快科技8月23日消息，NVIDIA不仅是全球最赚钱的AI硬件公司，现在也把重点放在大模型开发上了，为此不惜砸了70亿美元招人，要开发全球最强的开源大模型。    Abuja, Oct 10 (UNI) At least 76 people were killed after a boat capsized in the southern state of Anambra at the weekend, Nigerian President Muhammadu Buhari confirmed.



In a statement on Sunday, Buhari acknowledged that rescue agencies had officially reported the death toll of the tragedy on Friday and expressed his grief.



A passenger boat carrying 85 people capsized in the Ogbaru area of Anambra State on Friday, following rising floods in that area, according to the statement.



An investigation had been launched to determine "other causes of the incident" besides floods, Thickman Tanimu, the southeast zonal coordinator of the National Emergency Management Agency (NEMA), was quotes as saying by local media.



Following the tragedy, Nigeria's National Inland Waterways Authority and the NEMA have embarked on rescue and recovery missions in Anambra.



Buhari said he has directed other rescue and relief agencies to rush to the site, urging the relevant government agencies to check the safety protocols on local transport ferries to make sure such incidents can be avoided in the future.

UNI/XINHUA PRT。美国媒体报道，NVIDIA要跟AI初创公司Poolside达成一项合作协议，会以120亿美元的估值向公司投资10亿美元，同时花费60亿美元购买该公司的技术产权及部分研发人员。

二 | 这个合作模式跟NVIDIA去年底与LPU芯片公司Groq的合作相似，不是直接收购这家公司，而是重金获得技术授权及人才，公司还会继续独立运营。

三 | Poolside公司大约100多名员工将会加入NVIDIA公司，参与的则是NVIDIA自己开发与的大模型Nemotron，旨在帮助该系列大模型成为全球最强。众所周知，NVIDIA是强烈支持开源大模型的，之前也对中国的DeepSeek、Kimi等开源大模型表示赞许，该公司自己的Nemotron大模型也是开源开放的，只不过性能不算出众，达不到DS V4、Kimi K3等中国开源大模型的水平。NVIDIA作为全球最大的AI GPU供应商，而且利润丰厚，在资金及算力资源上优势独特，现在又招揽了一堆人才，Nemotron未来的性能跻身一线是很有希望的。NVIDIA自己下场开发最强的AI大模型，对OpenAI、Anthropic等VIP客户来说不是好事，但老黄这也是以彼之道还施彼身罢了。因为OpenAI、Anthropic两家公司也在大笔购买NVIDIA GPU之外的AI算力，包括谷歌及亚马逊，并且还在开发自己的AI芯片，他们不想依赖NVIDIA的芯片，NVIDIA自然也不能把大模型的希望都放在他们身上。

四 | 对NVIDIA来说，自己开发最强的开源AI还有别的好处，美国迟早要对中国的开源AI下黑手的，NVIDIA的大模型未来有机会抢这部分市场，他们是不希望DeepSeek、Qwen、Kimi、GLM等中国开源大模型成为美国公司使用的主要大模型。