Join our daily and weekly newsletters for the latest updates and exclusive content on AI coverage. Learn more
If you hadn’t heard, there’s a new AI star in town: In depththe subsidiary of Hong Kong Analysis (quant) Firm High-Fly Flyer Capital Management, sent shockwaves throughout Silicon Valley and the wider world with its release earlier this week of a new open source reasoning model, Deepseek R1which corresponds to the most powerful Openai available model O1 – and at a fraction of the cost for users and for the company itself (during training).
While the advent of Deepseek R1 has already reshuffled a very dominant, fast moving and intensely competitive market for new AI models – the previous months have seen Openai jockey with Anthropic and Google for the most powerful proprietary models available, while Meta Platforms often came up with “Pretty close open source rivals – The difference this time is the company behind the hot model is based in China, the geopolitical”Fremy“From the United States, and whose technology sector has been widely considered, until this moment, to be inferior to that of Silicon Valley.
As such, there has been no shortage of hand-wringing and existentialism from Western bloc tech, who are suddenly doubting Openai and Big Tech’s general strategy to throw more money and more compute ( graphics processing units, GPUs, powerful gaming chips in general Used to train AI models) towards the problem of inventing more and more powerful models.
Still, some Western tech executives have had a largely positive public response to Deepseek’s rapid rise.
Marc Andreessen, co-inventor of the pioneering Mosaic web browser, co-founder of the NetScape Browser Company and current general partner at the famous Andreessen Horowitz (A16Z) venture capital company, Published on x today: “Deepseek R1 is one of the most amazing and impressive breakthroughs I have ever seen – and as open source, a profound gift to the world [robot emoji, salute emoji]. “
Yann LeCun, the principal AI scientist for META’s Fundamental AI Research (Fair) division, published on his LinkedIn account:
“To the people who see Deepseek’s performance and think:
“China overtakes the United States in AI.”
You are reading this wrong.
The correct reading is:
“Open source models outpace proprietary ones.”
Deepseek took advantage of open search and open source (e.g. Pytorch and Meta’s Llama)
They came up with new ideas and built them on top of the work of others.
Because their work is published and open source, everyone can benefit from it.
This is the power of open research and open source. »
And even Mark “Zuck” Zuckerberg, founder and CEO of Meta Ai, seemed to seek to counter the rise Own Facebook post Promising that a new version of Facebook’s open source Family Llama Model would be “the leading edge model” when it comes out later this year. As he said:
“This will be a defining year for AI. In 2025, I expect that Meta AI will be the main assistant that will serve more than a billion people, Llama 4 will become the main cutting-edge model, and we will build an AI engineer that will start contributing more and more more code to our R&D efforts. To power this, Meta is building a 2GW+ data center that is so large it would cover a significant portion of Manhattan. We will bring ~1 GW of compute online in 2000 and end the year with over 1.3 million GPUs. We plan to invest $60-65 billion in CAPEX this year while significantly growing our AI teams, and we have the capital to continue investing in the years to come. This is a massive effort, and over the coming years it will boost our core products and businesses, unlock historic innovation, and expand America’s technological leadership. Let’s go build!“
He even shared a graphic showing the 2 Gigawatt Datacenter mentioned in his article superimposed on Manhattan:
Clearly, while he espouses a commitment to open source AI, Zuck isn’t convinced that Deepseek’s approach to maximizing efficiency while pulling far fewer GPUs than the big labs is the right one for the meta, or for the future of AI.
But with US companies increasing and/or spending registration monies New AI infrastructure As many experts have noticed quickly (due to advancements in hardware/chips and software), the question remains which vision of the future will ultimately win to become the dominant AI provider for the world. Or maybe it will always be a multiplicity of models with a smaller market share? Stay tuned, because this competition is getting closer and fiercer than ever.
