President and founder Emj Capital Eric Jackson analyzes the heavy AI competition, the future of Cybersecurity Stargate and Cryptocurrency Cybersecurity under Trump’s administration.
Latest Artificial intelligence (AI) The models launched by the Chinese startup Deepsek have encouraged discomfort in the technological sector after its appearance as a potential rival of leading companies based in the US.
Last month, Deepseek wrote in his work that his model Deepseek-V3 trained a computer power worth less than $ 6 million from what he says they were 2,000 NVIDIA H800 chips to achieve a level of performance with the most advanced models of Openi and Meta.
These chips are less advanced than the state -of -the -art chips on the market, which are subject to export controls, although Deepseek claims to overcome this deficiency with innovative AI training techniques. Deepseek’s ai assistant, triggering the Deepseek-V3 model, has surpassed Openai’s chatgpt As the best rated free app at the Apple App Store in the US
The appearance of Chinese companies asked questions about leading American technological companies that invest billions of dollars into advanced chips and large data centers used to train AI models. It also serves as a “moment of Sputnik” for the AI race between the US -ai China after the perception that they US had an advantage over their geopolitical rival in the emergence fields.
Chinese app Deepseek Hammers American stock with cheaper AI model open code
Aid Assistant Deepseek surpassed Openi’s Chatgpt in the Apple App Store. (Christoph Dernbach / Picture Alliance via Getty Images / Getty Images)
The quality of Deepseek’s models and reported cost effectiveness changed the narrative Chinese AI of the company They follow their American colleagues, who started after Baida published the first Chinese Chatgpt equivalent.
The Deepseek-R1 model has been released last week and 20 to 50 times is cheaper for use than OPENAI O1 models, depending on the task, according to the post on the company’s official Wechat account.
The R1 model is also an open source and is available to users for free, while the Openi -DO CHATGPT PO Plan costs $ 200 a month.
Silicon Valley Praise Chinese Ai startup Deepseek: ‘Deep Gift to the World’
Deepseek says his model has performed equally with the latest Openi and anthropic models with costs. (Getty Images / Getty Images)
Deepseek was created at the end of 2023. After controlling the shareholder Liang Weenfeng, co -founder of quantitative Hedge Fund Tall back, moved to create “a new and independent group to explore the essence [artificial general intelligence]. “
Artificial general intelligence (Agi) is also a goal that follows the Openi, which defines agi as autonomous systems that outweigh people in the most economically valuable tasks.
Get a job with Fox on a clicking movement here
“Deepseek did something really impressive with far fewer employees and financing than Openi, and in less time,” said Mark Malek, Cio of Siebert. Malek noted that Deepsek, “not competing with Openi,” and went to explain some differences between Deepsek and the more famous AI apps.
“Openai (and Google twins) is a wide general purpose tool, based on a large corpus of information. It can also be used to create closely focused apps. Openi has huge options in the processing of natural languages, while Deepsek is created to be specific to task,” he said Malek. “Now, this does not mean that Deepsek is not good. What is important to understand is that this is not the same as open, so he would logically take fewer resources.”
Reuters contributed to this report.
