---
title: "U.S. Stock Market News | Dominant Position at Risk, NVIDIA Falls Over 5%"
type: "News"
locale: "zh-CN"
url: "https://longbridge.com/zh-CN/news/210228054.md"
description: "On Tuesday, NVIDIA's stock price erased its initial gains, falling over 5% to $105.8. On the news front, Apple released the first preview of iPhone AI, followed by a paper stating that its artificial intelligence model was trained on Google's TPU (Tensor Processing Unit). The paper introduced the foundational language model developed to support Apple Intelligence features, including a 3 billion parameter model designed to run efficiently on devices and a large cloud-side model based on private cloud computing. According to the paper released by Apple, they trained the cloud-side AFM (Apple Foundation Model) on 8192 TPU v4 chips, using a sequence length of 4096 and a batch size of 4096, with a training of 63 trillion tokens. The edge-side AFM was trained on 2048 TPU v5p chips. Apple's decision indicates that some large tech companies may be looking for alternatives to NVIDIA's graphics processing units in artificial intelligence training"
datetime: "2024-07-30T15:29:02.000Z"
locales:
  - [zh-CN](https://longbridge.com/zh-CN/news/210228054.md)
  - [en](https://longbridge.com/en/news/210228054.md)
  - [zh-HK](https://longbridge.com/zh-HK/news/210228054.md)
---

> 支持的语言: [English](https://longbridge.com/en/news/210228054.md) | [繁體中文](https://longbridge.com/zh-HK/news/210228054.md)


# U.S. Stock Market News | Dominant Position at Risk, NVIDIA Falls Over 5%

According to the Zhitong Finance and Economics APP, on Tuesday, NVIDIA (NVDA.US) saw its stock price erase the initial gains, falling over 5% to $105.8. On the news front, Apple released the first preview of iPhone AI, followed by a paper stating that their artificial intelligence model was trained on Google's TPU (Tensor Processing Unit). The paper introduced the foundational language model developed to support Apple Intelligence features, including a 30 billion parameter model designed to run efficiently on devices and a large cloud-side model based on private cloud computing.

Apple's paper stated that they trained the cloud-side AFM (Apple Foundation Model) on 8192 TPU v4 chips, with a sequence length of 4096 and a batch size of 4096, conducting training with 6.3 trillion tokens. The edge-side AFM was trained on 2048 TPU v5p chips. Apple's decision indicates that some large tech companies may be looking for alternatives to NVIDIA's graphics processing units in artificial intelligence training

### 相关股票

- [NVIDIA (NVDA.US)](https://longbridge.com/zh-CN/quote/NVDA.US.md)
- [VanEck Semiconductor ETF (SMH.US)](https://longbridge.com/zh-CN/quote/SMH.US.md)
- [Direxion Semicon Bull 3X (SOXL.US)](https://longbridge.com/zh-CN/quote/SOXL.US.md)
- [GraniteShares 2x Long NVDA Daily ETF (NVDL.US)](https://longbridge.com/zh-CN/quote/NVDL.US.md)

## 相关资讯与研究

- [08:01 ETInnofiber, Acrolite, and PAVE Technology Co. Combine to Form Connexis Group, a Leading Provider of High-Performance Interconnect Solutions](https://longbridge.com/zh-CN/news/281362986.md)
- [TSMC Gets Nod to Deploy 3nm Process at Japan Fab](https://longbridge.com/zh-CN/news/281343240.md)
- [What's Going On With SEALSQ Stock Wednesday?](https://longbridge.com/zh-CN/news/281403873.md)
- [This Underrated Artificial Intelligence (AI) Infrastructure Stock Has Surged 80% in a Year. It Can Still Surge 53%.](https://longbridge.com/zh-CN/news/281402576.md)
- [Micron Revenue Smashes Expectations, but Wall Street Yawns](https://longbridge.com/zh-CN/news/281414796.md)