Callosum raises $10.25 million to unlock new AI systems that run on any chip, inspired by the brain
Founded by Cambridge scientists, Callosum is building systems-level software to make diverse chip architectures work together, breaking AI’s dependence on any single hardware provider
Inspired by how the brain uses variety, the approach already outperforms single-type chip systems by orders of magnitude
Funding will expand the London-based team, scale its software technologies and accelerate the pathway to future sovereign AI infrastructure
London, 26th February 2026: Callosum – the AI infrastructure company transforming how AI models are combined and run to solve hard real-world problems – has raised $10.25 million in funding led by Plural, supported by the UK’s Advanced Research and Invention Agency (ARIA). Angel investors, including Charlie Songhurst (Meta), Stan Boland (FiveAI), and John Lazar (Royal Academy of Engineering), also participated in the round. The funding will be used to expand the team, scale its software and source compute resources to accelerate development of sovereign AI infrastructure.
Callosum was founded by Danyal Akarca and Jascha Achterberg, who met during their PhDs at Cambridge, working on the intersection of the brain, computing and AI. Their world-class research has been published in several Nature journals, and they have worked at the likes of Intel and collaborated with Google DeepMind. The company is built on the core belief that, like the brain, the future of AI won’t lie in running bigger models on more of the same chip, but with different models optimised as a system, running across different chip types.
Challenging AI’s monoculture
Until now, AI development has assumed that scaling a single model on identical chips will deliver ever-greater intelligence. This requires extraordinary amounts of energy and capital and locks AI control in the hands of a small group, creating a trillion-dollar monoculture revolving around OpenAI and NVIDIA, with NVIDIA alone currently controlling ~85% of the AI GPU market and being the first company to reach a $5 trillion market cap. This monoculture has fostered the belief that superintelligence will emerge from a single “God-like” model.
But real-world problems are heterogeneous, complex and require different capabilities. In nature, intelligence doesn’t emerge from endlessly repeating the same component: the human brain didn’t develop by copying one neuron billions of times. It evolved to be efficient, flexible and resilient by combining many different types of cells, signals and specialised circuits. Together, this makes intelligence possible.
Diverse AI model systems that dynamically adapt and collaborate
Inspired by this principle, Callosum is building systems-level software that enables many different models to work together across diverse chip architectures. Rather than forcing every task onto identical hardware, its platform orchestrates components into a system that interacts, communicates and collaborates, reducing dependence on any single provider.
This system works by extracting performance benefits from different chip architectures and optimising each one for the task at hand. The benefits of this become clear in complex real-world tasks, such as autonomous computer use, which requires vast amounts of information and many different types of decisions. For problems like this, Callosum’s system already delivers 2x more accuracy, 7x faster performance and 4x lower cost relative to monoculture solutions. This makes it possible and economically viable to solve hard, real-world problems at scale.
Multi-cloud, multi-chip, global ambition
The impact of Callosum’s technology extends far beyond better benchmark scores. Its architecture scales differently by intelligently combining diverse hardware, fundamentally changing how large-scale compute can be built. This also creates an accelerated pathway for emerging chip technologies to reach the predicted $250+ billion market, challenging incumbents who have locked out competition by optimising the stack for their own hardware. As well, it opens the door to entirely new classes of AI, enabling model architectures that would be impossible to run efficiently on today’s homogeneous systems.
This flexibility has implications for national infrastructure, too. Callosum already works with multiple cloud partners, including AWS, Google and Microsoft and recently secured support from ARIA to unlock new capabilities in data centres with mixed-chip infrastructure, contributing to UK AI sovereignty while maintaining a global, multi-cloud approach. The company is targeting two customer segments: companies building multi-agent AI systems that need superior performance and new chip manufacturers seeking to demonstrate their hardware’s capabilities at scale.
Danyal Akarca, co-founder of Callosum, said: “Big labs are currently betting that one model will rule them all. We think that’s wrong and our work proves this. Nature shows that real intelligence emerges from many systems working together. We’ve brought together incredible talent to enable a paradigm shift in how we build intelligent systems to solve real-world problems, with the infrastructure to make that possible, on any chip, anywhere in the world.”
Jascha Achterberg, co-founder of Callosum, said: “Everyone assumed chip diversity was a disadvantage to be managed. We saw the opposite, that it’s an advantage to be exploited. We’re not optimising one algorithm on top of the existing stack. We’re using software to control all the levers across the entire system, extracting benefits from diversity that others dismiss. Plural understands this mission and we’re excited to build alongside them.”
Ian Hogarth, partner at Plural, said: “Danyal and Jascha have been published in several Nature journals and worked with some of the most prestigious AI and compute organisations in the world. Now they’ve built an incredible team to define and commercialise entirely new ground. Their vision for a multi-model, multi-chip future could be transformative and positions them to compete with the world’s biggest chip and model makers. These are serious founders tackling a serious mission, which is exactly what we look for at Plural.”
ENDS
Contact: antonella@burlington.cc
About Callosum:
Callosum is an AI infrastructure company building systems-level software that makes diverse chip architectures work together. Founded by Cambridge scientists Danyal Akarca and Jascha Achterberg, the company’s technology unlocks new AI systems through next-generation infrastructure, allowing the usage of any combination of chips. This delivers superior performance, lower costs and reduced dependence on any single hardware provider. Callosum is backed by Plural and is based in London. https://www.callosum.com/
About Plural:
Plural is an early-stage investment fund that backs the most ambitious founders on a mission to change the world through technology. Plural launched in June 2022 with the aim to give serious founders in Europe investors with experience to match their ambition. Based in Tallinn, Estonia, and London, UK, Plural’s mission is to have GDP-level impact on Europe, address systemic risks and reduce the opportunity gap worldwide through the companies it backs. https://pluralplatform.com