Spectral Compute secures £4.6m to tackle CUDA vendor lock‑in
This article covers Spectral Compute, a London startup, which announced a £4.6m seed funding round and published further details of its SCALE framework. The development aims to reduce vendor lock-in in high-performance computing and AI workloads by enabling existing CUDA code to run on non-Nvidia GPUs without rewrites or performance loss, affecting research labs, cloud providers and enterprises.
Spectral Compute, a London startup building software to make CUDA applications portable across GPUs, has announced a seed funding round and released more detail on its SCALE framework. The technology aims to reduce vendor lock-in in high-performance computing and AI workloads by letting existing CUDA code run on alternative GPUs without rewrites or performance loss, a capability that could change procurement and resilience strategies for research labs, cloud providers and enterprises.
Why it matters
Nvidia’s CUDA stack has become the default for many AI and high-performance computing workloads. That dominance has led to supply-chain dependence, premium pricing and limited choice for organisations that need large volumes of accelerated compute. New export controls and long lead times for GPUs have intensified those risks.
Spectral’s approach seeks to give teams the ability to use their existing CUDA code on other GPUs, starting with AMD hardware, which could lower costs, reduce single-vendor exposure and make it easier to match workloads to the most appropriate chip for price, performance or availability.
Inside the product
SCALE is a software framework that translates or adapts CUDA-based code so it can run natively on non‑Nvidia GPUs. Spectral positions SCALE as different from traditional porting tools by promising fast interoperability with minimal performance degradation and without lengthy rewrites. Today the product supports the AMD chip ecosystem, with additional architectures planned.
Early users include organisations in microprocessor design and high-performance motorsports. In those contexts, portability matters because teams run compute-heavy simulations and models where turnaround time and deterministic performance are critical. Spectral says customers are already using SCALE to diversify hardware stacks and avoid delays tied to a single vendor.
Spectral was founded in 2018 by Michael Søndergaard, Chris Kitching, Nicholas Tomlinson and Francois Souchay. The team combines experience in HPC, GPU programming and compiler design across domains such as computational fluid dynamics, high-frequency trading, AI and broadcasting. The startup currently employs 19 people and plans to expand its engineering headcount to support product development and market expansion.
Founder perspective
In the announcement, Michael Søndergaard, co-founder and CEO of Spectral Compute, said:
Companies shouldn’t have to bet their entire compute infrastructure on one chipmaker. The vision of SCALE is that teams can finally use their existing CUDA code on any GPU, whether that’s Nvidia, AMD, Intel, or others, without performance loss or costly rewrites. We’re giving developers and enterprises the freedom they’ve never had before.
### Who's backing the round
Who’s backing the round
Spectral secured a £4.6m seed round (reported as $6m) led by US venture firm Costanoa. The round also included participation from Crucible and a group of angel investors.
In the announcement, Tony Liu, Partner at Costanoa, said:
Spectral has solved one of computing’s toughest challenges: letting teams write CUDA once but run it anywhere. Their breakthrough technology eliminates one of the largest bottlenecks in AI infrastructure and HPC today, unlocking choice and innovation across the compute landscape.
If you're researching potential backers in this space:
- Explore our list of AI venture capital firms in the UK
- Browse our directory of AI angel investors in the UK
The bigger picture
The startup’s proposition comes at a moment of heightened attention to supply-chain resilience and regulatory pressure on export controls for advanced chips. For public and private research institutions, financial firms and cloud providers, the ability to switch hardware vendors more easily could be a tactical response to shortages, tariffs or policy-driven restrictions.
If SCALE delivers on performance parity, the technology could influence procurement decisions across Europe and the UK, where organisations are increasingly balancing onshore capacity, cost and geopolitical risk. The UK’s AI and HPC ecosystem—spanning universities, research councils and cloud operators—stands to benefit from tools that reduce vendor dependency and make investment in heterogeneous infrastructure more viable.
Spectral’s progress will be worth watching as it moves from early pilots to wider deployment, and as competing approaches from chip vendors and software vendors play out in the market. The outcome will matter for anyone buying or operating large-scale accelerated compute in the UK and Europe.
Click here for a full list of 7,233+ startup investors in the UK