What is Deepseek & How Is It Different?

DeepSeek is a Chinese AI company. It started in 2023 but made splashes late January 2024. A hedge fund manager named Liang Wenfeng founded it. Liang used money from his own hedge fund to build DeepSeek. The company built its AI model for about $6 million.

That cost is much lower than the billions of dollars that other groups spend. DeepSeek used many Nvidia A100 chips to train its model. That is important because the U.S. limits exports of advanced chips to China. Liang bought these chips before the ban.

DeepSeek calls its model open source. This means people can inspect or modify the code. Some experts say DeepSeek’s model works like ChatGPT.

Others are unsure if it will match the quality of big U.S. tech firms. Even so, many people are curious about DeepSeek. They want to see if it can compete in the world of AI without spending much money.

Why Are People Talking About DeepSeek?

DeepSeek’s app topped the charts on Apple’s App Store in the U.S. That was a big surprise. ChatGPT used to be the top AI app. This sudden rise got the attention of many investors. Some fear DeepSeek could change how AI firms build models.

They wonder if AI can be built for less money and still work well. Some also point to how quickly DeepSeek launched its model. It took only a few months to release. This short timeline hints that AI development might speed up.

Experts say DeepSeek caught investors off guard. Large U.S. companies like OpenAI, Google, and Meta spend billions on AI. DeepSeek spent far less and still claims strong results. That news worried some market watchers. They wonder if big spending is needed or if DeepSeek’s methods show a new path.

How Is DeepSeek Different?

DeepSeek uses “inference-time computing.” This means it only uses the parts of its AI model that fit each query. That saves money and energy. The company says this helps users get faster responses. It also keeps costs down. Other AI models often process bigger chunks of data. That can use more hardware and power. DeepSeek’s design might allow it to run on fewer resources.

People want to see if DeepSeek’s smaller budget can still handle big tasks. Many experts are testing the model. Some say it is good enough to compare with ChatGPT and other big players. Others think it will need more time to prove itself. The open-source nature of DeepSeek gives developers a chance to study it. They might adapt its code for other uses.

Is DeepSeek in the U.S.?

Yes. DeepSeek’s app is in the Apple App Store. You can also find it on their website. Some people had trouble signing up at first. The company said it faced attacks on its service. Because of that, DeepSeek had to limit new accounts. Existing users can log in as usual. DeepSeek is also on Google Play in some regions, but it is not as popular there.

DeepSeek’s presence in the U.S. has caused discussions about security and future rules. Some wonder if the government will step in. Others say it is too early to worry about that. DeepSeek’s main focus is AI content via chat. For now, it looks like a direct rival to ChatGPT.

How Is DeepSeek Affecting U.S. Tech Stocks?

The arrival of DeepSeek shook the stock market. Nvidia, a big chipmaker, saw its stock price drop. Some worry DeepSeek’s methods might reduce the need for expensive chips. Other AI-related stocks, like those tied to big tech, also fell. Investors wondered if AI would grow in a new way. Some experts say the market reaction might be too big. They believe U.S. firms will still buy top chips for their own models.

Some analysts see the drop in stock prices as a chance to buy shares at lower prices. They think the demand for AI will keep growing. They say that big businesses will likely stick with trusted providers. DeepSeek might not be an easy choice for large American companies. People are waiting to hear more from Meta, Microsoft, and other leaders about DeepSeek’s impact.

I Jumped In

So I went ahead and signed up, and gave it a try. Seemed to work much faster than ChatGPT out of the box. I also ran it through a bunch of our sports content examples we use in Coded Content. It actually performed as good or better than ChatGPT. Interesting?

But… Later on I logged in to try again and got this message. I guess hackers had something to say about the early DeepSeek success. Than 10 minutes later it worked 🙂

Knup’s Take on What is DeepSeek

DeepSeek is new and raising questions. It built an AI model fast and spent little. This shows that AI might not always require huge budgets.

Still, U.S. companies trust local providers. It is unclear if DeepSeek can gain a big following outside China. For now, DeepSeek’s success may push others to rethink their spending. They might look for cheaper ways to build AI tools.

We should keep an eye on DeepSeek’s progress. If it keeps growing, it may show that smaller teams can still shake up the AI world. The price seems right. The speed seems right. The quality seems right.

We will see how things progress.