Hello Reader,

This week DeepSeek shook the tech world. In this edition, we will learn why it crashed the market, why the market recovered, considerations of using it, and, importantly, the impact of DeepSeek on your job. Let's get started:

How It Started

DeepSeek R1 is a cutting-edge artificial intelligence model developed by the Chinese AI company DeepSeek. Launched on January 20, 2025, it has quickly gained attention for its impressive capabilities and cost-effectiveness. This is very similar to ChatGPT. However, this crashed the market. Why?

• DeepSeek was able to train a better model than ChatGPT but with 1/20th less cost. For reference, DeepSeek stated they only used $6.5M compare to OpenAI burning upto $125 Million.
• This means, the demand for NVIDIA GPU, and cloud infrastructure will go down, causing a market crash
• DeepSeek also open-sourced this model, which means anyone can take the code, enhance it, and customize it!

Why did the market recover, then?

Scale AI CEO Alexandr Wang, whose company provides data to OpenAI, Google, and Meta, dropped this bombshell that DeepSeek is lying about using fewer NVIDIA GPUs. Mr. Wang stated that DeepSeek might have used hundreds of NVIDIA GPUs. Initial testing proves this theory. This means demand for GPU and compute infrastructure will not go down. So the market recovered.

Data Privacy and Workaround

If you use the DeepSeek app or the website directly, it raises several security concerns:

DeepSeek collects extensive personal data from users, including:

• Email addresses, phone numbers, and dates of birth
• User inputs, including text and audio, along with chat histories
• Technical information such as device models, operating systems, IP addresses, and keystroke patterns1​

This data is stored on servers located in China, which is a significant privacy concern. Under Chinese law, all enterprises are required to assist with state intelligence operations. This means that user data managed by Chinese firms could be subject to government monitoring. The Chinese government could potentially access any information entered into the system, including sensitive personal details, financial information, trade secrets, or healthcare data.

Quite scary, right? There are a few workarounds:

• You can deploy the model in Amazon Bedrock (or Sagemaker) using a python notebook: https://community.aws/content/2sECf0xbpgEIaUpAJcwbrSnIGfu/deploying-deepseek-r1-model-on-amazon-bedrock​
• You can also deploy DeepSeek on your local laptop using Ollama, though it takes a good bit of storage: https://huggingface.co/deepseek-ai/DeepSeek-V3​
• Easy button: AI applications like Perplexity (https://www.perplexity.ai/) have deployed the model in US datacenters. You can select DeepSeek R1 model and run your queries.

How Can You Use This for YOUR Career

If you have been following me for some time, you know I hate endless theory crafting. My entire goal is to help us achieve a better career and job so we can prosper. Which brings me to this: How can we use what we know from the above arguments to our advantage for interviews and working projects? Let's look at the Gen AI layers:

• The bottom layer is the hardware layer, i.e., the silicon chips that can train the models. Example - AMD, NVIDIA
• Then comes the LLM models that get trained and run on the chips. Examples are DeepSeek, Open AI, Anthropic etc.
• Then comes infrastructure providers who provide an easier way to consume, host, train, inference the models. Examples include AWS, Azure, and GCP. This layer consists of managed services such as Amazon Bedrock, which hosts pre-trained models, or provision VMs (Amazon EC2) where you can train your own LLM
• Finally, we have the application layer which uses those LLMs. Some examples are Adobe Firefly, LLM chatbots, LLM travel agents etc.

Now, the important part - as you go from the bottom to the top, the learning curve gets easier, and so does the opportunity for new market players to enter. Building new chips requires billions of dollars of investments, and hence, it's harder for new players to enter the market. The most opportunities are in the top two layers. If you already know the cloud, then integrating Gen AI with your existing knowledge will increase your value immensely. It doesn't matter which LLM enters the market; you are well-equipped as long as you learn how to use and consume the model.

For example, if you are working in DevOps, learn MLOps; if you know K8s/Serverless, learn how you can integrate Gen AI with those; if you work in an application, integrate with managed LLM services to enhance functionality—you get the idea. I am focusing most of my time on this layer! Especially given that the new US president, Trump, just announced $500B funding for data centers for AI. Don't just think that; this will only benefit data scientists. This will impact all four layers shown above.

Question for you readers - Have you already started learning Gen AI? Are you thinking about it, or perhaps you think this whole thing is a farce and will blow over soon? Feel free to reply, and let me know!

​

If you have found this newsletter helpful, and want to support me 🙏:

Checkout my bestselling courses on AWS, System Design, Kubernetes, DevOps, and more: Max discounted links​

AWS SA Bootcamp with Live Classes, Mock Interviews, Hands-On, Resume Improvement and more: https://www.sabootcamp.com/​

Keep learning and keep rocking 🚀,

Raj