The era of "self-powered" is coming? Trump's new policies may ignite a super cycle in electric power equipment.

President Donald Trump said that under the "rate protection pledge" recently negotiated by the U.S. government, American tech giants building large-scale artificial intelligence data centers would have to bear their own power needs and related costs. Trump said in his State of the Union address to Congress that tech companies would be required to build dedicated power supply systems for their expanding AI infrastructure, rather than drawing extra power from local grids and significantly increasing the load. The consumer power rate protection pledge put forward by Trump essentially rewrites the "cost attribution" of the increased load of AI data centers into the policy framework. Large tech companies will either have to build their own on-site/behind-the-meter generation power sources, or they should not shift the pressure of additional power consumption onto local grids and residential electricity rates. On the engineering level, this will shift the power supply format of data centers from "waiting for grid expansion" to "project-provided power systems": more common combinations will be on-site gas turbine/generator units plus grid backup, or a mixed microgrid of "renewable + energy storage + adjustable load + small amount of stable power." The reason is very practical - training/inference-type areas are often far greater than 1GW continuous load, requiring high reliability (N-1/N-2) power supply, network connection timing, and power quality demands. The expansion, approval, network connection queue, and equipment delivery of the U.S. grid are all forming hard constraints. This policy direction of "rate protection + requiring tech giants to self-generate power" will objectively turn AI data centers from major power consumers into "power infrastructure investors," shifting the demand from just "grid access capacity" to a complete set of power equipment CAPEX including self-generated power supply, grid system, and on-site distribution system. This also means that the AI "electricity-eating monster" will bring an unprecedented "super bullish" trend to the electricity stock market. Trump demands that tech giants "voluntarily, self-build, and self-use" "Tonight, I am pleased to announce that I have negotiated a new rate protection pledge. Do you know what that is? We are telling all tech companies that they have an obligation to meet their own power needs." Trump said in his speech. "Our power grid infrastructure is severely aging. It can never withstand extreme figures like the massive scale of power needed. So I am telling them, they can build their own power plants. They will generate their own power and use it separately. This will ensure they have the ability to get enough power while also lowering your electricity costs," he said. He did not name which tech or electricity infrastructure companies are involved, nor did he specify how the plan will be implemented or executed. Reports suggest that the White House is expected to invite relevant companies in early March to officially implement this major initiative. Last month, tech giant Microsoft Corporation announced a plan to ensure that its AI data centers under construction do not raise consumer electricity prices, promising to minimize water usage and potentially exceed the amount of water it uses. This was while the Trump administration was considering potential actions to address the soaring costs of electricity. Wedbush, a Wall Street financial giant, said that with federal, state, and local governments accelerating their reviews of large data center power construction to address major concerns related to the construction of large-scale AI data centers, it is expected that other large tech organizations will follow suit soon. Earlier in January, the Trump administration announced a plan requiring large tech companies to foot the bill for the construction of new large power projects within the regional area managed by PJM Interconnection. PJM Interconnection is the largest grid operator in the United States, serving 67 million customers across 13 states and the District of Columbia. PJM also announced its plan, calling on the expanding or new data centers in the United States to voluntarily bring their own large-scale new power generation capacity, or they may face significant reductions in their power supply during peak demand periods. AI leads to electricity! The "super bullish" trend of electricity stocks is striking The global AI data center construction process led by Alphabet Inc. Class C, Microsoft Corporation, and Facebook's parent company Meta is in full swing, and this process is increasingly highlighting the importance of power resource supply. This is why the investment theme of "the end of AI is electricity" is becoming increasingly hot. What's even more significant is that if the "self-generation of power" path is eventually institutionalized in the United States and other regions, a significant portion of AI capital expenditure will undoubtedly be shifted to power equipment and network technology. The essence of the global AI competition is the competition of AI computing infrastructure, and the core foundation driving the AI computing cluster is a stable and massive power supply system. It is because of this that the power demand of AI data centers is skyrocketing at an unprecedented speed, with AI transforming into a "power-eating beast." Goldman Sachs Group, Inc. has significantly raised its global data center electricity demand forecast for 2030 relative to 2023, with the focus of this increase being on the United States: about 60% of the new power demand comes from the U.S., and data center capacity forecasts have been significantly raised. The reason why the global capital market has recently pushed the power equipment and network chain to become the "new main line" is also clear: the AI arms race has shifted demand from GPUs/servers to power generation equipment (combustion engines), transformers/switchgear, transmission and distribution expansion, and grid engineering and scheduling software. For example, Siemens Energy clearly benefits from the demand for AI-driven combustion engines and grid equipment, with performance and stock price performance directly linked to the "AI data center construction boom." In Europe, the largest grid operator E.ON focuses its additional investment on "preparing for the large-scale expansion of European AI data centers," and the European stock utility sector has become one of the beneficiaries of AI-related trends; in the United States, utilities like AEP and Exelon have raised their capital spending plans due to the demand from data centers. From the breakdown of the power system engineering perspective, "self-generation" does not mean "off-grid operation", and it is more common behind-the-meter power sources (large gas turbine/generator units, renewable energy + energy storage, or even nuclear power PPA) + redundant connection with utility, so the demand will simultaneously boost three chains: (1) generation side: combustion engines/power generator units, grid connection and control; (2) transmission and distribution side: substation, switchyard, GIS/circuit breakers, relay protection and SCADA; (3) data center campus side: medium/low voltage switchgear, UPS, bus ducts, distribution management, energy efficiency, and microgrid control. Benefitting from these three data center power chains, Schneider Electric has recently secured large orders/agreements related to U.S. data centers, while Siemens raised its profit guidance in the latest financial report, explicitly mentioning the sharp growth in business related to North American AI-driven data center infrastructure demands in the short term. Leaders in electricity like GE Vernova in the U.S. have the resilience to thrive with the cycle: they cover both grid infrastructure (Grid Solutions: switchgear, transformers, etc.) and power generation (Power: combustion engines, etc.), and their strong revenue outlook is closely related to the strong demand for data center power needs.