AI Startup: Recogni, AI startup, introduces a new computing method to reduce costs and energy requirements

Artificial intelligence software and chip startup Recognize He released a novel calculation method on Tuesday that could make its chips used to train and run artificial intelligence systems smaller, faster and less expensive to operate.

With the support of BMW, Bosch and venture capital firm Mayfield, Recogni develops specialized chips and software to enable AI inference — the process of machine learning. AI models making predictions or decisions on new and invisible data.

The company said the new patented system, called Paretouses a logarithmic approach that outperforms existing methods when running large AI models.

“It’s a huge leap across all the KPIs (key performance indicators) that influence silicon hardware system design when it comes to AI computing,” Recogni co-founder and vice president of AI Gilles Backhus told Reuters.

Current AI models, such as OpenAI’s GPT-4 and Google’s Gemini, require hundreds of thousands of energy-intensive mathematical operations to generate simple prompts in chatbots like ChatGPT.

Recogniz said its new system converts these multiplication operations into addition operations, significantly reducing energy consumption while maintaining accuracy.

The startup said it has already tested Pareto on AI models developed by Meta Platforms, Stability AI and others. Recogni, whose first chip was designed and manufactured on Taiwan Semiconductor Manufacturing Co.’s seven-nanometer process, said it was working with an unnamed partner to make Pareto more widely available and will announce the partnership in the coming months.

“We’re talking to companies that are putting hardware in data centers and offering it to the world to basically anyone who wants to rent it… that’s definitely one of the deployment paths we’re looking at,” Backhus added.