Doubleword, the leading self-hosted inference platform for enterprises, today announced a £9m Series A led by Dawn Capital.
The UK-based company - formerly TitanML - has already scaled into the US and secured partnerships with Snowflake and Dataiku. Doubleword also counts K5 Tokyo Black as an investor alongside leading AI entrepreneurs as angel investors, including Hugging Face CEO Clément Delangue and Dataiku CEO Florian Douetteau.
Doubleword is solving one of the biggest barriers to large-scale enterprise AI adoption: self-hosted inference. Inference is where AI delivers real-world value — from answering questions to generating images, it transforms models into business outcomes. As AI adoption grows, inference has become mission-critical — a capability enterprises must own and control — driving a shift toward self-hosted inference.
Self-hosting, however, brings with it the enormous task of building and maintaining performant, scalable inference infrastructure. As a result, many teams find themselves trapped in an ongoing cycle of assembling tools, hunting for specialised talent, and continually updating their infrastructure to keep pace. This is where Doubleword comes in.
Doubleword is purpose-built for enterprises. Its end-to-end solution enables enterprises to self-host AI models—open-source, proprietary, or fine-tuned—without having to build, maintain, or optimise complex infrastructure. By working with Doubleword, enterprises get a production-ready, future-proofed self-hosted inference platform, empowering them to:
• Deploy 1 or 1,000 models in their environment effortlessly, with one-click private deployment of any open-source, domain-specific, or custom model
• Manage all their deployments in one place, acting as the unified control layer across cloud and on-prem deployments with analytics, logs and alerts built in
• Scale AI on their terms, with proprietary inference optimisation research that optimises GPU usage, and dynamic infrastructure that is built to flex with their roadmap, not dictate it
The company was one of the earliest movers in the newly formed inference space, founded pre-ChatGPT by Meryem Arik (CEO), Dr Jamie Dborin (CSO), and Dr Fergus Finn (CTO). During their postdoctoral research at UCL, Jamie and Fergus realised techniques used in their quantum machine learning model compression were also well placed to improve AI inference performance. Since then, Meryem has established herself and the company as a key player in enterprise AI.