Join our daily and weekly newsletters to get the latest updates and exclusive content on our industry-leading AI coverage. He learns more
If you haven’t heard, there’s a new AI star in town: DeepsecThe subsidiary of Hong Kong-based quantitative analysis firm Quantum sent shockwaves throughout Silicon Valley and the wider world with the release earlier this week of a new open source thinking model, Deepseek R1, that matches the most powerful Openai model available, the O1. – And at a fraction of the cost to the users and the company itself (when trained).
Although the advent of Deepseek R1 has already rewired a diverse, fast-moving, intensely competitive market for new AI models – the previous months had seen Openai Jockeing with Anthropore and Google for the most powerful proprietary models available, while their meta platforms often came with it.” Its open source competitors are open enough – the difference this time is that the company behind the hot model is based in China, geopolitically “geopolitical”Tap“From the United States, whose technology sector has been widely viewed, up to this point, as inferior to Silicon Valley.
As such, it does not cause a lack of manual and existential technology by the US and Western countries, who are suddenly skeptical of Openai and the general general technical strategy of throwing more money and more (used to train AI models) towards the problem of inventing more powerful models.
However, some Western tech leaders have had a largely positive public response to DeepSeek’s rapid rise.
Marc Andreessen, co-inventor of the pioneering web browser Mosaic, co-founder of Netscape Browser and current General Partner at Famous Andreessen Horowitz (A16Z) venture capital company, Posted on X Today: “Deepseek R1 is one of the most amazing and impressive breakthroughs I have ever seen – and as an open source, Deepseek’s gift to the world.” [robot emoji, salute emoji]”
Yann LeCun, senior AI scientist in META’s Artificial Intelligence Research (FAIR) division, posted on LinkedIn account:
“For people who see Deepseek performance and think:
“China overtakes the United States in artificial intelligence.”
You are reading this wrong.
The correct reading is:
“Open source models go beyond proprietary models.”
Deepseek has taken advantage of open research and open source (such as Pytorch and Llama from Meta)
They come up with new ideas and build them on top of the work of others.
Since 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, Meta AI’s founder and CEO, seemed set to counter Deepseek’s rise with… Private post on Facebook Llama promises that a new version of the Facebook Open Source Model Family will be the “leading model of the art” when it is released sometime this year. As he put it:
“This will be a crucial year for Amnesty International. In 2025, I expect Meta AI to be the leading assistant serving over a billion people, Llama 4 will become the premier model of the art, and we will build an AI engineer who will begin to contribute increased code to our R&D efforts. To power this, Meta is building a 2GW+ data center so large that it covers a large portion of Manhattan. We will bring online ~1GW of COMPUTE in ’25 and will end the year with over 1.3 million GPUs. We plan to invest between $60 and $65 billion in Capex this year while growing our AI teams significantly, and we have the capital to continue investing in the years ahead. This is a tremendous effort, and over the coming years, it will drive our core products and core businesses, unlock historic innovation, and expand America’s technology leadership. Let’s go build!“
He even shared a graphic showing the Gigawatt data center mentioned in his post in Manhattan:
Clearly, even as he espouses a commitment to open source AI, Zuck isn’t convinced that Deepseek’s approach of improving efficiency while leveraging far fewer GPUs than major labs is the right fit for the description, or for the future of AI.
But with US companies raising and/or spending record sums on new AI infrastructure, and many experts noting a rapid decline (due to hardware/chip and software developments), the question remains which future will ultimately win out to become the world’s dominant AI provider. Or maybe it will always be a large selection of models each with a smaller market share? Stay tuned, because this competition is getting closer than ever.
