International Business Machines (IBM) is considering the using its in-house designed artificial intelligence chips to reduce the operational costs of its newly introduced cloud computing service.
In a Reuters report, Mukesh Khare, General Manager of IBM Semiconductors, said that the company is contemplating using a chip called the Artificial Intelligence Unit as part of its new “watsonx” cloud service.
IBM watsonx is the tech giant’s enterprise-ready AI and data platform. The platform is comprised of three products to help organisations accelerate and scale AI – the watsonx.ai studio for new foundation models, generative AI and machine learning; the watsonx.data fit-for-purpose data store, built on an open lakehouse architecture; and the watsonx.governance toolkit to help enable AI workflows to be built with responsibility, transparency and explainability, which will become available later this year.
According to reports, IBM aims to capitalise on the growing demand for generative AI technologies capable of producing human-like text, which comes more than ten years after Watson, its earlier AI system, failed to gain significant market traction. The high costs associated with the previous Watson system presented a challenge, which IBM aims to overcome this time by leveraging its own power-efficient chips to reduce cloud service expenses. Although IBM announced the existence of the chip in October, it did not disclose the manufacturer or its intended use.
Khare reportedly disclosed that Samsung Electronics, a collaborative partner of IBM in semiconductor research, manufactures the chip under consideration, and IBM is evaluating its integration into the watsonx platform. Although IBM has not specified a definitive release date for customers, Khare mentioned that the company already operates thousands of prototype chips.
In line with industry leaders like Google and Amazon, IBM has ventured into the development of its own AI chips. However, Khare emphasised that IBM’s chip is not meant to directly compete with Nvidia’s dominant semiconductors in the AI training market, which handle extensive data processing. Instead, IBM’s chip prioritises cost-efficiency in inference, where pre-trained AI systems are utilised for real-world decision-making.
Khare clarified that the current market demand lies in the realm of inference, and IBM aims to maximise its impact in this area. Rather than concentrating on AI training, which necessitates significant computational resources, IBM focuses on delivering efficient solutions for deploying pre-trained AI models.
