<p>Under the global AI boom, in addition to the focus on chip technology, its development is also dependent on power supply. The world's richest man, Elon Musk, recently stated in an interview that although the United States is currently leading in the AI competition between China and the U.S., China can narrow the gap due to its advantages in power production scale and is expected to lead the world in AI computing. In fact, a report from Goldman Sachs also confirms this statement, indicating that nearly all power grids in the U.S. face bottlenecks due to insufficient capacity, while China's reserve power capacity by 2030 will not only meet its own needs but will also exceed the expected total demand of global data centers by more than three times.</p>
<h2>Musk: China Will Conquer Chip Technology</h2>
<p>Last week, Musk stated in a podcast program called &#34;Moonshots&#34; that power supply is a limiting factor for scaling AI systems, and China's decisive advantage in the AI competition lies in its power production scale. He estimates that by 2026, China's power output will reach nearly three times that of the U.S., enabling it to support energy-intensive AI data centers.</p>
<p>Additionally, Musk believes that even if the U.S. is committed to restricting China's access to advanced semiconductors, the impact of these restrictions will weaken over time. He also mentioned that there are diminishing marginal returns in the cutting-edge field of chip performance, and China will eventually &#34;conquer chip technology.&#34; Even without the most advanced designs, China may catch up.</p>
<p>In fact, earlier, Microsoft CEO Satya Nadella also admitted that &#34;the biggest challenge currently is not computing power, but power supply.&#34; If power supply is insufficient, some AI chips may face the risk of idleness. Furthermore, NVIDIA CEO Jensen Huang stated last year that &#34;China will win the AI competition,&#34; citing a more favorable regulatory environment and lower energy costs.</p>
<h2>Morgan Stanley: U.S. Power Shortage by 2028</h2>
<p>If chips are viewed as the &#34;brain&#34; of AI, then power is its &#34;blood.&#34; When the &#34;blood&#34; cannot keep up, even if the &#34;brain&#34; develops advanced capabilities, AI can only stagnate. Currently, the U.S. AI industry indeed faces severe energy challenges. Morgan Stanley analyst Stephen Byrd warned last November that due to the enormous power demand from AI data centers, the U.S. could face a power shortage of up to 20% by 2028, with a potential gap of 13 to 44 gigawatts, equivalent to the annual electricity consumption of over 33 million households.</p>
<p>Samantha Dart, co-head of Goldman Sachs' global commodities research, also pointed out that the U.S.'s new power generation capacity is not keeping pace with growing demand. According to industry norms, power grids need to maintain at least a 15% reserve margin to create a safety buffer between peak load and available generation capacity, but currently, some U.S. power grids are still below that level. As data centers continue to drive electricity demand, the issue of insufficient reserves will further worsen in the future.</p>
<p>Goldman Sachs has previously indicated that U.S. data centers already consume about 6% of the country's electricity, and this proportion is expected to rise to 11% by 2030. However, the effective reserve power capacity indicator, which measures grid reliability, has dropped from 26% five years ago to 19%. Among the 13 regional power grids in the U.S., eight have reached or fallen below this critical threshold</p>
<h2>China Has Formed a Diverse Energy System</h2>
<p>In contrast to China's electricity supply situation, Liu Liehong, director of the National Bureau of Statistics of China, stated last March, &#34;In China, electricity is our advantage.&#34; Goldman Sachs also pointed out that after large-scale electricity construction since 2021, China has formed a diverse energy system in renewable energy, coal power, nuclear power, and other fields. It is expected that by 2030, China will have about 400 million kilowatts of backup power capacity, which is more than three times the expected total demand of global data centers.</p>
<p>According to domestic media, last year the national &#34;West-to-East Power Transmission&#34; scale reached 340 million kilowatts, accounting for about 23% of the country's maximum electricity load. By 2030, the capacity for optimizing the allocation of grid resources will continue to strengthen, and the scale of &#34;West-to-East Power Transmission&#34; will exceed 420 million kilowatts. It is also noted that China's power grid has become the world's largest transmission capacity, the highest operating voltage level, the largest scale of new energy grid connection, and the longest safety operation record of a super-large power grid.</p>
<h2>Inner Mongolia Becomes &#34;Prairie Cloud Valley&#34; with Apple and Alibaba Setting Up Shop</h2>
<p>China's rich and inexpensive electricity resources are becoming an important advantage for its AI industry development. According to domestic media reports, Chinese companies such as DeepSeek are relying on cost advantages to develop high-quality AI models at lower prices, effectively compensating for the domestic chip technology shortcomings.</p>
<p>Reports indicate that Ulanqab City in Inner Mongolia has become a new hub for China's AI industry, even being hailed as &#34;Prairie Cloud Valley.&#34; Thanks to the cool climate and low electricity prices, tech giants like Apple, Alibaba (9988), and Huawei have set up data centers here, and electric vehicle manufacturer XPeng (9868) is also training AI models and processing AI workloads in the area. Over the past five years, the region's GDP has grown by 50%, and the electricity consumption of data centers has increased more than sevenfold. As of June last year, the local area has attracted about $35 billion in computing power industry investment.</p>
<h2>Electricity Prices Are Half of Some Areas in the United States</h2>
<p>Since the launch of the &#34;East Data, West Computing&#34; plan in 2021, China has actively utilized the abundant electricity resources in the west to meet the data processing needs of the densely populated eastern regions. China's current installed power generation capacity reaches 3.75 terawatts, more than twice that of the United States. Data from the National Energy Administration shows that the electricity price obtained by Chinese data centers through long-term power purchase agreements is as low as 3 cents per kilowatt-hour, which is less than half of that in some areas of the United States. In addition, by 2028, China also plans to build a shared national computing pool by connecting hundreds of data centers, which can also be referred to as the &#34;National Cloud.&#34;</p>
<p>In the nuclear power sector, according to data from the China Nuclear Energy Industry Association last April, China has a total of 102 nuclear power units in operation, under construction, and approved for construction, with a total installed capacity of 113 million kilowatts, making it the world's largest for the first time. At the same time, the Yarlung Zangbo River Hydropower Station being built in Tibet is expected to generate three times the output of the Three Gorges Dam.</p>
<p>The report also cited Bernstein semiconductor analyst Lin Qingyuan, who pointed out that although China is still constrained by a shortage of advanced chips in the short term, its power advantage keeps it competitive in the AI race. &#34;The longer the AI competition lasts, the more opportunities China has to close the gap.&#34;</p>

TSLA

NVDA

<p>In the global AI competition, Musk pointed out that China is expected to narrow the gap with the United States due to its advantages in electricity production scale, predicting that by 2026, China's electricity output will reach nearly three times that of the United States. A Goldman Sachs report shows that the U.S. power grid faces capacity shortages, while China's power reserve capacity will meet three times the global data center demand. Musk believes that although the U.S. restricts China's access to advanced semiconductors, the impact will weaken over time. Morgan Stanley analysts warn that the U.S. may face a power gap of up to 20% by 2028</p>

- The article discusses the significance of power supply in the global AI competition, highlighting Elon Musk's view that China's electricity production scale could help it catch up with the U.S. in AI development.  
- A Goldman Sachs report corroborates this, noting that U.S. power grids face capacity issues, while China is projected to exceed global data center power demands by 2030, leveraging its diversified energy system.  
- Additionally, China's competitive advantage in low-cost electricity has attracted major tech companies to establish data centers, boosting its AI industry amidst a period of significant infrastructure investment.

Stheadline

Tesla

Morgan Stanley

NVIDIA

Goldman Sachs

China's victory in the AI competition relies on its power advantage. Musk estimates production will reach three times that of the United States, "the impact of chips will weaken over time."