Leo's Lightbulbs
Posts
Leo's Lightbulbs💡- DeepSeek are big lying liars, spent $1.6 billion and used 50,000 NVIDIA gpus to train their model

Leo's Lightbulbs💡- DeepSeek are big lying liars, spent $1.6 billion and used 50,000 NVIDIA gpus to train their model

Leo Rodman
February 10, 2025

So, as the headline says, the big news this week is that DeepSeek are big lying liars who spent $1.6 billion and used 50,000 NVIDIA gpus to train their model, then claimed they did it for 6 without a fleet of NVIDIA gpus, causing NVIDIA’s stock to take massive losses.

If this happened in the USA, DeepSeek’s executives would be going to jail for manipulating the stock market. Since it happened in China, either they’ll be let off the hook OR something really bad will happen to them. Since their actions hurt a US company, not a Chinese company or the Chinese government / national interest, I suspect they won’t be penalized.

DeepSeek might not be as disruptive as claimed, firm reportedly has 50,000 Nvidia GPUs and spent $1.6 billion on buildouts

The fabled $6 million was just a portion of the total training cost.

www.tomshardware.com/tech-industry/artificial-intelligence/deepseek-might-not-be-as-disruptive-as-claimed-firm-reportedly-has-50-000-nvidia-gpus-and-spent-usd1-6-billion-on-buildouts

Three things to know as the dust settles from DeepSeek

As the initial hype fades, here are the likely longer-term impacts.

www.technologyreview.com/2025/02/04/1110918/three-things-to-know-as-the-dust-settles-from-deepseek

DeepSeek might not be such good news for energy after all

New figures show that if the model’s energy-intensive “chain of thought” reasoning gets added to everything, the promise of efficiency gets murky.

www.technologyreview.com/2025/01/31/1110776/deepseek-might-not-be-such-good-news-for-energy-after-all

US proposes bill to criminalize DeepSeek download, up to 20 years prison, $100M fine

The future of AI is under scrutiny, with proposed laws threatening severe punishments for using Chinese AI, such as DeepSeek.

interestingengineering.com/culture/law-proposed-to-criminalize-deepseek

Anthropic has a new way to protect large language models against jailbreaks

This line of defense could be the strongest yet. But no shield is perfect.

www.technologyreview.com/2025/02/03/1110849/anthropic-has-a-new-way-to-protect-large-language-models-against-jailbreaks

Senator Hawley Proposes Jail Time for People Who Download DeepSeek

According to the language of the proposed bill, people who download AI models from China could face up to 20 years in jail, a million dollar fine, or both.

www.404media.co/senator-hawley-proposes-jail-time-for-people-who-download-deepseek

Texas governor orders ban on DeepSeek, RedNote for government devices

Texas ordered a ban on the Chinese AI startup DeepSeek on government devices days after the popular chatbot has shook investors and the tech community in recent weeks for its ability to compete with OpenAI.

apnews.com/article/texas-deepseek-apps-ban-3828a4743e9919398dfac0ba9d4a5c25

DOGE can’t use student loan data to dismantle the Education Dept., lawsuit says

Students don’t want loan data used in AI probe to slash DOE, according to lawsuit.

arstechnica.com/tech-policy/2025/02/lawsuit-tries-to-block-doges-access-to-student-loan-data

Andrew Ng is 'very glad' Google dropped its AI weapons pledge | TechCrunch

Andrew Ng, the founder and former leader of Google Brain, supports Google's recent decision to drop its pledge not to build AI systems for weapons. "I'm

techcrunch.com/2025/02/07/andrew-ng-is-very-glad-google-dropped-its-ai-weapons-pledge

ChatGPT comes to 500,000 new users in OpenAI’s largest AI education deal yet

Still banned at some schools, ChatGPT gains an official role at California State University.

arstechnica.com/ai/2025/02/chatgpt-comes-to-500000-new-users-in-openais-largest-ai-education-deal-yet

Not Gouda-nough: Google removes AI-generated cheese error from Super Bowl ad

Unlike Google search, AI writing assistant doesn’t even cite its sources.

arstechnica.com/ai/2025/02/not-gouda-nough-google-removes-ai-generated-cheese-error-from-super-bowl-ad

Anthropic CEO says DeepSeek was 'the worst' on a critical bioweapons data safety test | TechCrunch

Anthropic's CEO Dario Amodei claims DeepSeek generated sensitive bioweapons data in a safety test it ran.

techcrunch.com/2025/02/07/anthropic-ceo-says-deepseek-was-the-worst-on-a-critical-bioweapons-data-safety-test

Hugging Face clones OpenAI’s Deep Research in 24 hours

Open source “Deep Research” project proves that agent frameworks boost AI model capability.

arstechnica.com/ai/2025/02/after-24-hour-hackathon-hugging-faces-ai-research-agent-nearly-matches-openais-solution

DeepMind claims its AI performs better than International Mathematical Olympiad gold medalists | TechCrunch

An AI system developed by Google DeepMind, Google's premiere AI research lab, has surpassed the average gold medalist in solving geometry problems in an international mathematics competition.

techcrunch.com/2025/02/07/deepmind-claims-its-ai-performs-better-than-international-mathematical-olympiad-gold-medalists

Are companies falling out of love with OpenAI?

Equity is getting into Figure Robotics' move to ditch OpenAI for its own AI, plus the latest on startup hires, space telescope collabs, and Elon Musk’s latest moves.

techcrunch.com/podcast/are-companies-falling-out-of-love-with-openai

Report: Ilya Sutskever's startup in talks to fundraise at roughly $20B valuation | TechCrunch

The startup founded by ex-OpenAI chief scientist Ilya Sutskever is reportedly in talks to raise funding at a valuation of "at least" $20 billion.

techcrunch.com/2025/02/07/report-ilya-sutskevers-startup-in-talks-to-fundraise-at-roughly-20b-valuation

PromptLayer is building tools to put non-techies in the driver’s seat of AI app development | TechCrunch

The GenAI boom of the last few years has unleashed a wave of startups promising to support the process of prompt engineering -- i.e., coming up with

techcrunch.com/2025/02/07/promptlayer-is-building-tools-to-put-non-techies-in-the-drivers-seat-of-ai-app-development

An AI chatbot told a user how to kill himself—but the company doesn’t want to “censor” it

While Nomi's chatbot is not the first to suggest suicide, researchers and critics say that its explicit instructions—and the company’s response—are striking.

www.technologyreview.com/2025/02/06/1111077/nomi-ai-chatbot-told-user-to-kill-himself

Four Chinese AI startups to watch beyond DeepSeek

These companies are all in a position to compete both within China and beyond.

www.technologyreview.com/2025/02/04/1110942/four-chinese-ai-startups-deepseek

Reply

or to participate.