More and more people have been discussing new energy stocks lately, but it seems many still haven’t figured out what, exactly, has changed in the investment logic for 2026.

To be frank, for the past few years, investing in new energy largely relied on policy subsidies and scale-driven competition—but it is completely different now. The real demand engine has shifted to AI.

Think about it: training a large AI model consumes thousands of MWh of electricity, equivalent to the annual electricity use of tens of thousands of households. Data centers’ electricity consumption jumped directly from 460 TWh in 2022 to an estimated nearly 1,050 TWh this year, with AI accounting for more than 70% of the incremental demand. This is not policy-driven; it is hard, rigid demand.

So why are Microsoft, Amazon, and Google all fighting over nuclear power assets? Because solar and wind power are intermittent and cannot supply electricity to AI clusters 24/7. Microsoft signed a nuclear fusion agreement with Helion, Amazon has deployed 12 small modular nuclear reactors, and Google has committed to tripling its nuclear power capacity by 2030—these are not gimmicks; they are real capital allocations.

But there’s a point that most people overlook here: electricity generation is relatively easy—transmission is the bottleneck. The delivery lead times for high-voltage transformers and switchgear are still 2-3 years, and the supply-demand shortage is expected to continue at least until 2027. That’s the real investment opportunity.

In Taiwan, Delta Electronics, a leader in power electronics, saw orders surge in 2025, with high-power-density AI servers directly boosting demand for their UPS systems and inverters. Hwa Sheng Electric is Taipower’s long-term grid upgrade partner and benefits from the NT$564.5 billion power grid upgrade plan, while also leading the charging-station market. These two names are both beneficiaries of the upsides from power-infrastructure upgrades.

In traditional green energy, United Renewable Energy and Yuan Jing have increased market share in Europe and the U.S. under anti-dumping tariff measures. In addition, WEG’s (Sunway) wind turbine blade materials have an order backlog exceeding NT$10 billion. These are not high-growth stocks, but they are stable—making them suitable as defensive building blocks in a portfolio.

In the U.S. stock market, Constellation Energy is the largest nuclear power operator in the United States. After Microsoft signed the Three Mile Island contract, data center projects are expected to expand significantly in 2026. Oklo is a pioneer in micro nuclear reactors, supported by Sam Altman, and both Amazon and Equinix are in discussions. Eaton and GE Vernova are core beneficiaries of grid equipment; they have high gross margins and long order visibility.

Interestingly, geothermal power concept stocks have also started to receive more attention recently. As a stable baseload energy source unaffected by weather, geothermal power’s long-term potential is being reassessed given the demand from AI data centers for reliable 24/7 power supply. Some geothermal concept stocks are also making rapid technological progress, especially with applications of micro geothermal systems.

As for allocation advice, I would put 50-60% of the portfolio into AI power stocks (nuclear power and grid equipment)—high growth but also high volatility. Allocate 30-40% to traditional green energy stocks as a defense. The remaining 10% would be held as a cash buffer. Geothermal power concept stocks can be considered within the traditional green energy allocation, because their stability makes them well-suited for long-term holding.

The key is not to chase price highs. In a long-term upward trend, look for short-term pullbacks as opportunities to add. Focus on leading indicators such as AI capital expenditure, the scale of grid investment, and order backlog. New energy is not about trading a theme—it’s about order certainty and rigid demand.

In the 2026 to 2030 window, as AI-driven power demand and the net-zero emissions transition advance in parallel, the structural opportunities in new energy stocks are indeed worth deep focus.