DeepSeek breakthrough raises AI energy questions

Energy question?

In fact DeepSeek require less energy and data center needs.

https://www.reuters.com/business/energy/us-power-stocks-plummet-deepseek-raises-data-center-demand-doubts-2025-01-27/

A truly impressive achievement for DeepSeek CEO Liang Wenfeng, a hedge fund manager,

At the faction of the cost of any of DeepSeek rivals

A wake up call indeed

The firm says it developed its open-source R1 model using around 2,000 Nvidia chips, just a fraction of the computing power generally thought necessary to train similar programs.

Another claim, quote....

The Chinese firm claims it took just two months and cost under $6 million to build an AI model using Nvidia's less-advanced chips, and launched its iPhone app in the US last week.

Microsoft AI CEO Mustafa Suleyman, Sam Altman CEO of OpenAI better be paying full attention, along with both insufferable egos.

GAFA will be checking the small print on those energy and data centre contracts.

There is no business case for AI at the investment levels they cited. Not even at 1% of those numbers. People just won't pay the extra to interact a bit differently with a computer. It is not a magic solution to everything. It has been massively oversold.

DeepSeek has saved GAFA from flushing insane amounts of cash away. They should grab the eleventh hour exit opportunity whilst they can and save the cash for their next scam.

Efficiency improvements have so far moderated consumption despite growth in data centre demand.

A free and open source gift that is also an Anti-monopoly buster. One trillion dollars was wiped out. The Nvidia six hundred billion fall is the biggest in history. This despite the sanctions and hurdles put up by the US.

The score is 1-0. US responds with 100% tariffs on TSMC chips. The chips are going to go to where the demand is anyway. The next revolution may not come from Santa Clara County.

china will put money into its own advanced chip development.

China made it cheaper. A lot cheaper. Surprise surprise. Bad for big tech but good for everyone else and the planet. Less energy needed.

Lot's of hype, but let's see how it all plays out. Here's some great insight into why it might not be as spectacular and game changing as it sounds:

https://www.youtube.com/watch?v=Jq3egQZmyas

Points raised:

• **Lack of Transparency**: Experts question DeepSeek’s claims, citing its refusal to engage with media and lack of transparency. • **Unusual Behavior**: The company restricts overseas user registrations and avoids media visits, leading to speculation about its motives. • **Government Involvement**: DeepSeek’s founder has close ties to the Chinese government, which actively promotes the company’s success. • **“Genius AI Girl” Controversy**: Lau Le, presented as a key figure at DeepSeek, faces skepticism over her credentials, with some suspecting a publicity stunt. • **Misleading Cost Claims**: DeepSeek’s stated AI development costs omit key expenses, making its affordability claims questionable. • **Propaganda Concerns**: The hype around DeepSeek is likely state-driven - China has a long history of exaggerated technological achievements. • **Security Risks**: DeepSeek’s ties to the Chinese government raise concerns about potential data sharing with authorities.

Sorry for the formatting mess

Lack of Transparency: Experts question DeepSeek’s claims, citing its refusal to engage with media and lack of transparency.

Unusual Behavior: The company restricts overseas user registrations and avoids media visits, leading to speculation about its motives.

Government Involvement: DeepSeek’s founder has close ties to the Chinese government, which actively promotes the company’s success.

“Genius AI Girl” Controversy: Lau Le, presented as a key figure, faces skepticism over her credentials, with some suspecting a publicity stunt.

Misleading Cost Claims: DeepSeek’s stated AI development costs omit key expenses, making its affordability claims questionable.

Propaganda Concerns: The hype around DeepSeek is likely state-driven, with a history of exaggerated technological achievements.

Security Risks: DeepSeek’s ties to the Chinese government raise concerns about potential data sharing with authorities.

(I know this is an oxymoron) but people should calm down and not to hype stuff.

This is a Chinese product... and everyone should know that China is good making "shiny new things". But you just have to scratch it a little and the rust, the cheap and outdated technology, the hollowed systems and the corrupted conections with the Chinese governmet will come out soon (it has already begun)

What the market did (as always) is to freak out at the new shiny thing, drop the "old" thing that you knew it is safe (in this case NVIDIA) and buy the hype. With this only making it easier to the people who know and have the resources to buy the good thing while the masses are dreaming

China really burst their little AI bubble bigtime. As a prominent commentator has noted, this is yet another Sputnik moment.

China really burst their little AI bubble bigtime. As a prominent commentator has noted, this is yet another Sputnik moment.

We'll see

Please try and see whether ChatGPT and the other US AI apps collects info on you. Oh wait, you can't. Everything's closed off. They hiding something?

I'm a cynic in the sense that I firmly believe that whatever clever technology we come up with, we'll eventually end up using it to do terrible things.

So if we're going to keep trying to do that, we might as well try and make it a bit less environmentally destructive.

Many Ai industry ecosystem leaders are sceptical into the Chinese start-up DeepSeek's new chatbot claims, the government of China veracity, also the privacy and security of how/where DeepsSeek user data will be stored.

It is a shame, Ai development is screaming out for a competitive market model.

Not to be consigned used as a political football.

DeepSeek = Deep Thief hackers stole information to gain advantage, the Chinese do as they always do pilfer other countries technology to boost their own.