Clockwork.io, a Palo Alto, California-based AI infrastructure software company, has raised $31 million in new funding co-led by Premji Invest, Wing Venture Capital and Seligman Ventures. Existing investors NEA and e& Capital also participated in the financing, which brings the company’s total funding to $73 million.
Clockwork said it plans to use the capital to accelerate the rollout of its fault-tolerance software across AI training, inference and reinforcement learning workloads, expand enterprise adoption, and scale delivery through cloud partners.
Founded in 2018 by Yilong Geng, Deepak Merugu and Stanford professor Balaji Prabhakar, Clockwork develops software for improving the reliability and utilization of large GPU clusters. Its technology operates as a software layer between AI workloads and underlying infrastructure, providing observability, fault tolerance and traffic optimization across GPU environments.
Its fault-tolerance products include LinkPass, which reroutes traffic around failed network links, and TorchPass, which can move workloads from failing GPUs to healthy ones without requiring a full job restart. Alongside the funding announcement, Clockwork introduced multi-node platform snapshots and asynchronous application checkpoints designed to preserve the state of distributed AI workloads and support faster recovery.
Clockwork also announced production deployments at LinkedIn and Together AI, while existing customer WhiteFiber is expanding its use of the company’s software. LinkedIn said its deployment of LinkPass prevents tens of thousands of GPU-hours of downtime each month across its AI infrastructure fleet.
Clockwork previously announced a $21 million Series A led by NEA in March 2022.
