UK AI startup Callosum raises $100M to rethink how AI uses models and chips, backed by Sovereign AI Fund

Callosum, a London AI startup founded last year, has raised $100 million in seed funding to tackle one of the industry’s increasingly expensive problems: deciding which AI model should run on which chip.

The company announced Thursday it has raised a $100 million (£73.5 million) seed round led by Atomico, with significant participation from Plural, DCVC, and the UK Sovereign AI Fund. The deal is one of the largest seed rounds ever raised by a UK startup and marks the first disclosed investment from the British government’s £500 million Sovereign AI Fund.

“Today we announce our $100M seed round to redefine how humanity computes for AI’s next chapter. The round is led by Atomico with significant participation from Plural, DCVC, and the UK Sovereign AI Fund, alongside other world-class investors and angels,” the startup said in a blog post.

Founded in 2025 by Cambridge University neuroscientists Danyal Akarca, Callosum’s CEO, and Jascha Achterberg, its CTO, the company is betting that the next big AI infrastructure problem is less about finding a single superior model or chip and more about choosing the right combination for each job.

That question is becoming more pressing as companies gain access to a growing mix of AI models and specialized processors, each carrying different trade-offs in speed, cost, and capability.

Callosum is building systems software that routes AI workloads to models and chips suited to each task. The goal is straightforward: get the required result without automatically throwing the most expensive GPU or largest model at every problem.

“The defining question of the next decade will not be “which model is better?” but “which combination of models, on which silicon, together solve my problem under my constraints of cost and speed?”” Callosum said on its blog.

“This is heterogeneous intelligence: treating AGI [artificial general intelligence] not as one solution but as a space of capabilities matched to the problem at hand. It is how nature evolves intelligence, and we have already shown it to be the superior approach for many of the most important problems we want AI to solve. The advantage compounds as the explosion accelerates.”

AI’s Next Bottleneck May Be Efficiency, Not Intelligence

Callosum is arriving at a moment when the AI industry is spending staggering sums on compute. Nvidia’s GPUs remain central to much of the AI boom, yet startups and chipmakers are racing to offer alternatives optimized for inference, training, and specialized workloads.

Callosum’s thesis challenges a costly assumption buried inside that race: that better AI necessarily requires ever more expensive hardware.

Instead, the company wants software to decide what level of intelligence and compute a task actually needs. A simple request might run on a cheaper model and processor, while a difficult reasoning problem could be routed to something far more capable. At scale, those decisions could have major implications for AI infrastructure costs.

That approach helps explain the UK government’s interest.

The Sovereign AI Fund launched in April with £500 million to invest in British AI companies, support jobs, and strengthen the country’s position in AI infrastructure. Callosum is its first disclosed investment.

“AI is nothing without the chips that underpin it – and the eye-watering demand for them is only going to grow,” said Kanishka Narayan, the UK minister for artificial intelligence.

“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.”

Callosum has announced a partnership with AI chipmaker Cerebras Systems and already has deals involving Rebellions and Axelera AI, giving its software access to a broader mix of AI hardware.

The timing is striking. British AI chip startup Fractile is reportedly seeking roughly $600 million at a $6.5 billion pre-money valuation, and US chip startup Etched recently raised $700 million at a $21 billion valuation.

Those deals are centered on building more AI compute. Callosum is making a different bet: the next fortune may be made by deciding how to use all that compute more intelligently.