archy13 - stock.adobe.com
Digital Catapult is partnering with high-performance computing (HPC) firm Cray to give startups in its Machine Intelligence Garage accelerator access to supercomputing resources.
The innovation centre started up its accelerator in January 2018 and is now working with Cray to offer the participants expertise and access to its Accel AI lab, a suite of systems and tools geared towards machine learning, which will help the startups to develop AI services.
Digital Catapult’s head of technology and AI/machine learning lead tech, Anat Elhalal said Cray’s “expertise in HPC” will give the startups a much-need boost in their work.
Per Nyberg, Cray’s vice-president of market development, AI and cloud, said: “We are seeing a convergence between AI and supercomputing as organisations work towards taking AI projects out of beta and into production.
“The number of AI models that must be trained will only increase and organisations are coming to realise the challenges that will arise.”
At the accelerator’s launch, Digital Catapult CTO Marko Balabanovic said a lack of computational power is a big problem for startups in this field.
“You need a staggering amount of compute power to run this kind of stuff,” he said. “You are looking at these little startups having to spend tens of thousands of pounds on cloud compute to do one training run.”
Read more about supercomputing
- Enterprises interest in HPC is reportedly being fuelled by the rising demand for artificial intelligence-based applications and services, and Cray is one of a number of providers looking to cash in.
- Both Microsoft and Amazon are building up their HPC capabilities. The partnership with Cray could help to bolster Azure’s footprint.
This is illustrated by the organisation’s own poll of 68 startups, which found 60% of them felt constrained by the amount of compute power they had access to.
Digital Catapult originally selected five startups for the accelerator and is choosing a new cohort every six weeks. Participants remain in the programme for three years.
Balabanovic added: “What we’re hoping to see is that the startups will get their products out more quickly and will create better machine learning tools and get to their customers more quickly. They will basically get to the market quicker and be able to scale up their company quicker.”