Recently, I noticed a very interesting phenomenon: the logic behind new energy investments in 2026 has completely changed. In the past few years, everyone was focused on subsidies for electric vehicles and solar capacity, but now the focus has shifted entirely to the essential demand for AI-powered electricity and grid upgrades. This actually reflects a hardcore need — the electricity consumption of data centers is exploding exponentially.

According to the latest forecasts, global data center electricity use will surge from 460 TWh in 2022 to 1,050 TWh this year, with AI-related parts contributing over half. The power consumption of training a large AI model can reach several thousand MWh, equivalent to the annual electricity usage of tens of thousands of households. This means that intermittent power generation from traditional solar and wind energy simply cannot meet the demand; 24/7 stable power supply is the hard requirement.

Why are tech giants like Microsoft, Amazon, and Google all rushing to invest in nuclear energy? Because they understand this point deeply. Amazon plans to deploy 12 small modular nuclear reactors directly, and Google has committed to tripling its nuclear capacity by 2030. Power companies with nuclear and natural gas assets are receiving the highest premiums — a long-term structural opportunity.

But I think many people overlook another bottleneck — “easy to generate power, hard to transmit.” The global power grid is severely aging, with delivery times for high-voltage transformers and switchgear still as long as 2-3 years. The supply-demand gap is expected to persist until at least 2027. Data centers will account for over 8% of total electricity consumption in the US, up from 4% in 2023, directly boosting the revenue growth rate of power companies from 1% to 4-6%.

Of course, not all new energy investments need to follow AI-powered electricity. The long-term goal of global net-zero emissions remains unchanged; renewable energy will account for nearly 50% of global electricity by 2030. Traditional solar and wind energy have experienced overcapacity, and now they are entering a stage of supply stabilization, cost reduction, and demand recovery. Geothermal power concept stocks, as a baseload energy source, are also worth attention as a defensive foundation in the new energy portfolio.

Here are some opportunities I’ve summarized in Taiwan. Delta Electronics is a leader in power electronics, with high-power-density AI servers driving a surge in orders that will continue to grow through 2026, and automotive electronics are also expanding. Walsin Lihwa is a long-term partner of Taiwan Power Company, benefiting from the NT$564.5 billion “Strengthening Power Grid Resilience Construction Plan,” and is also a leader in the electric vehicle charging station market across Taiwan. United Renewable Energy has completed capacity optimization, with gross profit margins rebounding, benefiting from anti-dumping tariffs and technological upgrades in Europe and the US, with overseas module shipments expected to grow over 15%. Sunwoda, as a leading manufacturer of wind turbine blade materials, has a backlog of over NT$10 billion, with revenue growth projected at 18%. Yuanjing focuses on high-efficiency heterojunction and TOPCon products, with high overseas order visibility, expecting annual revenue growth of 12-15%.

On the US stock side, I see several particularly worth allocating targets. Constellation Energy is the largest nuclear power operator in the US, holding about 20% of the country’s nuclear capacity, and recently signed a 20-year contract with Microsoft to restart Three Mile Island, with a major expansion of data center projects expected in 2026. Oklo is a pioneer in micro nuclear reactors, supported by Sam Altman, with fast fission technology that is low-cost and quick to deploy, showing explosive potential amid AI power shortages. Eaton is a leader in smart grid technology, with transformer demand surging and delivery times extending to 24 months; the grid business is expected to grow over 25% in 2026. GE Vernova, spun off from GE’s power and renewable energy division, benefits from global grid upgrade investments, with record-high order backlogs and projected revenue growth of 15-18% in 2026. NextEra Energy is the largest renewable energy company in the US, with stable dividends and annual payout growth of over 10%, balancing the volatility of AI power stocks.

For practical investment advice, the risks in the new energy sector are indeed high, but the long-term return potential is also considerable. In portfolio allocation, I suggest 50-60% in AI power stocks aiming for high growth, 30-40% in traditional energy stocks as a hedge, and the remaining 10% in cash or bonds as a buffer. Most importantly, avoid chasing highs; look for short-term dips within the long-term upward trend as opportunities to add. Key indicators to monitor include AI capital expenditure, grid investment scale, and order backlogs — these are leading indicators. New energy stocks are not about hype but about order certainty and rigid demand.

The new energy cycle is long, and bear markets are often accompanied by policy cold spells, but every downturn is a starting point for a long-term bull run. Against the backdrop of the AI era and global net-zero transition, 2026 to 2030 is truly a structural opportunity window for deepening in new energy stocks.