Insider Brief
- IBM is focusing on smaller, specialized AI models optimized for specific business applications rather than large, general-purpose models requiring vast computing resources.
- The company believes that fit-for-purpose AI models provide better accuracy and lower costs, applying lessons learned from past AI ventures like Watson and Deep Blue.
- IBM sees AI’s economic benefits as widely distributed, comparing its adoption to the early internet, where both large platforms and small businesses thrived.
- Image: IBM
IBM CEO Arvind Krishna told Time Magazine that the company is prioritizing smaller, specialized AI models tailored for specific business applications rather than large, general-purpose models that require vast computing power.
Krishna argues that bigger AI models are not necessarily better.
He told Time: “So if I have a 10 billion parameter model and I have a 1 trillion parameter model, it’s going to be 10,000 times more expensive to run the very big model. Then you turn around and ask the question, if it’s only 1% better, do I really want to pay 10,000 times more? And that answer in the business world is almost always no. But if it can be 10 times smaller, hey, that’s well worth it, because that drops more than 90% of the cost of running it. That is what drove our decision. “
IBM’s strategy is to develop fit-for-purpose AI models that are cheaper to run while maintaining high accuracy for specific business tasks.
IBM’s AI approach stems from lessons learned from its early ventures. The company built Deep Blue, the first chess computer to defeat a human world champion, and later Watson, which won the game show Jeopardy! However, subsequent efforts to apply AI to broader challenges, such as medical diagnostics, faced setbacks. Krishna said IBM underestimated the complexities of industries like healthcare, including regulatory requirements and operational workflows.
Avoiding Large Generalist Systems?
That experience led IBM to rethink its AI strategy. Krishna disagrees with the notion that AI’s greatest potential lies in large, generalist systems.
“If you’re willing to have an answer that’s only 90% accurate, maybe,” Krishna told the magazine. “But if I’d like to control a blast furnace, it needs to be correct 100% of the time. That model better have some idea of time-series analysis baked into it.”
IBM sees greater value in AI models designed for targeted industrial and business applications rather than models optimized for broad, open-ended reasoning.
Krishna also believes AI’s economic benefits will be widely distributed. He compares AI’s development to the early days of the internet, where both large platforms and small businesses thrived.
“If I’m going to build a video streaming business, the more content you have, the more people you can serve,” said Krishna. “You get a network effect, you get an economy of scale. On the other hand, you have a shopfront like Etsy. Suddenly the person who’s an artisan who makes two items a year can still have a presence because the cost of distribution is extremely low.”
Quantum Computing Value
While AI remains a priority for IBM, the company is also heavily investing in quantum computing. Krishna says IBM is making progress on two key challenges in quantum computing: high error rates and coherence loss. IBM has extended qubit coherence times—how long a quantum bit maintains its quantum state—to about a tenth of a millisecond, with a goal of reaching a full millisecond.
“We feel over the next three, four, five years — I give myself till the end of the decade — we will see something remarkable happen on that front,” Krishna said.
IBM’s long-term strategy is to position itself as the leading provider of practical quantum systems, enabling industries such as materials science, pharmaceuticals, and energy to leverage quantum-powered discoveries.
Read more about IBM’s quantum strategy at The Quantum Insider.
TIME spoke with Krishna ahead of a ceremony in early February when he was awarded a TIME100 AI Impact Award.
