London AI Startup Callosum Raises $100 Million to Rethink How AI Uses Computing Power
A London startup is taking a different approach to one of the biggest challenges facing the artificial intelligence industry: how to use increasingly diverse computing hardware efficiently.
Callosum has raised $100 million (€85.4 million) in a seed funding round led by Atomico, with participation from Plural, DCVC, the UK Sovereign AI Fund and other investors and angels.
The company is developing software that allows AI applications to work across different models and types of computing hardware rather than being tied to one model or processor.
For an industry dealing with enormous demand for computing power, Callosum believes that flexibility could become increasingly important.
A Different Approach to AI Infrastructure
Much of today's AI infrastructure is built around large models running on powerful, standardized computing systems.
Callosum is taking a more flexible approach.
Its technology is designed to break an AI workload into individual tasks and determine which model and processor are best suited to handle each one.
The decision can take factors such as speed, cost and energy consumption into account.
In practical terms, this could mean that a complex AI application doesn't have to rely on one model or one type of chip for everything.
Instead, different parts of the workload could be directed to different systems.
Why This Matters as AI Gets More Expensive
The rapid expansion of AI has created an enormous demand for computing infrastructure.
Training and running advanced models can require significant amounts of processing power, while AI applications increasingly need to operate at scale.
At the same time, the hardware market is becoming more diverse.
NVIDIA remains a dominant force, but other semiconductor companies are developing processors designed for particular AI workloads. Companies such as Cerebras and Rebellions are pursuing different approaches to AI computing.
Callosum's software is intended to make it easier for businesses to take advantage of this growing range of hardware.
Rather than asking companies to choose one computing architecture, the platform is designed to coordinate different systems.
Backing From Major Investors
The size of the funding round is notable for a company at the seed stage.
Atomico led the $100 million round, with Plural, DCVC and the UK Sovereign AI Fund among the participating investors.
The investment reflects growing investor interest in companies working on the infrastructure beneath the AI applications that consumers see.
As AI adoption expands, investors are increasingly looking beyond model developers to the companies building the software, hardware and infrastructure required to operate those models efficiently.
Building a More Flexible AI Stack
Callosum describes its approach as heterogeneous intelligence.
The basic idea is straightforward: different AI models and chips can be good at different things.
Instead of treating one model or processor as the answer to every problem, Callosum wants software to determine which combination works best for a particular task.
The company has developed technology that can divide workloads into smaller components and direct them to suitable models and hardware.
That could potentially help businesses reduce computing costs while improving performance.
New Partnerships With Chip Companies
Alongside its funding announcement, Callosum has announced partnerships across the AI computing ecosystem.
One of the company's key partnerships is with Cerebras, a U.S. company known for developing large-scale AI processors.
Callosum is also working with next-generation semiconductor companies including Rebellions, as well as infrastructure providers and OEMs.
The partnerships are important because Callosum's model depends on having access to a broad range of computing options.
The more hardware platforms its technology can support, the more choices customers could have when deciding how to run their AI workloads.
The UK Is Betting on AI Infrastructure
The funding also comes as the United Kingdom is trying to strengthen its position in AI and semiconductor technology.
Callosum was previously the first company to receive investment from the UK Sovereign AI Fund, according to Sifted. The startup has also been included in the UK's broader plans to support domestic AI hardware capabilities.
That gives Callosum significance beyond its own fundraising.
For the UK, startups developing infrastructure around AI could help build a technology ecosystem that is less dependent on a small number of overseas providers.
What Callosum Plans to Do With the Money
The company plans to use the new funding to expand its infrastructure platform, develop additional partnerships and accelerate its Tailored Inference technology.
The platform is designed to provide task-specific AI capabilities through APIs while operating across different computing environments.
Callosum is also expected to expand its relationships with semiconductor and infrastructure companies as it builds out the platform.
A Growing Opportunity for AI Infrastructure Startups
The funding round points to a broader trend in the AI startup market.
The first phase of the AI boom was dominated by companies developing foundation models and consumer-facing applications.
The next phase could place greater emphasis on infrastructure.
Businesses want AI systems that are not only capable, but also affordable and efficient to operate.
That creates opportunities for companies that can help organizations make better use of available computing resources.
Callosum is targeting precisely that layer of the market.
The Road Ahead
The company's challenge now is turning its technology into a widely adopted infrastructure layer.
Supporting multiple models and processors is technically complex, and customers will ultimately need to see clear benefits in terms of performance, cost or energy use.
But the timing is favorable.
AI workloads are becoming more varied, while the computing market is becoming more fragmented. If those trends continue, software capable of managing different models and hardware could become increasingly valuable.
Key Takeaways
$100 million: Callosum's latest seed funding round.
€85.4 million: Approximate euro value of the financing.
Lead investor: Atomico.
Other backers: Plural, DCVC and the UK Sovereign AI Fund, among others.
Headquarters: London, United Kingdom.
Core technology: Software designed to coordinate AI workloads across different models and computing hardware.
Key partnerships: Cerebras, Rebellions and other infrastructure and hardware providers.
Focus: Improving AI performance, cost efficiency and hardware utilization.
The Bottom Line
Callosum's $100 million funding round highlights a growing problem in AI: having more computing power is not enough if that power cannot be used efficiently.
The London startup is building software designed to make AI workloads more flexible by allowing different models and processors to work together.
As AI systems become more specialized and the semiconductor market becomes increasingly diverse, that approach could become an important part of the infrastructure supporting the next generation of artificial intelligence.
For Callosum, the next challenge is turning its ambitious technology into a platform that companies around the world rely on to run AI more efficiently.
