The whole AI dynamic just changed...

Try Mammouth now at http://mammouth.ai/ NVIDIA is giving away AI models while supplying the chips that OpenAI depends on. So why compete with your biggest customers? This video looks at Nemotron, NVIDIA’s reported $26 billion investment in open models, and the logic behind making AI cheaper to use: every model still needs compute. Free models could mean more demand for GPUs, data centres and NVIDIA hardware. It’s a strategy with a familiar shape. CUDA helped make NVIDIA chips the standard for developers; Nemotron could keep more of the AI stack inside the same ecosystem. Meanwhile, OpenAI is moving into hardware with Broadcom, looking to reduce its reliance on NVIDIA and protect the value of its models. Google, Amazon, and Microsoft are building chips too. Then there’s Hugging Face, the open-model hub caught between OpenAI’s investment talks and NVIDIA’s $12.9 billion acquisition deal. Its role in this story includes a bizarre incident involving AI models trying to cheat on a hacking test. From Kimi and open weights to custom chips and local AI, this is a look at what happens when models become easier to replace, suppliers move into their customers’ business, and everyone tries to hold on to the most valuable part of AI. LinkedIn: https://www.linkedin.com/in/hariharan-jayakumar-silo Instagram: https://www.instagram.com/hariharan.jayakumar/ TimeStamps: 0:00 - NVIDIA and OpenAI Introduction 1:02 - Free Models 8:03 - Up the Stack, Down the Stack 14:00 - Hugging Face Sources: https://pastebin.com/3vgYbvkb

GPT-4正在为你翻译摘要中......

翻译全文