How A Mysterious Chinese Startup Disrupted The US AI Industry

Author

Benzinga Tech Trends
February 04, 2025

Remember when everyone thought building top-tier AI required billions in funding and cutting-edge chips? A small startup from Hangzhou flipped that script, causing a $1 trillion market meltdown in the process.

The Overnight Sensation That Wasn't

DeepSeek's January bombshell didn't just turn heads – it snapped necks. The company's R1 model didn't just match ChatGPT's capabilities; it did it for pocket change. We're talking $5.6 million versus the billions OpenAI burned through. If that sounds like a typo, it isn't.

By The Numbers: Silicon Valley's Wake-Up Call

  • DeepSeek's R1 model: $5.6 million to train (a claim by the Chinese startup).

  • Cost per million tokens: $2.19 (OpenAI charges $60).

  • Market impact: $600 billion wiped from Nvidia's value.

  • App Store performance: Knocked ChatGPT off the #1 spot.

Why Wall Street Is Freaking Out

This isn't just another AI startup story. DeepSeek's success with older, cheaper Nvidia H800 chips has forced investors to rethink the entire AI hardware premium. When a startup can achieve GPT-4 level performance using last-gen tech, those $40,000 H100 chips start looking pretty expensive.

The U.S. government isn't taking this lying down. Scale AI's CEO is reportedly heading to Washington for high-level talks about the DeepSeek situation. Meanwhile, ASML's chief has warned that export controls alone won't maintain Western AI dominance.

The Plot Twist: OpenAI's Accusations

Just when things were getting interesting, OpenAI dropped a bombshell of its own: allegations that DeepSeek trained its model using ChatGPT data. It's adding a layer of drama to an already spicy situation, raising questions about AI development ethics and intellectual property rights.

What This Means For You And Your Business

The Good: The emergence of DeepSeek brings significant opportunities for businesses looking to leverage AI technology. Most notably, companies can now access sophisticated AI capabilities at a drastically reduced cost compared to existing solutions. The open-source nature of DeepSeek's models provides unprecedented flexibility for companies to customize and adapt the technology to their specific needs.

The Red Flags: The primary issue revolves around data security, as DeepSeek's infrastructure requires data storage in China, subjecting it to local laws and regulations. Infrastructure vulnerabilities, including exposed API keys and unsecured chat logs, pose additional risks that businesses must evaluate.

This Week In Tech

This Week In Tech 

Google's Antitrust Battle With Epic Games

Alphabet Inc. faced skepticism from a federal appeals court in San Francisco regarding its efforts to overturn a jury verdict favoring Epic Games. The jury had previously sided with Epic Games in 2023, accusing Google of imposing restrictive policies on its Google Play Store. Google's argument that a similar case against Apple Inc. was dismissed did not seem to sway the court.

Nvidia's Instant Sell-Out

Nvidia Corporation launched its RTX 5090 and 5080 graphics cards last Thursday, only to see them sell out almost immediately. Despite Nvidia's prior warning about limited availability, many gamers were left frustrated as eager buyers snapped up the highly anticipated Series 5000 GPUs.

OpenAI's New Launch

Microsoft-backed OpenAI introduced the o3-mini, a new model in its reasoning series. The o3-mini model, previewed in December, promises to offer "exceptional capabilities" in science, math, and coding while maintaining low costs and reduced latency. It is the first small reasoning model from OpenAI to support developer features such as function calling and structured outputs.

TikTok Controversy

Former Greek Finance Minister Yanis Varoufakis warned that the TikTok ban controversy in the U.S. exposes a deeper issue: the concentration of power among tech platform owners. He argued that Meta Platforms pose similar concerns as TikTok, and cautioned against the concentration of power in the hands of a few "techno-feudal lords."

OpenAI's IP Concerns

OpenAI announced it has uncovered evidence that Chinese artificial intelligence startup DeepSeek allegedly used its proprietary models for unauthorized training purposes. The San Francisco-based company, backed by Microsoft, discovered suspicious activity last fall when security researchers identified individuals reportedly linked to DeepSeek extracting substantial data through OpenAI’s API.

That's all for this week! If you found these updates useful, you'll like more from this newsletter. Get deeper dives, hot takes, and all the latest tech news delivered straight to your inbox.