Recently studying investment opportunities in energy stocks for 2026, I found that the market logic has completely changed.

In the past few years, energy stocks were a game of policy subsidies and capacity expansion, with electric vehicles and solar cells making money through government support and economies of scale. But now it’s completely different—the real driving force has become AI’s rigid demand for electricity. This is the core logic behind investing in energy stocks.

According to the latest forecasts, global data center electricity consumption will surge from 460 TWh in 2022 to about 1,050 TWh in 2026, with AI-related portions contributing more than half. The electricity consumption required to train a large AI model alone reaches thousands of MWh—equivalent to the annual electricity usage of tens of thousands of households. The rack power density of AI data centers has already reached 50–100+ kW, which is 5–10 times higher than that of traditional data centers.

This brings several investment directions. First is nuclear power, because solar and wind are intermittent and cannot meet the 24/7 stable power supply requirements of AI data centers. Microsoft, Amazon, and Google are all making major investments in nuclear power. Amazon even plans to deploy 12 small modular reactors. The advantages of small modular reactors are factory prefabrication, fast deployment, and high safety, making them well suited for building near data centers.

Second is grid upgrades, which may be the biggest bottleneck. Power generation is easy, but power transmission is difficult. The global power grid is severely aging, and the delivery lead times for high-voltage transformers and switchgear will still be as long as 2–3 years in 2026. The share of electricity consumption from data centers in the United States will rise from 4% in 2023 to over 8%, directly driving power companies’ revenue growth rate from 1% to 4–6%.

Which Taiwan energy stocks are worth paying attention to? Delta Electronics is a leader in power electronics, and high power density AI servers are driving a surge in orders. Huacheng Electric Machinery is a long-term partner of Taiwan Power Company (Taipower), benefiting from Taipower’s NT$564.5 billion grid upgrade plan. United Renewable and Yuan Jing are both large solar module manufacturers, benefiting in 2026 from anti-dumping tariff effects in Europe and the U.S. as well as technological upgrades. Sunwoda is a leading supplier of wind turbine blade materials, with an order backlog of over NT$10 billion.

On the U.S. market side, Constellation Energy is the largest nuclear power operator in the United States, having signed a 20-year contract to restart Three Mile Island with Microsoft. Oklo is a micro nuclear reactor pioneer, supported by OpenAI’s CEO, and is ahead in progress with NRC approval expected in 2026. Eaton is a leader in grid intelligence; the high power density of AI data centers is boosting transformer demand, and orders are surging. GE Vernova covers high-voltage transformers, HVDC transmission, and wind power equipment, benefiting from global grid upgrade investment. NextEra Energy is the largest renewable energy company in the United States, with stable dividends; under the long-term trend of net-zero transition, its growth is well defined.

The investment recommendation is to allocate 50–60% of the portfolio to AI power stocks—high growth and high volatility; allocate 30–40% to traditional energy stocks for stable defense; and keep the remaining 10% in cash or bonds as a buffer. New energy stocks are volatile, so avoid chasing highs; in a long-term upward trend, treat short-term pullbacks as an opportunity to add positions. Focus monitoring on leading indicators such as AI capital expenditure, the scale of grid investments, order backlog, and technological iteration.

2026 to 2030 will be the most valuable structural opportunity window for new energy stocks. Whether it’s AI’s urgent power demand or the global transition to net-zero emissions, they all point in the same direction.