近日，市场调查机构Factorial Funds发布报告称，OpenAI在部署其文本转视频模型Sora时，可能需要高达72万张英伟达H100 AI加速卡，这一数字价值高达216亿美元。报告指出，一个月的Sora训练需要使用4200到10500片英伟达H100 AI加速卡。如果使用单片H100加速卡，生成一分钟的Sora视频需要渲染12分钟。这一消息透露了人工智能技术在视频生成领域的巨大投入和挑战，同时也展示了英伟达H100 AI加速卡在AI领域的关键作用。
Title: OpenAI’s Sora Video Model: A $21.6 Billion Bet on 720,000 Nvidia H100
Keywords: OpenAI, Sora Video Model, Nvidia H100
News content:
According to a recent report by market research firm Factorial Funds, OpenAI’s deployment of its text-to-video model Sora could require as many as 720,000 Nvidia H100 AI accelerator cards, valued at $21.6 billion. The report suggests that a month of Sora training might use between 4,200 and 10,500 Nvidia H100 AI accelerator cards. With a single H100 accelerator card, generating a one-minute video with Sora would require rendering for 12 minutes. This highlights the substantial investment and challenges in the AI-generated video field, as well as the crucial role of Nvidia H100 AI accelerator cards in AI technology.
【来源】https://www.ithome.com/0/758/413.htm
Views: 1