RUNNER — the actual computer (physical or virtual) that executes a CI job — compiling code, running test suites, and reporting the results. GitHub offers its own hosted runners; Blacksmith's core product is an alternative runner infrastructure designed to do the same job faster and more cheaply.
CODING AGENT — a category of AI tool that goes beyond simply suggesting code as a developer types (autocomplete) to autonomously planning, writing, testing and revising multi-step changes across a codebase, often with limited direct human supervision at each step. Examples widely used in 2026 include Claude Code, OpenAI's Codex, and GitHub Copilot's agent mode.
PULL REQUEST (PR) — a proposed set of code changes submitted for review before being merged into a project's main codebase; in modern workflows, both human engineers and AI coding agents submit pull requests, all of which typically need to pass CI checks before merging.
SERIES A / SERIES B — sequential stages of venture capital fundraising for a startup, generally representing increasing levels of maturity, revenue, and valuation; a Series B round like Blacksmith's typically follows a company that has already demonstrated meaningful product-market fit and is raising capital to scale that proven model further.
CORPORATE VENTURE ARM — an investment fund run by an operating company rather than an independent venture firm; GV, one of Blacksmith's backers, is the venture investment arm of Google's parent company Alphabet.
FREQUENTLY ASKED QUESTIONS
WHAT DOES BLACKSMITH ACTUALLY SELL? A cloud infrastructure product that runs software teams' continuous integration workloads — the automated build-and-test checks that run every time code is proposed — faster and more cheaply than GitHub's own hosted infrastructure, plus a coding agent, codesmith, that automatically diagnoses and fixes CI failures.
HOW MUCH MONEY DID THE COMPANY RAISE, AND FROM WHOM? $45 million in a Series B round led by Peak XV Partners, with existing investors Y Combinator and GV also participating. The round closed in March 2026 and was publicly announced on August 12, 2026.
WHAT IS BLACKSMITH WORTH NOW? The Series B values the company at $550 million, up from approximately $60 million at its Series A round less than a year earlier — a roughly ninefold increase.
WHY DID THE VALUATION JUMP SO MUCH IN SUCH A SHORT TIME? The company has pointed to a combination of factors: customer count grew roughly sevenfold over the same period (from about 800 to more than 6,000), CI job volume on its platform has been compounding at 5–10% weekly since the start of 2026, and annualised revenue reportedly grew from roughly $10 million to "tens of millions of dollars," all achieved with a lean team of around thirty employees.
WHO ARE BLACKSMITH'S CUSTOMERS? The company counts more than 6,000 companies as customers, including named clients Supabase, Clerk, Ashby, Mercury and Expensify — a mix of developer-tools, fintech and business-software companies.
WHAT WILL THE NEW MONEY BE USED FOR? Primarily expanding cloud compute capacity, which the company describes as its core cost driver, along with hiring engineers in New York and San Francisco and broadening its product suite beyond CI acceleration into a wider set of AI-assisted coding tools.
WHO DOES BLACKSMITH COMPETE WITH? GitHub Actions' own native infrastructure, automation features built into AI code editors like Cursor, and the native CI/CD tooling offered by major cloud providers including Amazon Web Services, Microsoft Azure and Google Cloud.
WHY INVESTORS ARE EXCITED ABOUT THE "VALIDATION LAYER"
To understand why a company solving what sounds, on paper, like a relatively unglamorous infrastructure problem could command a $550 million valuation, it helps to step back and look at how venture investors have been thinking about the AI software development stack as a whole through 2026.
Broadly, investors have come to see this stack as organised in roughly three layers. At the top sits the coding agents themselves — Claude Code, Codex, Cursor, GitHub Copilot and a rapidly expanding field of competitors — the tools that actually generate code based on natural-language instructions from a developer. This layer has attracted enormous amounts of capital and public attention, in part because it is the most visible, consumer-facing part of the AI coding revolution, and in part because it is where foundation model companies compete most directly with one another.
Beneath that sits a second layer: the infrastructure required to actually run, test, and validate whatever code the top layer produces, at the volume and speed AI generation now demands. This is where Blacksmith sits, alongside a small number of other emerging infrastructure companies focused on adjacent problems like automated code review, security scanning of AI-generated code, and deployment safety tooling.
A third layer, still earlier in its development as an investable category, involves the governance, compliance and audit tooling that larger enterprises are increasingly asking for as they deploy AI coding agents at scale — the guardrails that let a large, regulated company feel comfortable letting an AI agent submit pull requests in the first place.
Investors have increasingly argued that the middle layer — where Blacksmith sits — is unusually well-insulated from the intense, model-driven competition raging at the top layer. A venture firm betting on which specific coding agent will dominate the market five years from now is making a genuinely difficult, high-variance prediction, given how quickly the leaderboard among Claude Code, Codex, Cursor, Copilot and various open-source alternatives has continued to shuffle throughout 2026. A venture firm betting on the infrastructure that validates code regardless of which agent wrote it is making a comparatively lower-variance bet: as long as AI-generated code volume keeps growing in aggregate, demand for validation infrastructure grows with it, independent of which individual tool is currently winning developer mindshare.
That structural insulation from the "which model wins" question appears to be a meaningful part of what allowed Peak XV Partners to underwrite such an aggressive markup in this round, even in a funding environment where investors have grown noticeably more selective about paying premium valuations for AI-adjacent startups whose growth is directly tethered to a single, potentially displaceable underlying technology.
WHAT COMES NEXT FOR BLACKSMITH
With $45 million in fresh capital and a validated, rapidly scaling customer base, Blacksmith's next stretch will likely be defined by two parallel challenges. The first is a straightforward execution problem: continuing to scale compute infrastructure fast enough to keep pace with a customer base and job volume that, by the company's own account, has been compounding on a weekly basis — a scaling challenge that, if mismanaged, could just as easily produce reliability problems and customer churn as it could produce continued growth.
The second challenge is more strategic. Having proven the core thesis around CI acceleration, Blacksmith now has to decide how aggressively to expand into the broader "coding tools" category referenced in its own funding announcement — a move that would put it in more direct competition with a wider, better-funded set of companies building AI-native developer tools, rather than remaining a focused, best-in-class specialist in one narrow layer of the pipeline. That is a familiar tension for infrastructure startups at this stage of growth: broadening the product surface can unlock a larger addressable market, but it also risks diluting the deep specialisation that made the original product so effective in the first place.
For now, the market's verdict — at least among the investors who just wrote Blacksmith a $45 million check at a $550 million valuation — is that the company has more than earned the benefit of the doubt on both fronts.
PEAK XV PARTNERS: A NOTABLE LEAD INVESTOR
The choice of lead investor is itself a data point worth examining. Peak XV Partners — the firm formerly known as Sequoia Capital India & Southeast Asia before it split from Sequoia's global operations and rebranded — has, over roughly two decades of investing across India, Southeast Asia and increasingly the broader Asia-Pacific and US markets, built a portfolio spanning more than 450 companies and now manages upwards of $10 billion in committed capital across sixteen funds. While historically associated most closely with consumer internet and fintech bets across South and Southeast Asia, the firm has been steadily expanding its footprint in US-based, deep-technical infrastructure startups, and Blacksmith's Series B represents one of its more prominent recent bets in the AI developer-tooling category specifically.
Leading a round of this size, at this valuation, for a US infrastructure company is a signal of Peak XV's broader ambitions to compete directly with Silicon Valley-native venture firms for the most sought-after infrastructure deals in the current AI cycle — a category of startup that, historically, firms like Sequoia's US arm, Andreessen Horowitz and Benchmark have tended to dominate. That competitive dynamic among venture firms themselves is, in its own way, part of why valuations for standout infrastructure startups like Blacksmith have been able to climb so quickly: more large, well-capitalised firms are now actively competing for a relatively small pool of companies that can credibly claim to sit at a genuinely defensible layer of the AI stack.
THE ENGINEERING CHALLENGE BEHIND THE PITCH
It is worth spending a moment on why building faster CI infrastructure is a genuinely difficult engineering problem, rather than simply a matter of throwing more raw compute at the task — because that difficulty is central to why Blacksmith's specialised approach has proven defensible against the much larger cloud providers it competes with.
CI workloads have a distinctive, repetitive shape: the same codebase gets rebuilt and retested, over and over, often dozens or hundreds of times a day across a large engineering organisation, with each successive run typically differing from the last by only a small set of file changes. That pattern creates enormous opportunity for caching — reusing previously computed build artefacts and dependency installations rather than redoing that work from scratch on every single run — but only if the underlying infrastructure is specifically designed to recognise and exploit that repetition. General-purpose cloud compute, of the kind GitHub's own hosted runners and the major cloud providers offer, is built to handle an enormous diversity of workload types well, which by definition means it is not deeply optimised for any single one of them, including CI.
Blacksmith's pitch is that by building infrastructure exclusively around this one workload pattern — with caching, storage and compute provisioning all tuned specifically for the way CI jobs behave — it can extract meaningfully better performance and lower cost than infrastructure that has to remain flexible enough to serve every other kind of cloud computing task simultaneously. This is a well-worn playbook in enterprise infrastructure: specialised systems built around a single, well-understood workload pattern have repeatedly managed to outcompete general-purpose alternatives on that specific task, even when the general-purpose competitor has vastly greater overall scale and engineering resources. Whether Blacksmith can sustain that specialisation advantage as AWS, Microsoft and Google inevitably respond by investing more heavily in their own CI-specific optimisations is one of the more interesting open questions for the company's next phase of growth.
KEY FACTS AT A GLANCE
• Company: Blacksmith, founded 2024, San Francisco • Round: $45 million Series B, led by Peak XV Partners • Participating investors: Y Combinator, GV • New valuation: $550 million (up from ~$60 million at Series A) • Total funding to date: $58.5 million • Round closed: March 2026; publicly announced: August 12, 2026 • Customers: 6,000+, up from ~800 a year earlier • Core product: A faster, cheaper alternative to GitHub Actions, plus codesmith, a cloud coding agent that diagnoses and fixes CI failures • Stated use of funds: Expanding compute infrastructure, hiring engineers in New York and San Francisco, and broadening its coding tools suite
THE BOTTOM LINE
Blacksmith's Series B is a useful data point for anyone trying to understand where venture capital believes the AI software boom goes next. The easy, obvious bet over the past two years has been on the tools that write code faster. Blacksmith, and the investors now backing it at a $550 million valuation, are making a less obvious but arguably more durable wager: that the real, lasting value in an AI-saturated software industry won't only accrue to whichever coding agent writes the most impressive pull request, but also to the infrastructure quietly making sure all of that AI-generated code actually works before it reaches anyone's customers. It is, in effect, a bet on plumbing — unglamorous, largely invisible to end users, and, if Blacksmith's growth curve holds, increasingly indispensable.
A NOTE ON SOURCING
Figures in this article — including customer counts, funding amounts, valuation, growth rates and use-of-proceeds details — are drawn from Blacksmith's own public announcement and founder statements, corroborated by independent reporting from technology and business news outlets covering the round. Where a figure represents a company claim rather than an independently audited metric, such as performance benchmarks against GitHub-hosted runners, this has been noted explicitly in the text above rather than presented as verified fact.
DISCLAIMER: This article is intended for general informational and news purposes only and does not constitute investment advice. Performance and cost-savings figures attributed to Blacksmith's platform (such as claims of running workloads "twice as fast" or "cheaper" than GitHub-hosted runners) are based on the company's own comparisons and have not been independently verified. Readers evaluating developer infrastructure vendors should conduct their own benchmarking before making purchasing decisions.