Yes, NVIDIA lost almost a trillion dollars yesterday.

That’s more than a grande latte at Starbucks.

If you’ve not heard of DeepSeek, I promise you will within the next few days.

AI is incredibly expensive to run for one reason: GPUs.

A GPU is a graphical processor chip that was designed for video games.

If you think about video games, every pixel on the screen has to be repainted and calculated several times a second.

A normal CPU is made for multi-processing, but a GPU is designed for MASSIVE parallel processing.

AI models take advantage of GPU parallel processing to provide amazing performance against terabytes of data.

DeepSeek is a Chinese AI startup that has developed an AI model rivaling OpenAI and Google, but at a fraction of the cost

While OpenAI cost hundreds of millions, DeepSeek claims to have only spent $6 million.

AND . . .

DeepSeek isn’t reliant on expensive GPUs like other models.

AND . . .

IT’S OPEN-SOURCE (In case you’re worried about foreign bad actors)

So here’s why it’s different than other LLMs on the market:

• Efficient use of hardware: DeepSeek has optimized its models to run efficiently on less powerful hardware, reducing the need for expensive, high-end GPUs.

• Novel techniques: The company employs innovative methods like "mixture of experts" to activate only a portion of its model at a time, allowing it to run on consumer-grade hardware. It’s only using Think of it as a “highly-distilled” version of OpenAI.

• Focus on reasoning: DeepSeek has placed emphasis on developing models with strong reasoning capabilities, potentially opening up new applications for AI in business decision-making.

In other words, you could run your own DeepSeek LLM on your own “cheap” hardware.

No more per-token fees. No more data security concerns.

Can you see why OpenAI and NVIDIA are scared?

In the next version of my upcoming Build Your Own AI Assistant Boot Camp V2.0, I will be including DeepSeek as part of the training.

You will learn how to build your own AI assistant that you can host on your own hardware and escape the chains of the token mon$ter.

👉🏼 Click here to join the bootcamp!

Troy’s One Cool Thing of the Day

Key Features:

• 🎙️ Superhuman Accuracy: Provides highly precise speech-to-text conversion.

• 🤖 API Integration: Easily incorporate speech recognition into your applications.

• 🔧 Developer-Friendly: Designed for seamless integration into various tech stacks.

• 🧠 AI-Powered: Utilizes advanced artificial intelligence for improved performance.

• 🔌 Versatile Applications: Suitable for various industries and use cases.

The Bottom Line: AssemblyAI offers a powerful solution for businesses looking to leverage cutting-edge speech recognition technology, potentially improving efficiency in areas like customer service, content creation, and data analysis.

There’s a reason 400,000 professionals read this daily.

Join The AI Report, trusted by 400,000+ professionals at Google, Microsoft, and OpenAI. Get daily insights, tools, and strategies to master practical AI skills that drive results.

Sign up now for free and work smarter, not harder.

Disclaimer: Just a heads-up! Some of the links in this newsletter might be affiliate links, which means I could earn a small commission if you make a purchase. No extra cost to you… just a way to keep the magic coming!