London-based Callosum has raised $100M in seed funding to scale its software platform that automatically matches AI workloads to optimal models and hardware. The round was led by Atomico and the UK Sovereign AI Fund.
Callosum's platform solves a core infrastructure problem in enterprise AI: determining which machine learning model and computing chip combination best fits a specific task. The software aims to reduce costs and complexity as organizations deploy multiple AI systems across different hardware configurations.
The funding round included participation from Atomico, the UK Sovereign AI Fund, and other investors. The capital infusion positions Callosum to expand its team and accelerate product development as demand for AI infrastructure optimization grows.
The startup addresses a gap in the AI stack. As companies integrate various large language models, specialized AI models, and different processors—from GPUs to custom chips—they face increasing operational overhead. Callosum's matching software automates this allocation process, potentially saving significant computational resources and operational costs.
The UK Sovereign AI Fund's participation underscores Britain's strategic interest in building domestic AI infrastructure capabilities. The fund, backed by the government, invests in companies developing critical AI technologies for national resilience.
Callosum emerges in a competitive landscape where other firms are building AI infrastructure layers, including platforms for model serving, monitoring, and optimization. The company's focused approach on task-to-resource matching differentiates it from broader infrastructure plays.
The $100M seed round signals strong investor confidence in infrastructure-layer solutions. As AI adoption accelerates across enterprises, the operational complexity of managing multiple models and hardware types is driving demand for orchestration and optimization tools.
Callosum's timing aligns with growing enterprise maturity around AI deployment. Early-stage AI projects often use single models on standardized hardware. As deployments scale, organizations increasingly need sophisticated tools to manage model selection and hardware allocation across diverse workloads.
Travis Kalanick has renewed his criticism of venture capitalists, claiming only 1% are genuinely helpful. The statement comes after securing $1.7 billion for his robotics company Atoms.
A new startup is addressing a critical market inefficiency: despite hundreds of billions in annual AI spending, there's no standardized way to price compute or hedge against cost fluctuations.
Also, the autonomous driving startup spun out from Rivian, secured $150 million in Series D funding led by Prysm Capital. The round follows a $200 million funding announcement in March 2026.
Also, a Rivian spinout, has secured $150 million in Series B funding led by Prysm Capital. The capital will accelerate the company's expansion from electric bikes into autonomous delivery vehicles.