May 2025 Summaries

Blacksmith explains how its serverless CI platform uses multitenancy to make bursty, short-lived CI workloads economically viable by pooling capacity across customers rather than requiring each customer to provision for peak demand. Individual customers may need thousands of vCPUs during code pushes but have long idle periods, while aggregated demand from many customers smooths into a more predictable pattern, improving fleet utilization and lowering the cost per job. The company operates a fixed-cost fleet of bare-metal machines that launch temporary Firecracker microVMs, making average utilization the primary driver of gross margins; it reports that margins rise sharply as utilization increases. Usage patterns also vary by weekday, time zone, and region, with European and Asian customers helping fill lower-demand periods in a fleet largely used by US customers. Blacksmith has expanded from a European region to a US region to meet latency and compliance needs, and argues that adding customers improves efficiency, supports lower prices, and strengthens profitability through the economics of multitenant infrastructure.

Blacksmith announced a $3.5 million seed round led by Google Ventures and Y Combinator, with participation from several technology founders and executives, to expand its continuous integration infrastructure platform. The company says it serves more than 600 organizations and provides a drop-in alternative to GitHub Actions runners that it claims is faster and less expensive. Founded by engineers with backgrounds at Cockroach Labs, Faire, and Superblocks, Blacksmith aims to build a specialized CI cloud combining compute, observability, and security for short-lived, highly variable workloads. It argues that self-hosting CI on major cloud platforms can require costly overprovisioning and involve slow VM provisioning and storage configurations ill-suited to ephemeral jobs, so it operates its own bare-metal fleet using high-single-core-performance consumer CPUs and local storage. The funding will support hiring and further development of a roadmap that begins with fast compute, adds tools to identify performance regressions and flaky tests, and ultimately seeks to improve software supply-chain security.