最新资讯
reddit.com OpenAI 称 Astra 仅满足 10% 安全任务标准 dev.to FreeToken:在自有 GPU 上运行前沿 MoE 模型及其内存需求 dev.to 当 AI Agent 说“完成”时,究竟完成了什么? dev.to 部署 Underdog Bark 3.0 Pro:逐字转写、自定义后端与分块设置 dev.to 为艰难对话构建 AI:五个 Agent、一条安全正则与一位盲评监督者 techcrunch.com Palo Alto Networks 洽谈以 5 亿美元收购 AI Agent 安全初创公司 Clutch theinformation.com 黄仁勋被要求就涉华间谍指控向国会作简报 www.infoq.com Edera 推出 Object Capability Bounded Authority,试图将凭证与 AI Agent 工作负载隔离 wechat 这波杀疯了!旗舰级模型不限量、1个月视频额度免费领 bloomberg.com AI 抛售蔓延至信用市场,软银领跌亚洲美元债 dev.to 如何在专用 GPU 服务器上用 OpenAI 兼容 API 部署 LLM dev.to 审计 153 份 Claude Code 与 Codex 会话记录:User-Agent 与 URL 中的邮箱泄露
多元动态
登录 注册
大模型 techspot.com · @Skye Jacobs

三星与 OpenAI 合作开发下一代芯片,同时大规模部署 ChatGPT

原文 我的收藏

报道称三星正与 OpenAI 在下一代芯片上展开合作,同时在内部大规模部署 ChatGPT。Kim 在首尔记者会上表示,与三星电子的联合生产与研究是进展最大、获得认可最多的领域之一,但未提供芯片细节或说明具体内容。

AI推荐理由 TechSpot 报道三星与 OpenAI 在下一代芯片上合作,同时在内部大规模部署 ChatGPT。引述 Kim 在首尔记者会上的说法,称与三星电子的联合生产与研究是进展最大的领域之一,但未披露芯片细节。涉及 AI 芯片合作与大规模企业部署两条线索,但信息较笼统,缺少具体规格与规模数据。

60

正文

Serving tech enthusiasts for over 25 years. TechSpot means tech analysis and advice you can trust. What we know so far: OpenAI is working more closely with Samsung on chip development while also deploying ChatGPT within the South Korean company, adding another major technology partner as it builds out its AI infrastructure. Harrison Kim, general manager of OpenAI Korea, pointed to the companies' work on chip manufacturing and research as a key part of their expanding relationship. "One of the areas where we have made the most progress and gained the most recognition with Samsung Electronics is our joint production and research on the next-generation chips we are developing," Kim said at a press conference in Seoul. Kim did not provide details on the chips or explain Samsung's role. Samsung said it could not confirm information related to customers. The work comes after OpenAI's June announcement of Jalapeno, its first custom AI chip. OpenAI designed the chip with Broadcom for AI inference, and TSMC is set to manufacture it. OpenAI's interest in Samsung also reflects memory's growing importance in AI systems. Large models need substantial memory capacity and bandwidth to run quickly, particularly in data centers serving large numbers of users. Kim said demand for advanced memory will continue to rise as AI computing becomes faster and more complex. Samsung and SK Hynix signed letters of intent last year to supply memory chips for OpenAI's Stargate data center project. Kim said working with South Korean semiconductor suppliers is a priority for OpenAI's local office. // Related Stories Samsung's new portable SSDs are blazing fast over USB4, but will torch your wallet OpenAI faces Senate probe over Hugging Face breach as more rogue AI activity is uncovered Samsung is also one of OpenAI's biggest enterprise users. Kim said it was "one of the largest-scale deployments of ChatGPT globally," with employees in Korea and other markets using the tool for research and development, marketing, and sales. The company has been opening access to several generative AI products internally. In June, Samsung said employees in its Device eXperience division, which oversees phones, televisions, and home appliances, would be allowed to use ChatGPT, Google's Gemini, and Anthropic's Claude for work. OpenAI said adoption among South Korean businesses has increased sharply. The number of ChatGPT Enterprise users at companies and institutions in the country was about 28 times higher at the end of August than a year earlier. It did not disclose the total number of users. The company said businesses are increasingly using AI across departments, including legal, sales, finance, marketing, and operations, instead of limiting the technology to individual assignments. South Korea had the highest number of paying ChatGPT subscribers outside the United States as of late last year, OpenAI said. OpenAI expects product use to grow now that GPT-6 Astra has launched. The company said the model offers better performance in computer use, professional work, science, coding, and cybersecurity. Share this article: See more TechSpot in Google Add us as a preferred source and our reporting shows up first when you search. Add TechSpot Featured on TechSpot A developer used AI to revive 26-year-old Voodoo3 cards on Linux How "tokenomics" might save AI PCs

来源报道 · 1 篇 · 以下链接均前往原始媒体

← 返回精选动态 原文

更多内容