AI News HubLIVE
站内改写3 min read

Rebellions Aims to Take on Nvidia in AI Inference Chip Battle

The South Korean startup sees itself as capable of competing with the biggest AI infrastructure providers.

SourceAI BusinessAuthor: Shaun Sutner, Esther Shittu

Newsletter Sign-Up

Generative AI

SAP Plans to Turn Spreadsheet AI Startup Into Top Frontier LabSAP Plans to Turn Spreadsheet AI Startup Into Top Frontier Lab

byGraham Hope

Sponsored Content

Why Enterprise AI Stalls Before It ScalesWhy Enterprise AI Stalls Before It Scales

byLuke Norris

Generative AI

Google Ads in AI Mode Will Help Businesses Be DiscoveredGoogle Ads in AI Mode Will Help Businesses Be Discovered

byEsther Shittu

Generative AI

Beijing Lab at $20B as AI Investors Look to ChinaBeijing Lab at $20B as AI Investors Look to China

byGraham Hope

Data Centers

Schneider Electric, Foxconn Partner to Build Next-Gen Data CentersSchneider Electric, Foxconn Partner to Build Next-Gen Data Centers

byScarlett Evans

Generative AI

Rebellions Aims to Take on Nvidia in AI Inference Chip BattleRebellions Aims to Take on Nvidia in AI Inference Chip Battle

byShaun Sutner,Esther Shittu

Generative AI

Startup Gets OpenAI Backing to Overhaul Enterprise AI AutomationStartup Gets OpenAI Backing to Overhaul Enterprise AI Automation

byGraham Hope

Agentic AI

Meta Rolls Out AI Agent for Enterprises GloballyMeta Rolls Out AI Agent for Enterprises Globally

byGraham Hope

Generative AI

Anthropic Forced to Disable New Models by US GovernmentAnthropic Forced to Disable New Models by US Government

byGraham Hope

Generative AI

AI's Hidden Energy Bill: Why Visibility is Becoming Critical for EnterprisesAI's Hidden Energy Bill: Why Visibility is Becoming Critical for Enterprises

byScarlett Evans

Generative AI

Anthropic Forced to Disable New Models by US GovernmentAnthropic Forced to Disable New Models by US Government

byGraham Hope

Generative AI

Prompt: AI IPOs Raise a Question Enterprises Are Still Trying to AnswerPrompt: AI IPOs Raise a Question Enterprises Are Still Trying to Answer

byLiz Hughes

Agentic AI

Cohere North Mini Code Gives AI Developers More ControlCohere North Mini Code Gives AI Developers More Control

byEsther Shittu

Data Centers

Schneider Electric, Foxconn Partner to Build Next-Gen Data CentersSchneider Electric, Foxconn Partner to Build Next-Gen Data Centers

byScarlett Evans

Shaun Sutner,Esther Shittu

June 15, 2026|Updated June 16, 2026

The origin story of this young AI chipmaker’s name captures its ambitious goals.

“Everybody's looking for an alternative to the big incumbent in every industry. And here we are to lead one of the rebellions against the big incumbent in this AI industry -- Nvidia,” said Marshall Choy, chief business officer at Rebellions, on the Targeting AI podcast from AI Business.

The vendor, founded in 2020, set out from the start to focus exclusively on the fast-growing inference sector of AI infrastructure. With the meteoric rise of agentic AI, inference has become increasingly important as enterprises look to deploy autonomous AI agents that depend on it to make their next moves.

“We took a view of, ‘How can we add the most value to the market and to customers?’” Choy said. “And we thought that was by focusing on the inferencing side of things, to complement the existing GPUs that people have today doing training and enables them to actually use AI in a production sense, which is what inferencing is by definition.”

Related:Anthropic Forced to Disable New Models by US Government

Rebellions also goes its own way with the type of AI models for which it designs its chiplets. The vendor focuses more on mixture-of-experts model architectures and multi-modal models than on standard large language models. The startup is also closely associated with open source, particularly the PyTorch community.

As for Rebellions’ chiplets, they are the product of fragmenting a single large AI chip into smaller, specialized silicon units that combine compute and memory into a more compact, customizable package. Chiplets function like a neural processing unit specialized for diverse applications -- from call centers and customer service to healthcare AI systems.

“There's a lot of ways you could go about building an AI chip,” Choy continued. “The route that we took was we wanted to be able to have greater flexibility, greater performance, greater efficiency and lower cost, and pass those savings on to our customers.”

And regarding its dreams of vying with the AI market colossus Nvidia, Rebellions -- which has raised $850 million, is now valued at more than $2.3 billion and is expanding internationally -- is a key member of “K-Nvidia,” a multi-billion-dollar South Korean initiative aimed at building sovereignty in AI and semiconductors. The genesis of this name, as Choy explained, goes back to the idea of taking on the giant.

About the Authors

See more from Shaun Sutner

See more from Esther Shittu

Agentic AI

Cohere North Mini Code Gives AI Developers More ControlCohere North Mini Code Gives AI Developers More Control

byEsther Shittu

Agentic AI

Cohere North Mini Code Gives AI Developers More Control

Cohere North Mini Code Gives AI Developers More Control

Jun 15, 2026

Data Centers

Schneider Electric, Foxconn Partner to Build Next-Gen Data Centers

Schneider Electric, Foxconn Partner to Build Next-Gen Data Centers

Jun 15, 2026

Latest podcasts

Generative AI

Navigating the Future of Work with AI

Jun 1, 2026

Generative AI

How Dataminr Kept the Human in the Loop

May 19, 2026

Generative AI

A New Way of Searching for Jobs with LinkedIn

May 4, 2026

Generative AI

AMD's Vision for AI PCs in the Age of Agentic AI

Apr 21, 2026

See all

Read More