What does ai really mean for the market

Open Editor

Rout Khalaf, FT Editor, chooses stories that he likes to this booklet.

The most important situation in the finances over the past 40 is the emergence of successful buildings – all of the technology. We live in the World of Network and Increasing Compensation to the point. In factory in the PC Software (Microsoft), Internet Research (15 is the best thing to say all the restrictions of investments over the last 10 years – all stories.

It is only natural that everyone thinks, desiring or being afraid that the largest Ote technology, intelligence, would have economic skills. Something has happened in yesterday market was not another area of ​​surprise, or shocking, or bubble pops. Were values ​​in a bit Ai is unusual if ai is not successful – take the whole game.

It was often thought that AI has turned to be like the design of a car or aircraft. Such peace leads to the creation of large, virtuous dignified, but so much competing, to ensure that the most drained value is to clients. Now looks like a very important thing in AI.

Most of the students know the news now. Develeek, Chinese company for a AI model called AI called R1, compared to good models from companies such as Openai, Anthropic and Anthropic and ATHRAPIn, but used under a subordinate government and is used below the State-on- Art GPU pips. Deveeek has been adequate to the public information that can be able to run out of their own computers (but it does not have to create it).

This placing three tongels in a ai Ai. Toss the idea that the best AI results can only be achieved with the relevant Nvips chips and software; The idea that the only major tech companies can cause to be built and flee the upper AI model; And the view that only companies with their AI model may provide good AI applications.

What is the nvidia nvidia nvidia in the past two years was very high. The chips was the best miles on AI. Develeek build R1 and Nvidia is old, chips slowly, if US verbs allow to be sent to China. This indicates that the door is open to some who compete to the level of silk.

Then, the cost. Major tech companies look more or less than aI, because, the only way to build a better AI model to use a lot of contributions, costly. The completed model will be under the control of a large company built. The obstacle to the amount of money – that is why the BIB is to continue the spree of data data. But R1 suggests that companies with less money can use competitive models.

Finally, applications. “Submission” is where the user entembers AI: The questions are answered, tasks completed. It was envisaged that any application depends on the models of large company and humiliary servers. But models such as R1 can work well, and adequate adequacy, running from servers for all unwanted companies “in the rent” to the openai.

All of this has great results. But without Nvidia, a great Faller yesterday was not a wonderful second technology. Those halo hay companies are usually in place. True damage is made in stock after the economic institution. Group Team, Company covering atlantic, Vista, Washington State's Mubs, and NGR, NGOs were the best texas players, all of them

Gea Vernova, Eota Services and CATEE services build systems in data centers. The Oracle has announced the largest data center. Arsom and Arista provides GPU technology in data centers. All had whad:

Amazon, letters, Meta and Microsoft all are very investing in a AI data centers. Maybe they have spent part in money – but they can be free from spending less. Apple and Microsoft, which is the potential gas, which can possibly cause ai applications for AI models instead of models on their own, can be in a better position.

NVOSTA itself is not hard for knees. The language of the coding, CUDA, is still the level of the sector. And because the scientific model is more effective does not mean that the Elenetic model will not be beneficial to the top quaccal and the best chips of the Nvidia. While its shares drop almost 17 percent yesterday, returned only (very, very) September:

We should not be over the advertisement in R1. NASDAQ fell in 3%. That was a bad day, not fear. The idea of ​​winning – all – all the pain is hurt, they don't die. AI Bubble, if that's all, it may dream, but it doesn't have to buy yesterday. Not near.

One study

The wrong hope.

robert.armstrong@ft.com and Aiden.reitar@ft.com.