Following the news that AbbVie is the latest biopharma company to adopt new CS-2 wafer scale chip technology from Cerebras, which offers the computer power of 128 NVIDIA graphics processing units (GPUs), while consuming only a third of their energy;

Mike Orme, Consultant Analyst in the Thematic Intelligence team at GlobalData, a leading data and analytics company, offers his view:

“Saving time is crucial in competitive AI model training and inferencing. By doing everything on a single piece of silicon within a closely knit mesh of processors, fast memories, and compilers, Cerebras’ CS-2 shows up the core weakness of the Nvidia approach of using lots of different types of separate chips that are working in parallel to do the same job.

“The CS-2 is a formidable feat in silicon engineering. CS-2 technology is a speed fiend in comparison to a GPU cluster, and offers a very big bang per watt too. There are several other well-funded contenders with wafer scale offerings, but Cerebras is the clear pack leader for now.”

David Bicknell, Principal Analyst in the Thematic Intelligence team at GlobalData, a leading data and analytics company, offers his view:

“AbbVie joins the Argonne National Laboratory, GlaxoSmithKline, AstraZeneca and TotalEnergies in opting for Cerebras’ breakthrough technology. But, however much Cerebras may be turning heads among research and development (R&D) chiefs, it is unclear if it will be capable of scaling up production to meet any surge in demand triggered by its current success.

“This development doesn’t spell the end of Nvidia’s AI related GPU business, as not everyone needs a $2 million+ superchip. There’s still plenty of business for Nvidia, AMD and Intel in an AI-driven world.”

Dr Lil Read, Senior Analyst in the Thematic Intelligence team at GlobalData, a leading data and analytics company, offers her view:

“Employing natural language processing (NLP) will only strengthen biomedical R&D by harnessing the vast swathes of written text. This has the potential to shorten R&D timelines and boost collaboration.”