On August 28, 2024, I saw the Alibaba office building in Nanjing, Jiangsu Province, China.
CFOTO | Future Publishing | Getty Images
Alibaba Stocks soared Wednesday after Chinese behemoth revealed a new model of reasoning that claims can match DeepSeek’s global sensation R1.
Alibaba-listed Hong Kong stock ended its 8.39% stock to hit a new 52-week high – the company’s New York-traded stock rose about 2.5%. Alibaba shares have risen nearly 71% in Hong Kong so far.
Chinese Giant Thursday Revealed QWQ-32B is its latest AI inference model, which says “competitor cutting-edge inference models, such as DeepSeek-R1”.
Compared to DeepSeek’s 671 billion parameters, Alibaba’s QWQ-32B has 32 billion parameters with 37 billion parameters, which is actively involved in the inference process – a process of running real-time data through a trained AI model to generate predictions or solve tasks.
Parameters are variables used by large language models (LLMS) (AI systems that can understand and generate human languages) in predictions and decision-making. Lower parameters usually signal is more efficient in the case of an increase in demand for optimized AI, which consumes less resources.
Alibaba said its new model has achieved “impressive results” and the company can “continuously improve performance, especially in mathematics and coding.”
Since the unexpected launch of DeepSeek’s revolutionary R1 earlier this year, established and emerging AI players around the world are competing to produce more efficient, higher-performance models.
Chinese companies have been doubling their technology Alibaba invests in AI as it debuts its first model in 2023. Rapid profit rate hike In the December quarter.
“Looking forward, revenue growth of the AI-powered cloud intelligence group will continue to accelerate,” Alibaba CEO Eddie Wu said at the time.
Optimism surrounding AI development projects could lead to huge gains in Alibaba stocks and “incorporate the company’s earnings on a more upward trajectory”, “” Bernstein analysts say.
“The pace of innovation is very fast now,” Futurum Group CEO Dan Newman told CNBC’s “Squawk Box Europe” on Thursday. “When DeepSeek came out, that left everyone with questions, is it the final answer to Openai? Will current businessmen, Microsofts, Google or Amazon win large-scale investments?”
He stressed that large language models are becoming increasingly “commodified” as developers want to reduce costs and improve access with users.
“Just like we’ve seen this higher efficiency, this cost has dropped, and we’ll see usage as well. In the training era, this is the moment when Nvidia really built the initial AI boom, and it’s an important moment,” Newman said. “But the inferences about AI consumption is actually the future, which will multiply that amount.”
