IBM kicked off its annual Think partner and developer conference with a focus on making Watson available everywhere. Watson has traditionally been a solution only available through IBM’s Cloud services. But today, IBM announced it is opening Watson to run on any private, public, hybrid or multi-cloud solution. In addition, IBM is offering an on-premises solution running Watson. IBM is essentially looking to make Watson the most accessible Artificial Intelligence (AI) solution that can execute where ever the data resides, even if it is split between platforms.
IBM has been a leader in AI technology with the development of Watson, an intelligent platform capable of processing a wide range of information and adapting to almost any application. Since its introduction, Watson has been applied to a to almost everything from competing with humans in games to customer service to scientific and business modeling. Watson is recognized as an innovative breakthrough that is paving the direction of applied AI. During the keynote address, several companies, including Geico and Hyundai Financial, discussed how they are already leveraging Watson AI as the foundation of their business solutions.
IBM is now extended the reach of Watson platform (Watson Studio, Watson Assistant, Watson OpenScale, etc) on the new IBM Cloud Private for Data (ICP for Data), which can be deployed on IBM Cloud or any private, public, hybrid, or multi-cloud environment through the use of Kubernetes technology. Additionally, IBM announced an extension of Watson Machine Learning with the Watson Machine Learning Accelerator for high-performance GPU clustering.
While Watson will run on both Power and X86 systems, there is a performance benefit to running on IBM’s Power-based systems, and in AI, performance matters. Whether it is developing new AI models or processing information in real-time, delays can cost companies in productivity, quality, and/or safety. IBM’s Power AI platform combines multiple Power9 CPUs with Nvidia Graphics Processing Units (GPUs) with high GPU memory bandwidth. The PowerAI platform is the only GPU-accelerated platform that links the processors to the GPUs using Nvidia high-speed NVLink interface. The AC922 can be configured with up to two 20-core Power9 CPUs and up to six Tesla V100 GPUs. Multiple systems can be linked to scale to up to 256 GPUs. According to IBM, recent tests demonstrated the ability to developed AI models up to ten times faster than competing platforms. As a result, IBM is now making Watson available on the PowerAI platform for on-premises deployments. As part of the announcement, IBM is also renaming the PowerAI Enterprise platform to the “Watson Machine Accelerator”.
Watson has become a key foundation for the company’s product and service offerings, as a result, IBM is continuing to invest in new capabilities for the platform that will be introduced in the future. IBM also offers business services to assist customers in implementing AI in their business, which makes IBM stand out from the rest of the AI crowd.
While IBM did not highlight how the company’s other research efforts will impact the Watson platform, TIRIAS Research believes that IBM’s leadership in quantum and neuromorphic computing will further advance the platform and when combined with IBM’s business services, will secure the company a leading position in the evolving AI landscape.
The author and members of the TIRIAS Research staff do not hold equity positions in any of the companies mentioned. TIRIAS Research tracks and consults for companies throughout the electronics ecosystem from semiconductors and sensors to the cloud. Members of the TIRIAS Research team have consulted for IBM, NVIDIA, and other companies throughout the electronics ecosystem.
