(Image source: Photo.com)
With the rapid development of artificial intelligence (AI) technology, its demand for computing power and power has also increased exponentially.
In a recent YouTube interview, Meta CEO Mark Zuckerberg expressed concern about the power supply of AI data centers.
In an interview with Dwarkesh Patel's channel, Zuckerberg pointed out that although the difficulties encountered by IT companies in purchasing AI graphics processing units (GPUs) have eased recently, the power problem may become the next bottleneck in AI development. In particular, he mentioned that the construction of 1GW of data centers – the equivalent of a nuclear power unit – is not currently realistic because of the increasingly stringent management of the energy sector in various countries, which means that the approval of supporting energy facilities (including power stations, substations, transmission systems) for large data centers is slower. The approval process is complex.
Zuckerberg's concerns are not unfounded. With the widespread adoption of AI technology, especially in areas such as data centers, large-scale computing, and smart grids, the demand for electricity has surged. OpenAI's ChatGPT and other AI products are big consumers of electricity, consuming more than 500,000 kWh of electricity per day, which is equivalent to the average daily electricity consumption of 17,000 ordinary American households. This power consumption is undoubtedly a huge challenge for the existing power system.
Musk has expressed similar concerns on different occasions. He stressed that with the continuous advancement and widespread application of AI technology, there may be a shortage of electricity supply. He called attention to energy issues and suggested more effective energy use and management measures to address the power challenges posed by AI.
The artificial intelligence industry is one of the most forward-looking and innovative fields in the world today, and its rapid development covers machine learning, deep learning, natural language processing and other fields, bringing great change and innovation to all walks of life.
In the future, with the continuous progress of AI technology and the continuous expansion of application scenarios, power problems will become one of the key factors restricting the development of AI.
The R&D intensity of Chinese artificial intelligence enterprises far exceeds the average level of domestic enterprises
According to the "China New Generation Artificial Intelligence Technology Industry Development Report 2022", 197 listed AI companies in China have announced R&D investment data. Among them, the number of enterprises with R&D intensity between 5% and 10% is the largest, accounting for 36.35%; The R&D intensity of enterprises greater than 10% accounted for 33.5%, and the average R&D intensity reached 9.98%, far exceeding the average level of domestic enterprises.
Half of the world's AI companies are still unprofitable
According to the Global Enterprise Database, a total of 1,540 companies have disclosed their profit margins, and from the perspective of the distribution of profit margins, there are currently 610 companies in the profit range of 0-20%, and 798 companies with less than 0% profit margins, accounting for 51.8% of the total number of companies with published profit margins.
Technical analysis of the core layout of Chinese artificial intelligence enterprises
From the perspective of the distribution of core technologies of artificial intelligence enterprises, big data and cloud computing account for the highest proportion, reaching 41.13%. followed by hardware, machine learning, recommender systems and service robots, accounting for 7.64%, 6.81%, and 5.64%, respectively. Among the subsequent technologies, the Internet of Things, industrial robots, speech recognition, natural language processing, and graphic image recognition technologies accounted for 5.55%, 5.47%, 4.76%, and 4.72% respectively.
Bank of America's latest research report pointed out that with the rapid development of artificial intelligence technology, the power demand of the U.S. power grid will face unprecedented growth. Electricity demand in the U.S. is expected to grow at a compound annual growth rate (CAGR) of 2.8% between 2024 and 2030. BofA believes that this growth is mainly driven by four factors: the expansion of AI data centers, industrial growth, the adoption of electric vehicles, and the electrification of buildings, with the power demand of data centers being particularly prominent.
ARM CEO Rene Haas said that AI models such as OpenAI's ChatGPT require a lot of electricity. He noted that by 2030, AI could account for 20 to 25 percent of U.S. electricity demand, up from less than 4 percent today. Haas noted that these AI models require a lot of sustained power to operate effectively, and as the AI gathers more information, it becomes smarter and consumes more power.
Prospective Economist APP Information Group
For more research and analysis of this industry, please refer to the "Chinese Artificial Intelligence Industry Development Prospect Forecast and Investment Strategic Planning Analysis Report" of the Prospective Industry Research Institute
At the same time, the Prospective Industry Research Institute also provides solutions such as industrial big data, industrial research reports, industrial planning, park planning, industrial investment, industrial mapping, smart investment promotion system, industry status certificate, IPO consulting/fundraising and investment feasibility study, and specialized and special new small giant declaration. To quote the content of this article in any public information disclosure such as prospectus and annual report, formal authorization from Qianzhan Industry Research Institute is required.
More in-depth industry analysis is available in the [Prospective Economist APP], and you can also communicate and interact with 500+ economists/senior industry researchers. More enterprise data, enterprise information, and enterprise development are all in the [Qichamao APP], the most cost-effective and most comprehensive enterprise query platform.