Mistral board member Anjney Midha recently shared more details about how he had witnessed the breakthrough performance by DeepSeek six months back. He also admitted that this is just the start of the tech world's hunger for AI GPU.
At the time, DeepSeek rolled out its Coder V2 which served as a rival to OpenAI’s GPT-4 Turbo for several different types of coding tasks. This really put the firm on a new path of success to roll out newer and more advanced models in a few months. This was right through R1 which is said to be its latest breakthrough open source reasoning model that shocked the tech world. After all, seeing the launch of a model that beats all industry standards at a fraction of the cost is major development.
Despite all the major changes seen, the R1 will not prevent other leading AI foundation models from spending major sums of money to eat more GPU chips and give rise to more data centers at a fast pace.
The competition in the world is already fierce and this just motivates them to do more with all the computing power they can get. The goal is to get more efficient output from the same level of computing.
This does not mean Mistral is behind rivals such as Anthropic and OpenAI. Each one of those continues to raise more money than Mistral. We’ve just seen OpenAI get more investments worth a staggering $40B.
Today, Mistral is fighting head-to-head with them as it’s open source. This gives the company full access to free-of-cost tech labor from anyone who wishes to help as they use this project. Meanwhile, closed-source rivals keep secrets guarded and have to pay for any labor as well as computing power.
You do not really need $20B but what you do need is more computing than any open source model application. Hence, the firm is in the driving seat and they have the most compute than any other provider for open source tech today.
We see Facebook’s Llam as one of the biggest open-source models in the Western world, serving as a serious rival for Mistral. It would also get a lot of investments as stated by Zuckerberg recently. The company hopes to spend hundreds of billions on AI this year, making the company’s vision come to life. The figures quoted are estimated to be $60B this year alone on capital expenses such as data center creation.
Image: DIW-Aigen
Read next: Meta's WhatsApp Uncovers Israeli Spyware Hacking Civil Society Members
At the time, DeepSeek rolled out its Coder V2 which served as a rival to OpenAI’s GPT-4 Turbo for several different types of coding tasks. This really put the firm on a new path of success to roll out newer and more advanced models in a few months. This was right through R1 which is said to be its latest breakthrough open source reasoning model that shocked the tech world. After all, seeing the launch of a model that beats all industry standards at a fraction of the cost is major development.
Despite all the major changes seen, the R1 will not prevent other leading AI foundation models from spending major sums of money to eat more GPU chips and give rise to more data centers at a fast pace.
The competition in the world is already fierce and this just motivates them to do more with all the computing power they can get. The goal is to get more efficient output from the same level of computing.
This does not mean Mistral is behind rivals such as Anthropic and OpenAI. Each one of those continues to raise more money than Mistral. We’ve just seen OpenAI get more investments worth a staggering $40B.
Today, Mistral is fighting head-to-head with them as it’s open source. This gives the company full access to free-of-cost tech labor from anyone who wishes to help as they use this project. Meanwhile, closed-source rivals keep secrets guarded and have to pay for any labor as well as computing power.
You do not really need $20B but what you do need is more computing than any open source model application. Hence, the firm is in the driving seat and they have the most compute than any other provider for open source tech today.
We see Facebook’s Llam as one of the biggest open-source models in the Western world, serving as a serious rival for Mistral. It would also get a lot of investments as stated by Zuckerberg recently. The company hopes to spend hundreds of billions on AI this year, making the company’s vision come to life. The figures quoted are estimated to be $60B this year alone on capital expenses such as data center creation.
Read next: Meta's WhatsApp Uncovers Israeli Spyware Hacking Civil Society Members
