Laurent -

AI Singapore to set up on-premise HPC infrastructure

Singapore’s national AI programme is getting its own high-performance computing infrastructure to support more compute intensive workloads

Singapore’s national programme to bolster the country’s artificial intelligence (AI) capabilities is getting a boost with an on-premise high-performance computing (HPC) infrastructure that will be able to crunch workloads that used to be run on the cloud.

Laurence Liew, director for AI industry innovation at AI Singapore, said as the number and variety of AI workloads grew under the 100 Experiments project, it soon became necessary to acquire its own infrastructure.

A flagship initiative of AI Singapore, 100 Experiments was conceived to solve problems for businesses by using AI, as well as to support the growth of local AI technology startups.

Although smaller workloads could be run from the cloud, Liew said it was getting expensive to crunch compute intensive applications such as voice-to-text translation.

“Within a week, the researchers can easily incur $5,000 to $10,000 in cloud resources – that’s enough to buy a server in two weeks,” said Liew. “If you have the technical capabilities, it can be cheaper to run them on-premise.”

The on-premise HPC infrastructure, which comprises a range of Dell EMC PowerEdge servers, as well as switches built for applications in high-performance datacentres, will also address data privacy requirements of government agencies and large companies that will not allow sensitive projects to be run from the cloud.

Romain Bottier, HPC and AI senior solution architect for South Asia at Dell Technologies, said for programmes like AI Singapore that work with a wide variety of projects, there is no single solution that makes sense. “You need to have a hybrid approach to address different workloads and data governance challenges,” he said.

Read more about AI in APAC

During a media briefing, Liew revealed that AI Singapore has received project proposals from about 300 companies to participate in the 100 Experiments initiative, out of which 41 projects have been approved.

Over 30 of the approved projects, including a project to help Expedia improve online search accuracy for Asian languages, are currently in progress. “AI in no longer academic in Singapore, we have a lot of projects in deployment,” he said.

Besides laying the infrastructure foundation required to spur AI adoption, AI Singapore has also been helping to groom AI talent in the city-state.

In May 2018, it teamed up with local polytechnics to conduct AI workshops for students. This was followed by the AI for Everyone programme later in the year to educate the public on how AI can be used in their daily lives.

According to a report by Cisco and Oxford Economics, nearly a fifth of full-time workers in Singapore could lose their jobs from greater adoption of technologies including AI and robotics over the next decade, more so than other ASEAN nations.

The authors of the report noted that Singapore’s higher job displacement rate is not a case of “technology catch-up”, but rather a result of having an “exceptional” environment for innovation and digital transformation.

Read more on Artificial intelligence, automation and robotics

Data Center
Data Management