DEEPSEEK YT
The Rise of Deep Seek: A Game Changer in AI
Major Market Disruption
- On January 27, 2025, a staggering $589 billion was lost in market value in a single day, marking the largest loss for an American company.
- The cause was not major tech firms like OpenAI or Google but rather a Chinese startup named Deep Seek, which developed competitive AI technology with only $6 million.
Historical Context and Comparison
- Mark Anderson likened this event to the "Sputnik moment" for artificial intelligence, referencing the panic in the U.S. during the early space race when they realized they were falling behind.
- Liang Wfeng, founder of Deep Seek, had humble beginnings and no political connections; he studied at Zang University and began exploring quantitative trading while still in college.
Development of Deep Seek
- In 2015, Liang founded a hedge fund called High Flyer that utilized AI for algorithmic trading and amassed significant GPU resources over time.
- By 2021, High Flyer invested heavily in advanced computing infrastructure before U.S. sanctions limited access to such technology for other companies.
Launching Deep Seek
- In July 2023, Liang launched Deep Seek without external funding; it was entirely financed by his hedge fund where he retained an 84% ownership stake.
- The launch on January 20, 2025 saw their chat model surpass ChatGPT within six days and achieve top rankings across multiple countries with millions of downloads.
Performance Metrics and Cost Efficiency
- Deep Seek's model achieved impressive benchmarks in advanced mathematics and programming compared to established competitors like OpenAI's GPT models.
- Training costs for Deep Seek’s initial model were reported at $5.6 million—significantly lower than competitors' expenditures (e.g., GPT4 cost around $100 million).
Market Reaction and Consequences
- Despite the low training cost figure being somewhat misleading due to prior investments in infrastructure and research failures, it highlighted innovative optimization techniques used by Deep Seek.
- Following these developments on January 27th, Nvidia experienced a massive drop in stock value leading to substantial losses among major tech companies including Broadcom and AMD.
The Impact of AI on the Tech Market
The Shift in Investment Thesis
- Nearly $1 trillion has evaporated from American tech stocks due to a shift in investment thesis regarding artificial intelligence (AI). The prevailing belief was that having more chips and computational power would guarantee success.
- A Chinese open-source startup challenged this notion, demonstrating that efficiency and algorithms could compensate for hardware limitations, prompting OpenAI to rethink its closed model strategy.
OpenAI's Response to Competition
- Following a competitive threat, OpenAI launched ChatGPT 5.2 in December 2025, achieving impressive benchmarks and outperforming human experts in many professional tasks.
- Deepsic also advanced rapidly with multiple model releases, including V4, which boasts 1 trillion parameters focused on autonomous programming.
Price Dynamics and Market Democratization
- The competition initiated by Deepsic forced price reductions across the market; OpenAI's GPT5 nano became one of the cheapest options available.
- January 27, 2025, marked a significant day not just for stock market declines but as a turning point for democratizing access to AI technologies.
Practical Applications of AI for Businesses
- AI is now cheaper than hiring interns; businesses can leverage it effectively without incurring high costs. For instance, Deepsic offers significantly lower token input prices compared to competitors.
- Using Deepsic can lead to substantial savings; small businesses using it could spend as little as $3.50 monthly while accessing powerful tools at no cost through their web chat service.
Real-world Examples of Cost Savings
- Implementing AI can yield significant financial benefits; one accountant saved R$3,300 monthly by automating manual reconciliation processes.
- A marketing agency saved R$72,000 annually by utilizing AI for automated content creation instead of relying on expensive design services.
Challenges with Deepsic's Privacy Policies
- Despite its advantages, users should be aware of privacy concerns associated with Deepsic’s data handling practices.
Data Privacy and Censorship Concerns with DeepSec
Data Storage and Security Risks
- DeepSec stores user data on Chinese servers, including typing patterns, conversation history, and sent files.
- In January 2025, security researchers discovered an exposed database containing over 1 million conversation logs without protection.
- Although the issue was fixed within an hour, the exposure had already caused significant damage.
Censorship Issues
- The model does not respond to queries about certain Asian leaders or sensitive topics like Taiwan's independence, reflecting censorship in its programming.
- Users asking about Taiwan receive a response stating it is an "inalienable part of Chinese territory," showcasing the limitations imposed by the model.
Global Bans on DeepSec
- Several countries have banned DeepSec: Italy blocked it from the Apple Store; Australia prohibited its use in federal systems; Taiwan banned it for government use.
- The U.S. Navy has also restricted its usage, along with NASA and various states like Texas and New York implementing complete bans.
Recommendations for Use
- For personal use or learning purposes, DeepSec can be beneficial as it is free and powerful. However, caution is advised when handling sensitive company data.
- Alternatives such as Clouder or OpenAI's GPT are recommended due to clearer privacy policies and servers located outside China.
Future Developments in AI Technology
- Anticipation builds for V4's release around February 2026 during the Chinese New Year, which may introduce a model with 1 trillion parameters and a context window of 1 million tokens.
- This shift indicates that cutting-edge AI technology will become more accessible to businesses without requiring billions of dollars in investment.
Practical Implementation of AI
- The speaker emphasizes practical applications of AI rather than theoretical knowledge within their platform aimed at helping businesses implement AI effectively.
- Over 850 companies have benefited from this approach, achieving significant cost savings while increasing revenue efficiency.