Clockwork Systems Inc. has secured $31 million in a funding round co-led by Seligman Ventures, Wing Ventures, and Premji Invest. Existing investors New Enterprise Associates and e& Capital also participated. This investment brings the startup's total funding to $73 million.
The company provides a software layer designed to improve the efficiency of AI chip clusters by managing hardware failures. According to the startup, recovering from failures in large GPU clusters can take up to 90 minutes, during which compute resources remain idle. Clockwork's software provides telemetry to identify failures before a full restart is required.
Clockwork has introduced a new feature called TorchSnap, which captures multinode snapshots of distributed AI inference workloads. This allows jobs to resume from the point of failure without requiring code modifications. TorchSnap joins the company's existing tools, LinkPass for network failover and TorchPass for GPU migration.
The company's technology is currently used by organizations including LinkedIn, Together AI, and WhiteFiber. LinkedIn utilizes LinkPass to reroute AI traffic during hardware issues, while Together AI offers TorchPass as a service. Clockwork reports rapid adoption of its software across public cloud providers and enterprise fleets since its previous funding round.
