Singtel yesterday pitched a new AI cloud service to organisations in the banking, healthcare and other highly regulated sectors, seeking to answer a need to run such workloads locally to meet stringent requirements.
Called RE: AI, the service will offer AI computing performance through Nvidia’s much-sought-after graphics processing units (GPUs), whether to train chatbots to answer queries or to analyse large amounts of data, say, for market insights.
For now, Singtel is bringing Nvidia’s H100 GPUs to the market, though it said it is ready to deliver the top-end GB200 GPUs later as well.
In March, Singtel had said it would be making use of Nvidia’s latest GPUs to deliver a GPU-as-a-Service offering to Southeast Asia customers, using Singtel’s data centres in Singapore, Thailand and Indonesia.
At the height of the AI gold rush now, Singtel could potentially cater to a market that current hyperscale cloud providers, such as Microsoft, Amazon Web Services and Google Cloud, don’t yet offer.
While such global players have been gobbling up GPU supply from Nvidia and expanding the scale of AI services, they are yet to offer the AI performance in Singapore.
This gives the Singapore-based telecom operator a window of more than a year to deliver the AI performance needed by organisations that wish to crunch their data here in the city-state, said Manoj Prasanna Kumar, chief technology officer of Singtel’s Digital InfraCo unit.
He added that the cloud-based offering could also be offered in other markets outside of Singapore if there is demand in future.
At a launch at the company’s George Street premises, Singtel also inked agreements with several industry players to build an ecosystem around its AI offering to give it a boost.
Among them is a deal with the Advanced Remanufacturing and Technology Centre, SIMTech, Singapore Manufacturing Federation (SMF) and Singapore Institute of Technology to set up a centre of excellence that makes use of AI for advanced manufacturing.
Speaking at the event, Bill Chang, chief executive of Singtel Digital InfraCo, said 150 companies from SMF’s network are expected to be involved in trials with students and academia from SIT to develop smart AI models and applications for the sector.
There has been interest from a number of organisations in Singapore and the region and Singtel’s new offering will help them get their AI projects up and running quickly, he added.
One example would be using AI to train chatbots to help answer common queries for healthcare organisations.
One Singtel partner, Hippocratic.AI, develops such AI-powered chatbots to help speak to patients. Through Singtel, it can develop relevant answers based on sensitive medical data that is stored locally within a country.
Of course, telecom operators such as Singtel will need to move fast and secure their customers as AI becomes more commonly used by businesses.
In the move to the cloud in years before, many telcos had failed to make use of their close relationships with corporate customers, allowing hyperscale players to come in and deliver more advanced, scalable and attractive offerings.
So, besides selling AI performance, Singtel now also wants to pitch its Paragon “all-in-one” platform that helps customers quickly get basic building blocks in place when they start on their AI journey.
The “orchestration” platform promises to connect up to other cloud providers as well, so a customer can monitor AI workloads on other hyperscale cloud providers in a centralised dashboard, said Kumar.
Plus, Singtel will be able to offer connectivity services to customers that prefer a single provider for their various requirements, he added.
