Lohnt es sich, in KI zu investieren, nachdem Deepseek den Markt auf den Markt gebracht hat? – Nach der Ankunft eines kostspieligen KI -Modells aus China ringen US -amerikanische Märkte und Akademiker mit dem endgültigen wirtschaftlichen Wert der Technologie.

From the article

>A.I. chatbots are fun, sometimes even useful and, until recently, endowed with the uncanny ability to mesmerize investors and fuel the U.S. stock market.

>But the excellent performance of a new, relatively cheap artificial intelligence engine from a Chinese start-up, [DeepSeek](https://www.nytimes.com/2025/01/27/technology/what-is-deepseek-china-ai.html), has perturbed the market and complicated the A.I. story.

>Investors are re-evaluating prominent companies swept up in A.I. fever, including [Nvidia](https://www.nytimes.com/2024/09/13/business/nvidia-stock-market-ai.html), [Meta](https://www.nytimes.com/2025/01/30/business/dealbook/meta-microsoft-ai-spending-deepseek.html), Alphabet, Microsoft, Amazon, Tesla and the private start-up OpenAI. The notion that full-blown superhuman intelligence is imminent has spurred the-sky-is-the-limit valuations, as well as concerns about the political and social [risks](https://www.nytimes.com/2023/05/30/technology/ai-threat-warning.html) posed by advanced intelligence.

>One immediate question: Is the main approach to developing A.I. in the United States — pouring billions of dollars into chips and infrastructure — worth the expenditure for all companies if similar results can be achieved far more cheaply? [DeepSeek’](https://www.nytimes.com/2025/01/28/technology/china-deepseek-ai-silicon-valley.html)s lower-cost innovations add urgency to bigger, longstanding financial questions: How much are artificial intelligence companies really worth, and what will the broader economic value of A.I. ultimately be?

DeepSeek requiring less money to train algorithms is great (if, and only if that’s found to be replicated by a third party). But the fact of the matter is that the bulk of the processing that data centres are used for is INFERENCE not TRAINING.  So DeepSeek or not, it makes no difference to companies like NVDA.