正在加载视频...

视频加载失败

The battle lines are clear: OpenAI/Google want centralized control through cloud-based AI. Nvidia is enabling distributed, democratic AI development at the edge. This isn't just about technology - it's about the future of AI control.

72,468 次观看 • 1 年前 •via X (Twitter)

18 条评论

GC Cooke 的头像
GC Cooke1 年前

For context: Today, most AI runs in massive data centers owned by companies like OpenAI and Google. Every time you use ChatGPT, your query travels to their servers, gets processed, and comes back to you. This centralized model creates a fundamental problem:

GC Cooke 的头像
GC Cooke1 年前

It concentrates power in the hands of a few tech giants. These companies control: • What AI models can and can't do • Who gets access to the technology • How much it costs to use • What happens to your data OpenAI has been aggressive about maintaining this control...

GC Cooke 的头像
GC Cooke1 年前

According to Marc Andreessen on Joe Rogan: When meeting with the Biden administration, they discussed regulating only 2-3 companies for AI development. This mirrors what Sam Bankman-Fried attempted with crypto regulation. But Nvidia just disrupted this power play:

GC Cooke 的头像
GC Cooke1 年前

The Jetson Nano is designed specifically for AI: It can run AI language models locally - without needing cloud connections. Think of it like a Raspberry Pi specifically optimized for AI applications. This $249 device fundamentally changes how AI will develop:

GC Cooke 的头像
GC Cooke1 年前

Instead of all AI running in centralized data centers... We're moving to "edge computing" - where AI runs directly on devices. Imagine your smart home devices, cars, and phones all running their own AI models. The implications are massive:

GC Cooke 的头像
GC Cooke1 年前

1. Privacy & Security When AI runs locally, your data stays on your device. No more sending sensitive information to big tech companies. No more questions about how your personal data is being used or stored. But the revolution goes deeper:

GC Cooke 的头像
GC Cooke1 年前

2. Speed & Reliability Local AI means instant responses - no internet latency. Your AI works even without connectivity. No more waiting for cloud processing or dealing with server downtime. And here's the game-changing part:

GC Cooke 的头像
GC Cooke1 年前

3. The Data Revolution According to Ilya Sutskever, OpenAI's co-founder: The era of pre-training AI on existing data is done. The future of AI data will come from the edge.

GC Cooke 的头像
GC Cooke1 年前

Edge data is fundamentally different: It's real-time information from actual device usage and human interactions. Unlike static internet data, it captures context and real-world behavior. This creates a massive power shift in AI development:

GC Cooke 的头像
GC Cooke1 年前

The companies building edge devices - not cloud providers - will control this valuable data. That's why the Jetson Nano is so disruptive: It puts AI development in everyone's hands, not just tech giants. The power moves from centralized clouds to distributed edges.

GC Cooke 的头像
GC Cooke1 年前

We're witnessing a pivotal moment: Will AI be controlled by a few powerful companies? Or will it be distributed among millions of developers and creators? The revolution begins at the edge.

GC Cooke 的头像
GC Cooke1 年前

A bit more about me! I'm Graham. Author, innovator, entrepreneur, and one of the first European employees at Google. I built products with them that generated over $2B+ in revenue. Currently working on @bravaxyz for effortless, secure, stablecoin yields, on autopilot.

GC Cooke 的头像
GC Cooke1 年前

Thanks for reading! - you're awesome! Follow me @thegrahamcooke for more. Like/Repost the quote below if you benefited:

The Information 的头像
The Information1 年前

OpenAI is betting on a little-known startup to stay ahead of Elon Musk in the supercomputer race.

ic.purpz 的头像
ic.purpz1 年前

$icp decentralised crypto cloud

UHD K1ng 的头像
UHD K1ng1 年前

Such a AI father. He is the boss of the bosses and brilliant in every speech. However product should be 350 and 16 GB VRAM due 8 GB not enough even for today LLM that has excellence marks. 16 GB and call it Pro or Ti or Max #JensenHuang, @nvidia , @NVIDIAAI , @NVIDIAGeForce

It’s Scott 的头像
It’s Scott1 年前

you may want to pay attention to this.

BIONASYS 的头像
BIONASYS1 年前

@hedera makes this possible! To own your data and verify the dataset the AI was trained on!

相关视频