Donald Trump and his energy secretary Chris Wright insist that solar power will never be a reliable source of energy. The US president has wound down tax credits for solar power, restricted solar development on federal land and raised tariffs to discourage companies from importing solar panels from China and elsewhere in southeast Asia.
Even so, has solar power really had its time in the sun? Solar stocks dipped following Trump’s return to office this time last year and the subsequent removal of subsidies on projects not yet underway in July’s ‘One Big Beautiful Bill Act’ (OBBBA). This appears to be a short-term setback, however, as solar energy participates in the urgent ramping up of electricity capacity needed to power the AI data-centre build out.
Speed, space and subsidies
According to the International Energy Agency, solar will account for around four-fifths of the global increase in renewable power capacity over the next five years, driven by lower costs, faster permitting timeframes and easier installation – all of which favour solar over wind and nuclear power.
Furthermore, despite Trump’s dismissal of solar power, the US is unique among developed Western nations for its size and space, making many places ideal for solar projects. States such as Arizona, New Mexico and Texas all benefit from plenty of flat empty terrain, reliable sunshine and competitive financing. As an example, the $500bn (£380bn) Stargate data-centre project in Abilene Texas, which will require enough power for a whole city, is ideal for installing largescale solar fields to help with power generation.
“While the cost to build energy sources has traditionally been measured in megawatts, AI’s hunger for power is now being measured in gigawatts
For the US solar industry – and the trend of re-shoring more broadly – there is still a government tailwind.”
Also, the OBBBA retained favourable tax treatment for domestic manufacturers of solar modules, panels and trackers. The 45X production tax credit, introduced under Biden’s Inflation Reduction Act, remains in place, benefitting US-made components – albeit under a slightly accelerated phase-out schedule. For the US solar industry – and the trend of re-shoring more broadly – there is still a government tailwind.
We can go further still by quantifying the economics of solar generation using the ‘levelised cost of energy’ (LCOE), which measures the average cost per megawatt hour to build and run a new energy generation plant. Lazard calculates this figure each year across energy sources and, in June 2025, concluded that – without subsidies – the LCOE ranges of solar ($38-78/MWh) and wind ($29-64/MWh) remain significantly below traditional gas ($48-107/MWh) and coal ($74-143/MWh).
Solar can now prosper without subsidies due to technological innovations over the last decade, lower running costs and little exposure to commodity prices. That said, solar is by no means a stand-alone solution as it does not solve the need for reliable baseload power. The grid requires a level of ‘always available’ energy for many industrial applications and, more than cost, this intermittency is the main pushback on renewables.
This drawback does not limit the use of solar, though – it just changes the ways it is used in the energy ecosystem. While solar has a future as part of the wider generation mix, for the next decade or so, its role will be to fill the gap between five-year and 15-year build-outs required for gas and nuclear power stations respectively. The real ace up solar’s sleeve comes later as the problem of affordable large-battery storage is solved. This will allow solar’s traditionally intermittent generation to become functionally equivalent to baseload.
The resolution of political and policy uncertainty – even one resulting in a worse absolute outcome – can act as a powerful catalyst in the stockmarket.”
While the cost to build energy sources has traditionally been measured in megawatts, AI’s hunger for power is now being measured in gigawatts – for example, OpenAI CEO Sam Altman says he wants to build data-centres that can produce a gigawatt’s worth of AI training every week.
One gigawatt of energy capacity can power around 800,000 US homes for one hour and is equivalent to the power generation of 1.9 million solar panels. Reaching OpenAI’s implied goal of 250 gigawatts worth of AI training by 2033 will require a huge amount of energy capacity and likely include vast numbers of solar panels.
If Altman is serious, then the US will need to adopt an ‘all of the above’ approach to energy generation – especially as electrification of the grid ramps up to help reach Net Zero in other areas, from cars and buses to chemical refineries and steel furnaces.
Opportunities for investing
Since the OBBBA passed on 4 July, the Invesco Solar ETF (TAN) has counterintuitively risen more than 26%, outpacing the S&P 500’s 9.37% return (to 31 October). The resolution of political and policy uncertainty – even one resulting in a worse absolute outcome – can act as a powerful catalyst in the stockmarket.
One standout performer is Nextracker, which makes specialised equipment that enables solar power plants to track the sun to optimise performance, adding value to solar projects for utility companies, which form over 95% of Nextracker’s business.
Generic tracking equipment increases solar energy yields by 20% to 30%, but Nextracker’s technology can add around a 5% improvement on top of that, making it a go-to choice in the industry. The stock is up 169% since the beginning of the year – and up 48% since the OBBBA was signed into law.
Breaking down the economics of solar in the wake of its subsidy removals shows it pays to be active in both industry and stock selection. The sun has not set on the solar industry.
Dan Scott Lintott is an investment analyst at De Lisle Partners
