IBM has unveiled an upgraded computer chip that could give its customers an AI boost, potentially reshaping how some industries handle data-intensive tasks.
For big businesses still relying on IBM’s powerhouse mainframe systems, the Telum II processor could mean smarter, faster operations in an increasingly AI-driven world. Banks might spot fraud more quickly, insurers could process claims more accurately, and retailers may offer more personalized shopping experiences.
The AI chip market is booming, with demand outstripping supply as businesses worldwide clamor for more computing power. Nvidia, AMD, and Intel have surged in revenues, while tech giants like Google, Amazon, and Microsoft are developing custom AI processors. This chip rush stems from the growing ubiquity of AI applications, from chatbots to autonomous vehicles. The resulting competition has spurred innovation, driving down costs while pushing the boundaries of what’s possible in machine learning and data processing.
Experts say the AI chip revolution could profoundly influence commerce. Retailers with powerful AI hardware can now analyze customer behavior in real-time, tailoring promotions and inventory to maximize sales. Financial institutions are deploying sophisticated fraud detection systems that can process millions of transactions per second, potentially saving billions in losses. Supply chain managers use AI to predict disruptions and optimize routes, cutting costs and improving efficiency.
At the Hot Chips 2024 conference in Palo Alto, California, IBM showcased its next-generation Telum II processor and Spyre Accelerator. Both are designed to supercharge AI capabilities within IBM’s Z mainframe systems, with an expected launch in 2025.
“Developed using Samsung 5nm technology, the new IBM Telum II processor will feature eight high-performance cores running at 5.5GHz,” IBM stated in its announcement. This represents a significant leap forward from the previous generation, with the company adding that the processor will include “a 40% increase in on-chip cache capacity, with the virtual L3 and virtual L4 growing to 360MB and 2.88GB, respectively.”
Perhaps most notably for AI applications, each accelerator on the Telum II is expected to improve “by 4x, reaching 24 trillion operations per second (TOPS).” The boost in processing power is designed to handle more complex AI models and larger datasets, which is critical for businesses dealing with ever-increasing amounts of information.
The Spyre Accelerator, meanwhile, represents a new addition to IBM’s offerings. According to the company, it “will contain 32 AI accelerator cores that will share a similar architecture to the AI accelerator integrated into the Telum II chip.” This accelerator can be connected to IBM Z systems via PCIe, potentially allowing for even greater AI processing capabilities when combined with the Telum II processor.
However, Dev Nag, CEO of QueryPal, cautions against overhyping the announcement. “These are not general-purpose AI chips that will compete directly with NVIDIA’s GPUs or similar offerings from AMD,” Nag told PYMNTS. “It’s unlikely to be adopted by cloud vendors or see widespread use outside of IBM’s existing mainframe customer base.”
While impressive for mainframe users, this development likely won’t reshape the broader AI landscape. Nag points out that the mainframe market “peaked in revenue about two decades ago” and represents a specific enterprise computing segment.
For those more familiar with cloud computing trends, IBM’s focus on mainframe systems might seem puzzling. However, mainframes still play a crucial role in specific sectors. Many of the world’s largest banks, insurance companies, and retailers continue to rely on these systems for their core operations, valuing their reliability, security, and ability to process vast transactions.
The chip innovation is about modernizing existing systems rather than revolutionizing AI computing. The company’s proposed “ensemble method of AI” – blending traditional and new AI models – could offer unique benefits for specific industries. As IBM explains, “Using ensemble AI leverages the strength of multiple AI models to improve overall performance and accuracy of a prediction as compared to individual models.”
The new chip approach could be valuable in scenarios like fraud detection. IBM suggests that “traditional neural networks are designed to provide an initial risk assessment, and when combined with large language models (LLMs), they are geared to enhance performance and accuracy.”
With its latest announcement, IBM is giving its mainframe customers a path to incorporate advanced AI without overhauling their systems. It’s a strategic move to keep these powerful, if aging, systems relevant in an increasingly cloud-dominated world.
“Infusing AI into enterprise transactions has become essential for many of our clients’ workloads,” IBM noted in its announcement. The company highlighted that its “AI-driven fraud detection solutions are designed to save clients millions of dollars annually,” underscoring the real-world impact of these technological improvements.
However, Nag emphasizes the limited scope of this innovation. “The mainframe market, while still significant for certain industries like finance and insurance, represents a niche segment of enterprise computing, albeit broadly supported across the Fortune 500,” he says. He also points out that “Nvidia GPUs boast higher revenue than all of IBM’s System Z mainframe sales.”
We’re always on the lookout for opportunities to partner with innovators and disruptors.
Learn More