IBM is planning to use its own artificial intelligence (AI) chips to lower the costs of operating cloud computing service, Mukesh Khare, general manager of IBM Semiconductors, told Reuters at a semiconductor conference in San Francisco.
Cost Reduction: IBM’s primary motivation for using its own AI chips is to lower the costs associated with operating its cloud computing service. The high costs of previous AI systems, such as Watson, hindered their market traction. By leveraging their own chips, IBM aims to improve cost efficiency, which could make their cloud service more competitive in the market.
Power Efficiency: IBM emphasizes that their in-house AI chips are power efficient. This aspect is significant as it directly contributes to reducing operational costs for the cloud service. Power-efficient chips consume less energy, resulting in lower electricity bills and potentially enabling IBM to offer competitive pricing to customers.
Generative AI Technologies: IBM aims to capitalize on the current boom in generative AI technologies. These technologies, which can produce human-like text, have gained substantial attention and demand. By incorporating their AI chips into the new “watsonx” cloud service, IBM seeks to offer a compelling solution in the generative AI space.
Partnership with Samsung Electronics: IBM has collaborated with Samsung Electronics for semiconductor research and has selected them as the manufacturer of their AI chips. This partnership demonstrates IBM’s focus on leveraging industry expertise and resources to develop and deploy their AI technology effectively. Samsung’s manufacturing capabilities likely play a crucial role in ensuring the scalability and availability of the AI chips.
Differentiation Strategy: IBM’s decision to design its own AI chips aligns with the approach adopted by other tech giants like Google and Amazon.com. By developing proprietary chips, IBM can differentiate its cloud computing service in the market. This strategy allows them to tailor the hardware and optimize it for specific AI workloads, potentially offering superior performance or cost advantages over competitors.
Focus on Inference: IBM’s AI chip is designed primarily for inference, the process of using a trained AI system to make real-world decisions. This indicates that IBM is strategically targeting the segment of the AI market where it can have the most immediate impact. By focusing on inference instead of training, which requires substantial computational resources, IBM aims to address the current demand for deploying AI systems in practical applications.
Timing and Availability: While IBM has thousands of prototype chips already in operation, there is no specific timeline for when they will be made available to cloud customers. The lack of a set date suggests that IBM is likely undergoing further testing and refinement before a commercial release. Timing is crucial in the fast-paced technology industry, and IBM must strike a balance between ensuring the chip’s readiness and capitalizing on the market demand.
Overall, IBM’s consideration of using in-house AI chips for its cloud computing service reflects a strategic move to reduce costs, leverage power efficiency, and capitalize on the growing demand for generative AI technologies. By designing their own chips, partnering with Samsung, and focusing on inference, IBM aims to differentiate itself in the market and provide a competitive offering to customers. The successful deployment of their AI chips could position IBM as a prominent player in the cloud computing and AI industries.