AI chip and software startup Recogni introduced a new computing method on Tuesday that could make its AI training and operating chips smaller, faster, and less costly.
Supported by BMW, Bosch, and venture capital firm Mayfield, Recogni creates specialized chips and software to power AI inferencing, allowing trained AI models to make predictions or decisions based on new data.
The company revealed that its new patented system, Pareto, uses a logarithmic approach that surpasses current methods when running large AI models.
“It is a huge leap in all of the KPIs (key performance indicators) that influence silicon hardware system design when it comes to AI computing,” Recogni’s co-founder and VP of AI, Gilles Backhus told Reuters.
Today’s AI models, like OpenAI’s GPT-4 and Google’s Gemini, require vast amounts of power for the numerous mathematical operations needed for simple chatbot prompts on platforms like ChatGPT.
Recogni stated that its new system transforms these multiplication operations into additions, which cuts down on power usage while preserving accuracy.
The startup has already tested Pareto on AI models from companies like Meta Platforms and Stability AI.
Recogni’s first chip was designed and manufactured using Taiwan Semiconductor Manufacturing Co’s seven-nanometer process. The company is working with an unnamed partner to expand Pareto’s availability and plans to announce this partnership soon.
“We are speaking to companies that are putting hardware in data centers and offering it to the world to whoever wants to basically rent it … that’s definitely one of the deployment routes that we’re considering,” Backhus added.
