Hosted.ai raises $19M Seed round to help neocloud service providers utilise wasted GPU capacity
Hosted.ai has raised a $19 million Seed round led by Creandum, with additional investment from People Ventures, and Repeat VC. The funding will be used to roll out Hosted.ai’s GPUaaS technology, which allows GPU cloud operators (known as ‘neoclouds’) to run multiple workloads on the same GPU (graphics processing unit) simultaneously, achieving up to five times the output from their existing infrastructure.
The AI buildout frenzy is the largest and fastest infrastructure project in world history, yet the way GPU compute – the underlying framework that enables the AI boom – is bought, sold and operated today is stuck in the past. Businesses are still forced to rent whole servers, even if their customers only need a fraction of the capacity. This has a particularly punishing impact for neocloud operators, who have significant amounts of capital tied up in hardware running at 40% utilisation, with every idle hour compounding losses.
Hosted.ai was founded by a team that has worked together for 15 years across three generations of infrastructure. Its platform acts as a connective tissue, sitting between AI workloads and the hardware to manage GPU memory in real-time, so multiple jobs can safely share the same processors without performance loss.
For infrastructure operators, higher utilisation means existing hardware can support far greater workloads, enabling operators who were running at thin or negative margins to reach profitability within 12 months, without additional capital expenditure. Ultimately, Hosted.ai is developing software that will allow AI developers to access consumption-based pricing that leads to significantly lower costs.
The technology also aims to diversify the global AI infrastructure market, which is currently dominated by a handful of US hyperscalers. For European companies and public sector organisations facing data sovereignty rules, more efficient regional infrastructure could provide an alternative to relying on overseas providers. A federated GPU marketplace also allows service providers on the Hosted.ai platform to publish spare capacity and subscribe to capacity from other providers, therefore seamlessly extending their geographic reach without owning the underlying infrastructure.
Ditlev Bredahl, CEO of Hosted.ai said: “The GPU infrastructure powering AI today is incredibly wasteful. Customers are forced to rent GPU as a static resource, because that is the only way operators can sell it – and AI workloads leave about 60% of that GPU idle, on average. Using dynamic memory management, the Hosted.ai platform allows workloads to access full GPU capacity, while it intelligently optimises workload placement to maximise the utilization of each GPU in the data center. The result is elastic, efficient GPU that transforms economics for operators and their customers.”
Carl Fritjofsson, Partner at Creandum said: “Founder-market fit like this is so rare – Ditlev and his team have built this business twice before, for the dotcom era and then the cloud era. The team have unmatched technical experience and are better positioned than anyone to use this edge in building for the GPU era, a market that will be orders of magnitude larger than the two generations that came before.”
For more startup news, check out the other articles on the website, and subscribe to the magazine for free. Listen to The Cereal Entrepreneur podcast for more interviews with entrepreneurs and big-hitters in the startup ecosystem.
