AI chip startup Rebellions hopes to challenge Nvidia with AI inference chip: CEO

By Kim Boram

SEONGNAM, South Korea, Dec. 16 (Yonhap) -- The head of South Korean artificial intelligence (AI) chip startup Rebellions Inc. said Tuesday the company is aiming to become a global player and challenge industry leader Nvidia Corp. with its inference-focused chips.

"Our vision is to compete with Nvidia on the same stage at some point in the future," Park Sung-hyun, the startup's co-founder and chief executive officer (CEO), said during a media event held in Seongnam, south of Seoul.

"I know Nvidia does not see us as a competitor yet, but our ultimate goal is to compete with Nvidia and take some of its market share," he said.

Founded in 2020, Rebellions is an AI inference chip designer and developer, and is among a small number of startups that have successfully commercialized AI chips.

Earlier this year, it raised 350 billion won (US$237.4 million) in funding from global big tech companies, including Samsung Electronics Co. and British chip design company Arm.

"For the past five years, the global AI infrastructure market has been searching for a next-generation neural processing unit (NPU) partner, and Rebellion was selected through that process," Park said.

Rebellions supplies ATOM, its first-generation NPU, or an AI accelerator, designed for small data centers and enterprise applications, to SK Telecom Co. for use in the telecom operator's AI telephone assistant service A., also known as A-DoT.

"ATOM is doing well with A., which handles more than 50 million calls a day," he said. "This provides us with a solid reference case for real-world deployment of our chip."

As part of its next strategy to expand globally, Rebellions recently launched the second-generation Rebel-Quad chip, which combines four AI inference Rebel chips to support large-scale data processing.

The company said the new product is targeting governments and companies that seek to establish AI infrastructure outside the Nvidia-dominated ecosystem.

"We expect this second-generation chip to serve as one of the most competitive alternatives to Nvidia GPUs," Park said. "We hope this chip will be our key asset for the next five years."

[email protected]
(END)