据市场调查机构Factorial Funds近日发布的报告显示，OpenAI在部署其创新的文本转视频模型Sora时，可能面临巨大的硬件需求。该机构预计，在模型运行的峰值阶段，OpenAI将需要多达72万张英伟达的顶级H100 AI加速卡，这将是一笔价值高达216亿美元的巨额投资。
Factorial Funds的报告指出，为了训练Sora模型，OpenAI在一个月内的硬件使用量可能在4200到10500片H100加速卡之间。这意味着，OpenAI的这一项目不仅在规模上，而且在计算资源的密集度上都将创下新纪录。值得注意的是，据估计，使用单片H100加速卡生成1分钟的视频内容，需要耗时12分钟进行渲染，凸显了这一技术背后的复杂性和计算密集程度。
这一预测引发了业界对AI技术发展成本的讨论，同时也反映了OpenAI在推动人工智能边界扩展时所面临的挑战。英伟达的H100加速卡作为目前市场上最强大的GPU之一，其在高性能计算和深度学习任务中的应用，无疑将为Sora模型的开发提供强大的计算支持，但同时也带来了高昂的经济负担。
OpenAI的Sora模型一旦成功部署，有望开启全新的多媒体内容生成时代，但其背后巨大的硬件投入也揭示了尖端AI技术发展的现实问题。随着技术进步的步伐不断加快，如何平衡技术创新与成本控制，将成为未来AI领域发展的重要议题。
英语如下：
**News Title:** “OpenAI’s Sora Text-to-Video Model: Estimated to Require 720,000 NVIDIA H100 Cards, Valued at $216 Billion”
**Keywords:** OpenAI, Sora model, NVIDIA H100
**News Content:** According to a recent report by market research firm Factorial Funds, OpenAI may face substantial hardware requirements in deploying its innovative Sora text-to-video model. The agency estimates that during peak operation, OpenAI could require as many as 720,000 top-of-the-line NVIDIA H100 AI accelerator cards, representing a massive investment of $216 billion.
Factorial Funds’ report suggests that for training the Sora model, OpenAI’s hardware usage could range between 4,200 and 10,500 H100 cards within a single month. This indicates that not only in scale but also in computational intensity, OpenAI’s project is set to establish new records. Notably, it is estimated that generating one minute of video content using a single H100 card might take 12 minutes of rendering, highlighting the complexity and computational intensity behind the technology.
This projection has sparked industry discussions about the cost of AI technology development and the challenges OpenAI encounters in pushing the boundaries of artificial intelligence. NVIDIA’s H100 accelerator cards, among the most powerful GPUs in the market, will undoubtedly provide robust computational support for the Sora model’s development, albeit with a significant financial burden.
If successfully deployed, OpenAI’s Sora model is poised to usher in a new era of multimedia content generation. However, the substantial hardware investment underscores the practical issues in the development of cutting-edge AI technologies. As the pace of technological advancement accelerates, striking a balance between innovation and cost control will emerge as a critical issue in the future of the AI sector.
【来源】https://www.ithome.com/0/758/413.htm
Views: 1