If the creators of DeepSeek are not lying, then neural networks require much less energy than previously thought.
In just one night, the Chinese startup DeepSeek turned the artificial intelligence industry upside down with its new chatbot, but its main advantage is high energy efficiency. This conclusion was reached by the Oilprice publication, which focuses on energy carriers.
According to some reports, DeepSeek's capabilities are comparable to those of such high-profile and expensive megaprojects as ChatGPT. NPR believes that the Chinese chatbot “can compete” with industry leaders such as OpenAI and Google, despite the fact that it was created with lower costs and computing power. Experts have compared its entry into global markets to the launch of the Soviet “Sputnik-1” into space, which suddenly and unexpectedly overshadowed the entire US technology sector.
The financial consequences of the launch of the second DeepSeek model, R1, are already huge, as it provoked a massive sell-off in AI shares. This has reduced Nvidia's capitalization by more than half a trillion dollars — the self-proclaimed world leader in artificial intelligence computing, similar damage, albeit to a lesser extent, has been suffered by Tesla, Google, Amazon, and Microsoft.
DeepSeek's approach to developing its AI model completely overturned everything the world knew about this technology, and may change the direction of the industry forever. If previously there was an opinion that large companies would dominate this area due to the availability of free money to invest, then the Chinese startup has disproved this myth, achieving results with a very limited budget.
At the same time, high demand creates the prospect of big earnings for energy suppliers. According to media Axios, energy companies, ranging from small reactor startups to existing utilities and gas companies, consider data centers as a critically important market for the United States. DeepSeek, on the other hand, could shake it up a lot with its claims about energy efficiency, creating an uncertain situation.
“DeepSeek appears to demonstrate that training high-performance models can consume much less electricity than previously thought. At the same time, DeepSeek also significantly reduces the cost of using AI output, which could lead to consumers using it much more, which in turn leads to greater demand for electricity,” — explained John Larsen of the research firm Rhodium Group.
Predictions of the growth of AI energy demand have been vague at best, with official forecasts fluctuating over a wide range of figures. The controversy surrounding DeepSeek only highlights the significant and already existing volatility in the nascent sector, Oilprice stressed.
Even if the Galaxy S25 Edge ends up being "just" a thinner Galaxy S25, that…
Vodafone has become the first company in the world to successfully make a video call…
Belgium had to temporarily restrict access to its airspace. All due to a technical failure…
The official statement was published in Oleh Kiper's Telegram channel. Dear citizens!, – the appeal said.…
38-year-old Salvan Momika, known for burning the Quran in Sweden, was shot dead in his…
A man was detained in Pakistan who shot his 15-year-old daughter dead over her TikTok…