London-based Callosum secures €85.4 million Seed funding led by Atomico to connect AI models and chips
London-based AI infrastructure startup Callosum has raised €85.4 million ($100 million) in a Seed round led by Atomico. The funding will help the company develop its “heterogeneous intelligence” approach, which aims to connect different AI models and chips so they can work together more efficiently.
The round also received significant backing from Plural, DCVC, the UK Sovereign AI Fund, and other investors and angel backers. Callosum said the funding will help it build infrastructure that can route AI workloads across different types of models and chips based on factors such as speed, cost, and energy use.
The company was founded in 2025 by Jascha Achterberg and Danyal Akarca. Its latest funding follows an €8.76 million ($10.25 million) pre-Seed round announced in February 2026. Callosum has also become the first company backed by the UK Sovereign AI Fund.
“As AI moves from training to inference, the question is no longer just who builds the best models. It’s how we deliver intelligence efficiently, across an increasingly diverse landscape of models and chips,” said Niklas Zennström, CEO of Atomico.
He added, “That’s what makes Callosum so compelling. They’re building the software layer that intelligently routes AI workloads to the right compute, making AI faster, more efficient and more accessible without adding complexity.”
The Seed round comes after Callosum raised €8.76 million ($10.25 million) in a pre-Seed round in February 2026. The round was led by Plural and backed by 22 investors, with support from the UK’s Advanced Research + Invention Agency (ARIA).
Founded in 2025 by Jascha Achterberg and Danyal Akarca, Callosum believes the future of AI will depend not on a single powerful model or the best AI infrastructure, but on connecting and managing multiple AI models across different types of chips.
Along with the funding, Callosum said it has received the first investment from the UK Sovereign AI Fund. The company has also been included in the UK government’s £1.1 billion plan to strengthen AI hardware.
“AI is nothing without the chips that underpin it – and the eye-watering demand for them is only going to grow. In the race to develop and use AI, success will depend not just on having access to those chips, but on using them as efficiently as possible,” said Kanishka Narayan, Minister for Artificial Intelligence, UK Government.
Callosum says new AI models and specialised chips are developing at the same time, creating an opportunity to use different technologies together to improve AI performance while reducing costs. As part of its Seed funding, the company highlighted three major developments to support its vision of “heterogeneous intelligence.”
The first is a series of global computing partnerships. This includes a “flagship partnership” with Cerebras Systems to deliver low-latency, multi-agent AI to customers at scale. Callosum is also working with chipmaker Rebellions, along with other infrastructure providers and OEMs around the world.
The second development is what Callosum calls “programmable heterogeneity.” This technology breaks AI workloads into smaller tasks and sends each task to the model and chip best suited to handle it. The approach is designed to balance factors such as cost, energy use, and speed, helping businesses run AI more efficiently.
The third development is the launch of “Tailored Inference” for production use. The API platform is designed to improve AI performance and reduce costs, with Callosum already seeing benefits in areas such as cybersecurity and finance.
Andrew Feldman, CEO of Cerebras Systems, said, “By integrating Cerebras into Callosum’s platform, we’re making ultra-low-latency inference available exactly where it creates the greatest impact, enabling customers to build AI systems that simply weren’t practical before.”
Callosum’s founding team brings experience across neuroscience and hardware-software co-design, with team members having backgrounds at institutions and companies including Cambridge, Oxford, MIT, Imperial College London, Microsoft, ETH Zurich, and Intel. The company has since expanded its team with engineers, technical experts, and operational staff across the UK and US.
In a February 2026 blog post, Callosum said the AI industry’s traditional approach of building larger models on the same chips is becoming less effective. The company believes most real-world AI applications need different types of intelligence working together rather than relying on one dominant AI model.
Callosum also sees computing infrastructure as one of the biggest challenges facing AI companies. Its “Tailored Inference” product aims to address this by providing APIs that route AI workloads across a global network of chip and infrastructure partners. The system continuously evaluates different combinations of AI models and hardware to help customers balance cost, speed, and performance as technology changes.