5D·

The great deepseek scam?

I would like to share with you my theory on the deepseek issue.

AI has the potential to turn the current world order upside down. Computing power is the new oil of the 21st century. Until now, everyone was firmly convinced that this was associated with high acquisition costs (R&D, infrastructure...). In order to extend its lead in the field of AI, the USA announced the 500 billion dollar Stargate project project. This represents an unprecedented level of investment that could cause other regions of the world to fall behind. other regions of the world fall behind.

Now my theory comes into play. Only days after the announcement of this investment package, news suddenly spread about a Chinese AI start-up that is supposedly as good as OpenAI, but can do the whole thing for virtually nothing. The information on the 6 million dollar development costs is as questionable as it is unverifiable.

The direct consequence of this news is that major investments in AI are being questioned. Why invest 500 billion in AI when you only need 6 million? The Chinese government with its puppet deepseek has now achieved its goal. They want to blow up Stargateblock it or simply take the wind out of its sails. SoftBank $9984 (+0,97 %), OpenAI and Oracle $ORCL (+0,01 %) will not be deterred from the project, but potential investors such as Apollo, Brookfield or sovereign wealth funds from the Middle East could be unsettled and reconsider their investments. be unsettled and reconsider their investments.

Meanwhile subsidizes China continues to secretly subsidize its own AI industry with billions. As private investors are unwilling or unable to invest so much money in China, they have resorted to this perfidious strategy in times of the cold war of technology probably the new normal.

What do you think of my theory? Or do you think it was all just a coincidence, there is no intention behind it and deepseek is a great thing?

10 Comentarios

Meister L.@Lumimyrsky

I see it similarly. A model that is cheaper or more efficient would not be so unlikely. But 6 million compared to 100 billion sounds very questionable.

Also the fact that allegedly only 2000 Nvidia chips were used with 5nm is simply strange

•

BamBamInvest@BamBamInvest

@Lumimyrsky I agree and the Chinese will then make it available to the whole world 😂 if it were that easy ....

•

Timo@Aktienorang-Utan

The Chinese just want to take the air out a bit 😮‍💨

•

Tenbagger2024@Tenbagger2024

Of course, something like this is also politically instrumentalized. Just look at how many substantial subsidies have flowed into this start-up. Something like this would be impossible in Germany. But it also shows us how important the topic of education is. And how important it is to support creative minds.

•

Alexxela@Alexxela

"Some analysts and developers are also questioning Deep Seek's total development costs, as the $6M noted in its research paper refers specifically to chip usage on its final training run." - which from other articles is actually just the tip of the iceberg and the cost to get to this point should be exponentially greater.

•

Philip S.@phswiss

I feel none of us knows. But the USA's dominance in this area should be a concern for everyone, so I welcome any disruptor. Ideally European. 😉

•

PikaPika0105@PikaPika0105

@phswiss European would be great 🥹 I would also be ok with a Japanese or Indian one. But China and the US nahhh

••

Charmin@Charmin

Quantum leaps are not unusual for young technologies. It should be quite easy to verify whether the costs are correct, as the project is open source. If there was something significantly wrong, someone would probably have cried out long ago.

••

Alexxela@Alexxela

@Charmin I'm not an IT specialist, but with open source you can see the programming behind it and not which hardware was used in the training and all previous phases and how long the training lasted, etc., right?

•

Charmin@Charmin

@Alexxela But I can train with my hardware myself and then see whether it really is less time-consuming.

••