IBM CEO Arvind Krishna Says DeepSeek An Indicator Of What's To Come, Predicts AI Models Will Use Just 1% Of Today's Energy In 5 Years

IBM IBM CEO Arvind Krishna believes the future of artificial intelligence lies in smaller, faster, and vastly more energy-efficient models — and he says DeepSeek is a glimpse of what's ahead.

What Happened: Speaking at SXSW 2025, Krishna praised DeepSeek, a Chinese-developed AI model that shocked the industry with its small size and high performance.

While many large language models continue to grow in complexity and cost, Krishna said that kind of scale won't be sustainable — and change is coming.

"If a model is about one-tenth to one-fifteenth the size and equally effective, it is 1% of the energy cost," Krishna said. He added that within five years, "most of the models are going to be using 1% of the energy of what they're using today."

See Also: 74% Of DeepSeek’s Output Mimics OpenAI’s ChatGPT, New Study Reveals: ‘It Does Raise Questions About Its Development’

While DeepSeek's origins remain somewhat opaque, Krishna suggested it may have been built using a larger "teacher" model — but its efficiency nonetheless proves a broader point.

"We've been on this path ourselves for about two years now," he said, reinforcing IBM's focus on building smaller, distilled models that retain performance without the massive compute costs.

Subscribe to the Benzinga Tech Trends newsletter to get all the latest tech developments delivered to your inbox.

Why It's Important: As generative AI continues to scale across industries, concerns over energy use and environmental impact are mounting.

Large models require enormous computing power, translating into high costs and carbon footprints. Krishna's prediction signals a major pivot toward leaner architectures that could democratize access to AI and reduce infrastructure burdens — a shift with wide-ranging implications for tech companies, governments, and the planet.

DeepSeek’s assertion that it developed its AI model at a significantly lower cost than its U.S. competitors contributed to Nvidia Corp shedding $600 billion in market value in just one day.

The AI model also played a role in a broader $1 trillion market downturn.

Reports suggest that U.S. authorities are investigating DeepSeek for potential violations of Nvidia chip embargo regulations.

Check out more of Benzinga’s Consumer Tech coverage by following this link.

Read Next:

• DeepSeek Loses Top Spot As Tencent’s Yuanbao AI Becomes The Top iPhone App In China Amid Intensifying AI Battle

Disclaimer: This content was partially produced with the help of Benzinga Neuro and was reviewed and published by Benzinga editors.

Photo courtesy: Shutterstock