Useful information
Prime News delivers timely, accurate news and insights on global events, politics, business, and technology
Unique our daily and weekly newsletters to obtain the latest updates and exclusive content on the coverage of the leader in the industry. More information
If you hadn’t heard it, there is a new AI star in the city: Deep searchThe subsidiary of the quantitative analysis firm High-Flyer Capital Management based in Hong Kong, has caused shock throughout Silicon Valley and worldwide with the launch in early this week of a new model of large reasoning reasoning model Open, Deepseek R1, which coincides with the most powerful available OPENAI O1 model, and a cost fraction for users and for the company itself (by training it).
While the arrival of Deepseek R1 has already reorganized a constantly scrambled market, of rapid movement and intensely competitive for new AI models, in the previous months OpenAi competed with Anthropic and Google for the most powerful owner models available, while the metaplataformas to They often arrived with “open source rivals close enough; the difference this time is that the company behind the fashion model is based in China, the geopolitical”enemy friend”From the United States, and whose technological sector was widely considered, so far, as lower than that of Silicon Valley.
As such, it has caused many handshakes and existentialism by the US technicians and the western block, which are suddenly doubting OpenAi and the general strategy of large technologies to spend more money and more computing ( Graphics processing units, GPU, powerful games of games typically). used to train AI models) towards the problem of inventing increasingly powerful models.
However, some Western technological leaders have had a positive public response to the rapid promotion of Deepseek.
Marc Andreessen, co-inventor of the pioneer Mosaic web browser, co-founder of the Netscape browsers and current general partner of Famous Andreessen Horowitz (A16Z) risk capital company, Posted in X today: “Deepseek R1 is one of the most surprising and impressive advances that I have ever seen and, as an open source, a deep gift for the world (robot emoji, greeting emoji).”
Yann Lecun, head scientist of the Fundamental Research Division of the Meta (Fair), published in his LinkedIn account:
“For people who see Deepseek’s performance and think:
“China is surpassing the United States in AI.”
You are reading this badly.
The correct reading is:
“Open source models are surpassing the owners.”
Deepseek has benefited from open and open source research (for example, Pytorch and Meta flame).
They came up with new ideas and built them on other people’s work.
Because his work is published and open source, everyone can benefit from him.
That is the power of open research and open source. ”
And even Mark “Zuck” Zuckerberg, founder and executive director of Meta AI, seemed to try to counteract Deepseek’s boom with his own publication on Facebook Promising that a new version of the family of open source AI models, Llama, would be “the latest generation leader” when it is launched at some point in this year. As he said:
“This will be a decisive year for AI. In 2025, I hope that Meta AI is the leading assistant who provides services to more than one billion people, calls 4 becomes the latest leading model and we will create an artificial intelligence engineer who will begin to contribute with increasing amounts of Code to our research and development. efforts. To boost this, Meta is building a data center of more than 2 GW that is so large that it would cover an important part of Manhattan. We will put online ~ 1gw computing in 25 and we will finish the year with more than 1.3 million GPU. We are planning to invest between 60,000 and 65,000 million dollars in capital expenses this year and, at the same time, significantly increase our AI teams, and we have the capital to continue investing in the coming years. It is a huge effort and, in the coming years, will boost our main products and businesses, unlock historical innovations and expand US technological leadership. Let’s build!“
He even shared a graphic that shows the 2 gigaw data center mentioned in its superimposed publication in Manhattan:
Clearly, even when defending a commitment to open source AI, Zuck is not convinced that Deepseek approach to optimize efficiency and take advantage of many less GPUs than the main laboratories is correct for the finish line or for the future of AI.
But with US companies raising and/or spending records in new infrastructure of AI that, according to many experts, depreciates quickly (due to the advances of hardware/chip and software), the question remains what vision of the future will win in the end To become the best. Dominant supplier in the world. Or maybe there will always be a multiplicity of models, each with a smaller market share? Be attentive, because this competition is increasingly at odds and fierce than ever.
Source link
