consultantsreview logo

Consultants Review Magazine

Indian Largest Supercomputer with 8 Exaflops to be Developed by Microsoft-backed G42

By Consultants Review Team Wednesday, 11 September 2024

According to a senior corporate executive, Microsoft and Mubadala-backed G42 is slated to develop India's largest supercomputer, with eight exaflops of processing capacity that can do millions of trillions of floating-point operations per second. The beta version of the Hindi language big language model, an AI engine that enables users to converse in Hindi, English, and Hinglish (a combination of Hindi and English), was also disclosed by Manu Jain, CEO of G42 India.

He mentioned that earlier this year, Prime Minister Narendra Modi traveled to Abu Dhabi and that India and the UAE had inked a memorandum of understanding on digital infrastructure. According to Jain "The partner selected to carry out this Memorandum of Understanding is us. One of the three topics covered in the MoU was the establishment of a massive 2 gigawatt data center in India. This has the potential to treble India's data centers' current capacity. The third was co-developing AI models in India, according to Jain. The second was constructing one of India's greatest supercomputers, up to 8 exaflops, which we (G42) are creating alongside Cerebras."

However, he omitted to indicate when the supercomputer and data center will be built up. "We are overjoyed to announce the introduction of Nanda, the major language model in Hindi, today. The highest Himalayan peak in India inspired the name of the LLM, according to Jain.

He claimed that a 13-billion parameter model, which was trained on around 2.13 trillion tokens from language datasets, including Hindi, was used to train NANDA. By ensuring that the rich history and complexity of the Hindi language are reflected in the digital and AI world, NANDA is helping to usher in a new era of AI inclusion. Acting CEO of G242 group firm Inception, Andrew Jackson, stated that NANDA "exemplifies G42's unwavering commitment to excellence and fostering equitable AI."
 

Current Issue