① US hedge funds maintained a stable exposure to AI-related investment topics in Q3, but reduced their investment in the semiconductor industry for the first time since Q2 2022. ② OpenAI's newly revealed “US AI Infrastructure Blueprint” conceals its grand vision of nuclear power; Hwang In-hoon also expressed optimism about nuclear energy. ③ Judging from the trend of tech giants, they mostly chose small modular reactors (SMR).
“Science and Technology Innovation Board Daily”, November 21 — US hedge funds are pouring into the nuclear power sector.
The AI industry has brought explosive growth in electricity demand, and nuclear power is seen as one of the key directions. According to Goldman Sachs's latest analysis of 697 funds with total asset management of more than 3 trillion US dollars, US fund managers bought shares of nuclear power companies in a big way in the third quarter of this year in order to seize the tide.
Goldman Sachs's “Hedge Fund VIP List” lists the stocks most favored by fund managers. Under “nuclear power heating,” several nuclear power concept stocks have already joined this list.
For example, the power company Vistra Corp (VST), which currently has an installed capacity of about 6.4 GW of nuclear power in the US, owns the Comanche Peak nuclear power plant in Texas, the Beaver Valley nuclear power plant in Pennsylvania, and the Perry and Davis-Bese nuclear power plants in Ohio;
Another independent power company, Talen Energy Corp (TLN), also entered the list in the third quarter. Talen Energy's Susquehanna steam power plant in Pennsylvania has a power generation capacity of 2.7 GW and is one of its core assets.
Goldman Sachs pointed out that overall, hedge funds maintained a stable exposure to AI-related investment themes in the third quarter, but reduced their investment in the semiconductor industry for the first time since the second quarter of 2022.
▌AI “energy saver”? Tech giants pile up their layout
AI energy consumption has always been a high-profile topic, and more and more large technology companies have turned to nuclear power to meet the growing power needs of data centers.
OpenAI's recently revealed “US AI Infrastructure Blueprint” conceals its grand vision of nuclear power. OpenAI proposed that US investment in AI will bring tens of thousands of jobs, GDP growth, modern power grids including nuclear power, a number of new chip manufacturing facilities, and billions of dollars of investment from global funds. Furthermore, the US Navy operates about 100 small modular reactors (SMR), and the Navy's expertise can be used to build more civilian SMRs.
In an earlier interview, Hwang In-hoon also said that more and more data centers require renewable energy, and nuclear energy is a good choice. “Nuclear energy is fantastic as an energy source, a sustainable energy source. But it won't be the only energy source. We will need energy from all sources and balance energy availability, cost, and long-term sustainability.”
Meta and Google parent company Alphabet both mentioned that building new nuclear power facilities may help meet future surges in electricity demand.
It's worth noting that judging from the trends of tech giants, they mostly chose small modular reactors (SMRs):
Oklo, the chairman of OpenAI Altman, is working hard to launch the company's first SMR in 2027;
Google has signed an agreement with nuclear startup Kairos Power, which will build seven SMRs to power Google's data centers;
Amazon Cloud AWS announced that it has signed cooperation agreements with three companies to initially invest hundreds of millions of dollars to develop SMR;
Oracle has obtained permission to build three nuclear reactors, all SMRs to power AI data centers...
Compared with traditional large-scale nuclear power plants, SMR has advantages such as small footprint and short construction cycle, and can be built closer to the power grid. Generally speaking, the power output capacity of an SMR is 300 megawatts, which is about one-third of a traditional large nuclear reactor.
Haitong International Securities pointed out that in the medium to long term, SMR may become the highest priority energy investment direction for large US technology companies.
According to the International Atomic Energy Agency, about a quarter of the new nuclear capacity is expected to come from SMR by 2050. Thanks to high interest and investment in nuclear power technology (especially SMR technology) from end users, especially large technology companies, it is expected that global nuclear power will develop optimistically. Based on the development of SMR technology and the active promotion of large technology companies, analysts expect the first batch of commercial SMRs to be put into use around 2030. SMR's commercial installed capacity will grow from a 1 GWe explosion to 122.25 Gwe in 2030-2050, with an average additional installed capacity of 6.11 GWe added in 2030-2050, with a compound growth rate of 27%.
Implemented to specific links in the industrial chain, the CITIC Securities research report on November 20 proposed investments in terms of nuclear fuel (uranium), nuclear power equipment, spent fuel treatment, and nuclear power operators.
Overseas analysts recommend focusing on: 1) overseas nuclear fuel suppliers; 2) large equipment manufacturers; 3) small modular nuclear reactor construction companies; 4) core equipment and automation; 5) nuclear power operation.
Domestic suggestions to focus on: 1) Benefiting from the approval, restart and accelerated construction of global nuclear power projects, natural uranium demand is expected to exceed expectations, and is optimistic that uranium prices will rise in the medium to long term; 2) pressure vessels, steam generators, nuclear main pumps, and nuclear pump bearings in the nuclear island equipment chain with a stable competitive pattern and high added value; 3) the nuclear power valve process is expected to increase further in unit value; 4) technology is expected to achieve breakthroughs, and domestic production prospects will replace broad spent fuel storage and transportation equipment; 5) nuclear power operation.