↑ ↓ Navigate ↡ Select ESC Close

DeepSeek - The Chinese AI That Crashed The Markets

2025-01-28 Science & Technology
325.3k
11.4k
1.8k
Matt Wolfe
Matt Wolfe
954.0k subscribers

Unlock all features

FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.

Description

Let's talk about DeepSeek and DeepSeek-R1. Discover More: πŸ› οΈ Explore AI Tools & News: https://futuretools.io/ πŸ“° Weekly Newsletter: https://futuretools.io/newsletter πŸŽ™οΈ The Next Wave Podcast: https://youtube.com/@TheNextWavePod Socials: πŸ–ΌοΈ Instagram: https://instagram.com/mr.eflow ❌ Twiter/X: https://x.com/mreflow 🧡 Threads: https://www.threads.net/@mr.eflow 🟦 LinkedIn: https://www.linkedin.com/in/matt-wolfe-30841712/ Resources From Today's Video: DeepSeek Stock Impact: https://www.cnn.com/2025/01/27/tech/deepseek-stocks-ai-china/index.html Andreessen DeepSeek Praise: https://x.com/pmarca/status/1882719769851474108 NVIDIA Market Analysis: https://x.com/mreflow/status/1883975293683351722 DeepSeek V3 Paper: https://github.com/deepseek-ai/DeepSeek-V3/blob/main/DeepSeek_V3.pdf DeepSeek Research: https://arxiv.org/pdf/2501.12948 DeepSeek R1 Guide: https://www.datacamp.com/blog/deepseek-r1 DeepSeek Homepage: https://www.deepseek.com DeepSeek Signup Issues: https://www.businessinsider.com/deepseek-issues-account-registration-limited-china-phone-numbers-malicious-attack-2025-1 Chinese AI Context: https://x.com/hxiao/status/1882649545982849031 NVIDIA Stock Drop: https://www.investopedia.com/nvidia-stock-worst-day-since-2020-deepseek-ai-wall-street-8780814 Market Impact Analysis: https://www.investing.com/news/stock-market-news/nvidia-sheds-589-billion-in-market-cap-on-deepseek-concerns-most-in--history-3831096 DeepSeek Training Discussion: https://manifold.markets/AndrewPeterson960c/did-deepseek-lie-about-the-number-o AI Compute Demand: https://x.com/garrytan/status/1883037020207153580 Jevons AI Paradox: https://x.com/satyanadella/status/1883753899255046301 Jevons Definition: https://en.m.wikipedia.org/wiki/Jevons_paradox Janus Model Release: https://x.com/rowancheung/status/1883917681642070282 Janus Technical Report: https://github.com/deepseek-ai/Janus/blob/main/janus_pro_tech_report.pdf Let’s work together! - Brand, sponsorship & business inquiries: [email protected] #AINews #AITools #ArtificialIntelligence Time Stamps: 0:00 Intro 1:02 Explaining DeepSeek-V3 4:11 Explaining DeepSeek-R1 6:37 Why People Are Freaking Out 8:18 DeepSeek is a Side-Project 9:09 Counter-Arguments To Freak Out 13:00 Jevon's Paradox 15:12 How To Use DeepSeek Yourself (Plus Examples) 20:41 DeepSeek Janus-Pro 22:35 Final Thoughts

Top Comments (10)

@sgjohnlim 2025-01-28

Thank you US for forcing the Chinese to think out of the box with lousier GPU. Love the competition, as consumers get better pricing.

1.3k 129 replies
@BrooksCardenas 2025-01-29

DeepSeek’s claims are definitely shaking things up! If true, it could signal a major shift in the AI landscape. However, I’m curious if this is just a short-term reaction or if it could have lasting effects on the market. What’s your take on its long-term impact?

937 8 replies
@austenlucksted 2025-01-28

So basically a side hustle from a few Asian crypto bros tanked our stock market.

869 56 replies
@superfliping 2025-01-28

So happy for Deepseek R1. Open source is growing. No more full corporate control.

741 31 replies
@YongLi-np3wg 2025-01-28

Happy Chinese New Year and enjoy the free gift!

520 14 replies
@Lenzspot 2025-01-28

Deep Seek has everybody in Deep Thought about the future of Ai

357 16 replies
@vainezaiven6677 2025-01-28

This is exactly what the AI space needing--a real kick in the pants to keep the big companies nervous. We need to champion open-source models wherever they come from.

277 2 replies
@FullFinnoy 2025-01-28

DeepSeek has changed this way of thinking: while the model's size is impressive (671 billion parameters), only the necessary 37 billion are activated at a time. It's like having a massive team of experts where only the right people are called upon to handle a specific task. Why does this matter? Training costs: Traditionally, training models has cost up to $100 million. DeepSeek's approach? Just $5 million. Computing power: The required number of GPUs has dropped from 100,000 to just 2,000. API costs: 95% cheaper than before. Hardware requirements: Models can now run on standard gaming PCs, no longer just on expensive data center equipment. This means that AI development is no longer exclusively in the hands of large tech companies. Now, even small teams and businesses can develop and utilize cutting-edge AI without a billion-dollar budget.

186 10 replies
@flickwtchr 2025-01-28

China didn't impose the restrictions, the US imposed export restrictions on the advanced GPUs. Kind of a big difference there.

169 12 replies
@craigrussell7542 2025-01-28

The tech bros are about to find out how much empathy the world that they exploit has for their sorry asses.

161 9 replies

Unlock the Data Inside
Turn Videos into Knowledge

  • Get FREE 10/day: transcripts, summaries, chats
  • Chat with videos, export text & PDF
  • $1 free API credit for RAG, chatbots & research

Free forever plan β€’ All features unlocked

App screenshot