Chinese researchers have introduced SpikingBrain-1.0, a new AI system that runs in a very different way from the large models that dominate today’s AI chatbots. The launch took place on September 8 at the Institute of Automation, Chinese Academy of Sciences. The project is led by Li Guoqi and Xu Bo.
Elevate Your Investing Strategy:
- Take advantage of TipRanks Premium at 50% off! Unlock powerful investing tools, advanced data, and expert analyst insights to help you invest with confidence.
Unlike systems that rely on heavy computing power, SpikingBrain works by mimicking how human brain cells send signals. Instead of processing every piece of data all the time, it’s “spiking neurons” that only fire when needed. This cuts energy use and training demands. In addition, the system needed just 2% of the training data that current models use.
Efficiency and Possible Uses
The team reports that SpikingBrain can work much faster in some cases. In a test where the model had to process one million tokens of context, one version ran more than 26 times quicker than a Transformer model. That makes it useful for tasks that involve very long files or records. Examples include reviewing legal contracts, scanning medical notes, running physics tests, or mapping DNA sequences.
China has also linked this research to work on new computer chips. Earlier work by the same group helped design the “Speck” chip, which runs at very low power. The country also introduced a “Darwin Monkey” supercomputer with more than two billion artificial neurons. Both projects show how China is trying to grow its own brain-inspired tech base and lower its reliance on foreign chips.
Impact on the AI Chatbot Market
For investors following the artificial intelligence space, the key intrigue lies in how this technology fits into the market for chatbots and large language models. Currently, companies such as OpenAI’s ChatGPT, Microsoft’s Copilot (MSFT), Alphabet’s (GOOG) Gemini, and Meta Platforms’ (META) Llama operate on Transformer models. These models require high spending on servers and graphics chips, with suppliers like Nvidia (NVDA) seeing direct gains.
If systems like SpikingBrain prove to match performance at lower cost, then the economics of chatbots could change. A model that runs on less data and fewer chips would mean lower costs for training and use. That could lower the barrier for new players, while also pressuring current leaders to test fresh designs.
Using TipRanks’ Comparison Tool, we lined up several leading companies developing AI chatbots similar to ChatGPT. This side-by-side view helps investors better understand each stock as well as the broader AI chatbot market.
