Deep seek: The Low-Cost AI That Has Changed Everything

“It’s a side project. I call it Deep seek.” The viral photo of Deep seek founder and CEO, Liang Wenfeng, in front of a whiteboard took the internet by storm some months ago, following the release of the relatively new Chinese AI chatbot, Deep seek. The release of the chatbot caused a huge disruption in the AI field as multiple companies scrambled to limit the damage, while investors rushed to protect their wallets. But what exactly is Deep seek, and why did it cause such disruption?

What is Deep seek?

Deep seek is an AI development company based in Hangzhou, China, founded by Liang Wenfeng in 2023. It operates as an independent AI research lab with no outside venture capital funding. The firm is entirely self-funded and has no external valuation—a unique model among VC-backed startups. Their earliest AI model launched in November 2023, but it was the Deepseek R1 model, released in January 2025, that really propelled the company into the public eye.

How Deep seek Disrupted the AI Market

At first glance, Deep seek might seem like just another chatbot in a crowded field. Yet it boasts traits that set it apart. Whereas GPT‑4 and Claude 3 were trained on budgets of hundreds of millions of dollars, R1 matches their performance on a budget of just six million dollars. It was born as a way to utilize the company’s GPU surplus—hence Liang’s “side project” tag.

Whenever a new innovation hits the market, the lucky companies behind it usually get a huge overvaluation as investors pump their stock prices in hopes of profiting, while consumer and investor hype creates a feedback loop that results in massive, often unjustified valuations. This scenario is more commonly known as a bubble. Given time, however, hype fades—or a new service enters the market that sets a more realistic standard—triggering a bubble burst. A popular example is the Dotcom bubble of the 1990s, when companies like Google and Microsoft became overvalued due to internet hype. When that bubble burst, it triggered a U.S. economic recession.

Before Deep seek hit the market, companies like Open AI, Anthropic, and Nvidia—closely tied to the AI sector—were significantly overvalued due to investor hype. The cost of AI had been set at hundreds of millions of dollars by industry titans, and that was just the price of entry. But when the R1 model was dropped and its low training cost announced, people immediately understood what was coming. The AI bubble burst as investors began pulling out of Nvidia, Open AI, and other industry giants. In just a few days, Nvidia lost about 18% of its valuation, with a whopping $600 billion (yes, billion) of market capitalization wiped away. Other companies suffered similar losses. In total, Deep seek caused the destruction of almost two trillion dollars in market capitalization in less than a week—setting a record for the greatest single market cap loss in history.

Features of Deep seek

Apart from its low training cost, Deep seek also has different core values that set it apart from other tech giants. The company believes in open-source resources and released the R1 architecture openly, allowing developers to draw inspiration from it. As of now, many tools and programs have been created that run on the R1 architecture—free of charge.

Deep seek uses a number of technical methods to keep its costs down, opening up this exciting area of tech innovation to many more individuals and firms who were previously locked out due to high costs.

Deepseek’s story is a prime example of innovation and the inevitable progress of technology. This once-restricted field is now open to new ideas, as methods have been created to keep costs low and outputs high—strikingly similar to how computers once cost millions to produce and now sit in the palms of everyone’s hands for a relatively affordable price.

Written by Adam Chege.