Callosum Technologies Secures $100M Seed Funding for AI Workload Optimization

Callosum Technologies, a London-based startup founded by neuroscientists Danyal Akarca and Jascha Achterberg, has successfully raised over $110 million in funding to enhance AI workload orchestration. The company emerged from stealth mode on February 26, 2026, with a significant $10.25 million in pre-seed funding, followed by a remarkable $100 million seed round in August 2026, led by Atomico and supported by other notable investors including Plural, DCVC, and the UK’s Sovereign AI Fund.

The startup’s software aims to optimize AI inference by breaking down workloads into subtasks and assigning them to the most suitable combination of chips and models. This innovative approach is designed to improve accuracy by up to 2x, increase speed by 7x, and reduce costs by 4x compared to traditional homogeneous GPU setups.

In a strategic partnership, Callosum has teamed up with Cerebras Systems to tackle the complexities of enterprise AI, particularly in financial services. By integrating Cerebras’ wafer-scale AI accelerators into its platform, Callosum aims to enhance performance, achieving a reported 4x improvement, a 70% reduction in compute costs, and a 10% increase in task success rates.

As AI inference becomes increasingly critical, with estimates suggesting that two-thirds of AI compute will focus on inference by 2026, Callosum’s technology is positioned to support ultra-low-latency tasks across various cloud platforms, making it particularly valuable for complex, multi-step AI workflows.