Join our everyday and weekly newsletter for the latest updates and exclusive content for top AI coverage in the field. More information
If you haven’t heard, in the city there is a new star AI: Deepseek, a subsidiary company for quantitative analysis (TA) in Hong Kong (TA) Management Capital Management, has the scent of shock waves throughout Silicon Valley and wider world with its release new open -Source of a large model of reasoning, Deepseek R1, which corresponds to the strongest available O1 Openi and for a fraction of the cost of the user and society itself (during its training).
While the arrival of Deepseek R1 has already moved consistent topsy-turvy, fast-moving, intensively Compeittany Market for new AI-pre-established months saw Openai Jockey with anthropic and Google for the strongest proprietary models available while the meta platform with closed “closed” Opponents with an open source code-a time the difference is that the company is a hot model in China, the geopolic “fremy” of the US and whose technological sector was considered Silicon Valley at this point.
As such, this did not cause a lack of hand and existentialism from American and western block technicians who suddenly doubt Openai and general big-tech strategy of throwing more money and computing units (graphics, GPU, strong playing, strong games, strong chips are usually used to train models AI) towards the problem of inventing ever stronger models.
Yet some Western technology leaders had a large extent a positive public response to Deepseek’s rapid ascent.
Marc Andreessen, co -authors of a pioneering mosaic browser, co -founder of Netscape browser and current general partner in the famous company Andreessen Horowitz (A16Z), published on X today: for the world (Robot Emoji, greeting Emodji).
Yann Lecun, the main scientist AI for the Meta Fundamental AI Research (FAIR) division, published on his LinkedIn account:
“For people who see Deepseek’s performance and think:
“China overcomes the US in AI.”
You’re reading it wrong.
The correct reading is:
“Models with open source code overcome proprietary.”
Deepseek takes advantage of open research and open source (eg Pytorch and Lama of Meta)
They came up with new ideas and built them on other people’s work.
Because their work is published and an open source, each can benefit each.
This is the power of open research and open source. ”
And even Mark “Zuck” Zuckerberg, founder and CEO of META AI, seemed to see Deepseek’s rise with his own Facebook post, which promises that the new version of the Open-source AI model Family Llam would be the “main state of art” , when it is released sometime this year. As he said:
“” “”For AI it will be a defective year. In 2025, I expect Meta AI to be a leading assistant who serves more than 1 billion people, Llama 4 will become the main state of the art model and build engineer AI, who will contribute to the growing Amants of Code to our RR & D efforts. To power it, meta builds 2GW+ data centers, which is so large that it would cover a significant part of Manhattan. We will bring online ~ 1GW calculation in ’25 and end with more than 1.3 million GPUs. This year we plan to invest in Capex 60-65B in Capex and at the same time the average growth of our AI teams, and we have capital to continue investing in the coming years. This is a huge effort and will manage our basic products and business in the coming years, unlock historical innovations and expand the American technological leadership. Let’s build!“” “”
He even shared graphics showing data centers 2-Gigawatt listed in his post on Manhattan:
Obviously, even if it espies an obligation to open-source AI, Zuck is not reasonable that Deepseek’s approach to optimizing efficiency and use of much less GPUs than the main laboratory is for meta or the future of AI.
But with American companies that increase and/or spend record amounts on the new AI infrastructure, which many experts have marked quickly (because of hardware/chip and software progress) remain a question that the vision of the future will eventually win to become becomes a dominant AI provider for the world. Or maybe there will always be many models with a smaller market share? Stay tuned because this competition is coming and raging than ever.
