除夕当天一觉醒来，美股纳指跌超3%，英伟达市值蒸发近三个字节跳动！国产大模型DeepSeek持续撼动美国霸权位置。
1月28日凌晨，DeepSeek发布了全新的视觉多模态模型Janus-Pro-7B。该公司称，这一模型在多项基准测试中击败了OpenAI的DALL-E 3和Stable Diffusion，成为AI领域的新标杆。
The Hangzhou-based AI company DeepSeek dropped its newly unveiled Janus-Pro-7B image generator Tuesday morning. The company said that its new model outperforms OpenAI's DALL-E 3 in benchmarks, while its app has overtaken ChatGPT in U.S. App Store downloads.
华尔街：慌得不行
DeepSeek的低成本高效能模式彻底颠覆了硅谷的“算力至上”逻辑。其最新模型仅需600万美元的算力成本，远低于美国同行的数十亿美元投入，这直接导致美股科技股集体暴跌。截至当地时间1月27日收盘时间，英伟达股价单日暴跌17%，市值蒸发近6000亿美元，创下华尔街历史最大单日跌幅。
U.S. tech stocks faced significant pressure on Monday as investors reassessed the economic implications of Chinese startup DeepSeek's rapidly advancing models. Nvidia led the carnage, plunging nearly 17 percent to erase almost $600 billion in market value – the largest single-day loss in Wall Street history – while the Nasdaq tumbled 3.1 percent and the Philadelphia Semiconductor Index cratered 9.2 percent, its worst day since March 2020.
硅谷：认真学习 华盛顿：敲响警钟
与此同时，硅谷AI巨头正紧急采取行动。据 Information 网站报道，脸书母公司 Meta 连夜成立四个专门研究小组来研究DeepSeek 的工作原理，并基于此来改进旗下大模型 Llama；OpenAI CEO Sam Altman也在社交媒体上表示，DeepSeek的R1模型“令人印象深刻”，并称其为“新的竞争对手”。
特朗普更是直言不讳地表示，DeepSeek的出现“为美国行业敲响了警钟”，并称赞其“更快、更便宜”的AI技术是一种“积极的发展“。
"The release of DeepSeek AI from a Chinese company should be a wake-up call for our industries,” Trump said as he speaks at the 2025 House Republican Members Conference Dinner in Miami, Florida.
"Today and over the last couple of days, I've been reading about China and some of the companies in China, one in particular, coming up with a faster method of AI and much less expensive method. And that's good, because you don't have to spend as much money. I view that as a positive, as an asset," he added.
图：当地时间2025年1月27日，美国佛罗里达州，美国总统特朗普在特朗普国家多拉尔迈阿密酒店出席2025年共和党问题会议并发表讲话。/VCG
