callosum.comCrawled Oct 24, 2026
Callosum - Co-evolving chips & intelligence
Callosum secures a **$100 million seed round** to advance AI computing infrastructure. The funding is led by Atomico, with participation from Plural, DCVC, and the UK Sovereign AI Fund. The round includes contributions from additional top-tier investors and angels.
Callosum designs high-performance computing systems by reengineering architecture from first principles. The company demonstrates how this approach enables large-scale AI workloads to achieve leading performance while improving efficiency. This paradigm extends beyond AI to domains like scientific computing and other high-performance applications.
Callosum explores how breaking down intelligence problems into components reveals heterogeneity as a key driver for innovation. The approach mirrors evolutionary processes, enabling new capabilities, efficiency gains, and solutions that were previously unattainable. The company focuses on making this heterogeneity programmable to unlock advanced problem-solving methods.
Callosum applies *Tailored Inference* to optimize AI workloads by combining customizable models with specialized silicon, tailored to specific computational demands. The approach targets partners’ most challenging efficiency problems by dynamically adjusting model composition and hardware allocation. This method aims to improve performance for tasks requiring precise, workload-specific optimization.
Callosum secures seed funding to advance its work on co-designing chips and AI systems. The startup focuses on integrating hardware and software to optimize performance for next-generation intelligent applications. This funding supports research into specialized architectures that adapt to evolving computational demands.
Callosum bets on scaling AI by deploying larger models, identical hardware chips, and vast datasets. The company designs its models, inference systems, and compute infrastructure around this scaling approach, treating it as the sole viable method for AI advancement.
Callosum scales heterogeneous compute to build AI infrastructure systems that outperform homogeneous alternatives in speed, cost, and capability. The company targets a new era of AI hardware by combining diverse processing units for optimized performance. This approach aims to address limitations in traditional, uniform computing stacks.