share_log

AI“用电荒”下,谁在加码核电?比尔·盖茨准备再豪掷数十亿美元

Who is investing more in nuclear power amid the AI "electricity shortage"? Bill Gates is ready to invest billions of dollars again.

Gelonghui Finance ·  Jun 17 16:40

The end goal of AI is electrical utilities.

The phrase "The ultimate goal of AI is power" has been repeatedly mentioned by tech giants including Huang Renxun, Musk, and Ultraman.

Now, Microsoft is accelerating its layout.

Recently, Microsoft founder Bill Gates said that he will continue to invest billions of dollars in new-generation nuclear power to meet the increasing demand for electricity in the United States.


Bill Gates Talks Nuclear Energy

Last weekend, Bill Gates revealed that his personally founded start-up company, TerraPower LLC, broke ground in Wyoming to build its first commercial reactor.

He said that this year, Tesla's capital expenditures related to AI will reach $10 billion, about half of which are internal expenses, mainly Tesla-designed AI inference computers, sensors in cars, and Dojo.Over 1 billion dollars have been invested and billions more will be invested in the future.

Since 2008, TerraPower has been exploring simpler and cheaper reactors and is expected to build a new reactor by 2030.

Gates said that the construction of the nuclear power plants was originally planned to be completed in 2028, but because it relied on fuel manufacturing in Russia, the original plan had to be changed and all the fuel for these nuclear power plants will now be purchased from the United States or allied countries.

TerraPower's factory has received support from the US Department of Energy.Previously, the United States has been actively exploring solutions for nuclear power plants to alleviate the electricity demand of AI data centers.

It is reported that the reactor design uses liquid sodium instead of water as a coolant and adds molten salt for heat storage to increase power generation. Gates said that this would be the safest nuclear power plant.

heHe hopes to find an energy source that is emission-free and cheaper than coal.

Regarding the original intention of TerraPower's establishment, Gates said that he not only considered the health costs of coal emissions but also that it is being overtaken by natural gas. Therefore, the goal is to compete effectively with natural gas.

"Once we produce the tenth unit and the cost of spare parts drops, we should not expect people to buy electricity at a higher price. The risk should be borne by these private investors and the US government. We believe that this very aggressive cost goal can be achieved. This is the reason why we founded this company and the reason why we believe it can contribute not only in the United States but also globally to address climate challenges."

Regarding nuclear energy investment, Gates said that innovation can improve profitability.If these climate investments really succeed, he will donate the money to the Gates Foundation to do more in areas such as malaria and tuberculosis.

From the perspective of the US economy, Gates said that building and operating these factories in the next five years will bring a large amount of employment. He hopes to build dozens of new nuclear power plants in the United States and strive to build more than 100.

From the perspective of the economy in the USA, Gates said that building and running these factories in the next five years will be important.This will bring a large amount of employment to the US economy.He hopes to build dozens of new nuclear power plants in the United States and strive to build more than 100.

He said that no matter who wins the election this fall, he is "very confident" in continuing to expand the United States' nuclear power plant inventory.


The 'power shortage' facing AI.

In recent years, carbon-free nuclear energy has increasingly been seen as a critical field for addressing climate change, and more and more companies are promoting small reactors.

Data shows that as many as 25 countries announced at last year's COP28 climate change conference in Dubai that they will increase their nuclear power capacity threefold.

Currently, with the frenzied race of AI computing power, the demand for electricity has spread to renewable energy such as nuclear and photovoltaic.

According to Goldman Sachs' forecast,From 2023 to 2030, data center power demand will grow at a compound annual rate of 15%.This growth trajectory is expected to increase the proportion of data centers in total U.S. electricity demand to 8% by 2030, up from about 3% today.

Moreover, this trend is expected todrive about $50 billion in U.S. electricity investment by 2030.

By 2030, an additional 47 GW of generating capacity will be needed to meet the growing demand for U.S. data center power, which is expected to be met by about 60% natural gas and 40% renewable energy.

Morgan Stanley's earlier data showed that from 2024 to 2027, power demand would increase by 318 terawatts, almost equivalent to the needs of a densely populated country, which could be met by building 39 large nuclear reactors.

Earlier this year, the AI gurus at the forefront raised the alarm of aAI"power shortage".

Tesla founderMusksaid that by 2025, there won't be enough power to run the chips that AI needs. ChatGPT consumes 500,000 kWh per day, which is more than 17,000 times the average U.S. household electricity consumption.

Now ChatGPT is responding to 200 million requests per day, which is causing a lot of energy consumption, and various requests will continue to increase, and energy consumption will continue to increase, so he predicts that next year's energy consumption may not be sustainable.

Nvidia founderJensen Huang.Similar warnings have also been issued.

He pointed out that the ultimate goal of AI is photovoltaics and energy storage. We cannot just focus on computing power, but need to consider energy consumption more comprehensively. If we only consider computers, we need to burn the energy of 14 Earths.

Founder of OpenAIUltramanHe also has proposed such a view. He believes that the development of future AI technology will depend heavily on energy, especially the progress of photovoltaic and energy storage technologies. The future of AI technology depends on energy, and we need more photovoltaic and energy storage.

The translation is provided by third-party software.


The above content is for informational or educational purposes only and does not constitute any investment advice related to Futu. Although we strive to ensure the truthfulness, accuracy, and originality of all such content, we cannot guarantee it.
    Write a comment