The solar industry is booming again this year. The U.S. installed 14.5 GW of large-scale solar so far this year. Solar was 75% of grid capacity additions in the first half of this year. Residential solar is also growing 14% a year. On larger projects, solar was over half of all new generation added to the U.S. grid in 2025.
Over recent years, the cost of solar panel modules has decreased substantially due to improved photovoltaic technology and manufacturing scale. This makes them more feasible, and the trend will continue. According to nonprofit Sun Day campaign, solar has been the largest source of new generating capacity added each month for 21 consecutive months, since September 2023.
Look at this graph from the EIA, showing the tremendous lead from utility-scale solar
Solar farms are increasingly being built together with battery storage systems. Around half of all battery storage systems are a combined solar and battery system.
Looking at solar growth through the lense of data center expansion, the future looks bright. Large data centers are seeing a growing trend in 2025, where hybrid power sourcing is combining on-site generation with grid backup.
According to Deloitte and RMI, hyperscale operators have responded to four to seven year grid interconnection delays by building behind the meter natural gas turbines or solar plus battery microgrids.
The U.S. Department of Energy’s Industrial Efficiency and Decarbonization Office funded the 2024 United States Data Center Energy Usage Report by Lawrence Berkeley National Laboratory. Lawrence Berkeley Lab found that the carbon intensity of electricity used by data centers was 48% higher than the U.S. average. This is partly due to many of them being in states with dirtier energy sources, like coal.
They also showed that 4.4% of all the energy in the US now goes toward data centers. Clean energy is definitely needed in this area. The big tech data centers keep getting bigger and bigger. They all need power backup and solar will be a part of that.
Where The Growth Comes From
Data centers, by current forecasts, will grow 15-23% CAGR and by 2030, are projected to consume between 8% and 12% of total U.S. Electricity. S&P Global’s 451 Research adds that total grid power demand from data centers may triple by 2030. This is with the normalcy bias that AI Co-Pilots and Agents won't see faster adoption. I see data center power draw eventually being much higher than forecasts, especially after 2030. I'll get to that in a moment.
Data centers are just a piece of the power draw. Overall, U.S. Electric grid growth from now until 2030 is forecasted to grow at a base case of 2.5% to roughly 3% per year. The most growth is forecasted for battery storage, coming in at 116% growth through 2035. Solar is second at 64% growth.
There is a bit of a problem, because after decades of flat electricity usage, demand is surging, but the infrastructure is old. Furthermore, 67% of electric utilities’ spending in 2024 was on replacements ($63 billion), while only $32 billion was spent on new lines and substations. The need for alternatives, like solar, has never been greater.
I'm particularly bullish on the energy demand because I see a strong potential that the growth from Agentic AI and AI Co-Pilots and Bitcoin mining exceeds the estimates. AI inference is already projected to take a large share of total U.S. electricity in the near future. In a couple years, it is going to surpass AI training. The estimates going five years out vary widely, and no one really knows the exact timeline and impact, of course. The EIA and Deloitte has 2030 estimated at 945 TWH to 1065 TWH. Even if that is just the base-case we are still talking about an incredible power draw.
Gen AI is growing at over a 30% CAGR. The AI assistants and AI Co-Pilots market is currently expected to grow at a slightly higher CAGR than Gen AI at mid 30% to mid 40% through 2030 according to Stanford and Grandview. So, before we have even seen the pre-seed startups still building come yet it is already tremendous expectations. Even Gen AI isn't fully adopted yet with the consumer. Only about half of adults are using it. So, the upside to the growth the next 10-15 years is hard to fathom.
The Agents and Co-Pilots are what will truly be beneficial to businesses. This growth is going to takeoff. It's like going from a dial-up internet modems in the 90s to ethernet or DSL broadband. Nvidia CEO Jenson Huang expects every company to have an "AI Factory." AI factories industrialize the creation of intelligence by making AI generation repeatable, scalable, and measurable across sectors.
This BOFA graph below illustrates a scenario. I like it not for the exact number estimates but rather to show how the inference could start to skyrocket exponentially to a large percent of total power usage very soon.
90% of enterprises currently anticipate spending more on AI in the future. Some sources already have enterprise adoption at over 70%. My relative is building an AI Co-Pilot for auto technicians that improves their productivity and revenue. They have been using clunky software for years. So, I think normalcy bias is discounting how widespread and fast the newest software will roll-out and see speedy adoption.
Another potential big driver of energy demand is Bitcoin. Bitcoin, whether one likes it or is skeptical, has been an outperforming asset since its beginning. If the price appreciation continues its bullish trend it will draw significantly more power as time passes.
Nextracker NXT
Nextracker NXT sells solar trackers and energy yield management systems. Their business model is built around large-scale solar farms that power utilities, corporations, and regional grids. Nextracker is the industry leader, holding about 70% U.S. market share.
I looked through Nextracker's latest quarterly and historical data. I'm glad I decided to look at this one, because it is the number one holding in the solar ETF TAN.
It has rightfully owned the #1 spot, being the global market leader based on gigawatts shipped for ten straight years.
The only other U.S. based competitor I found with a big U.S. presence is Array Tech ARRY. The key difference between the two is their size. Nextracker is a $14 billion company, and ARRY is only a $1.5 billion company. Even so Array has installed an impressive 80GW worldwide.
Over the last five years, Nextracker's revenue has grown at a 22% compounded annual growth rate. This is extremely good growth for being a $13 billion company. Less than 7% of Fortune 500 companies with market caps of $10 billion or higher in 2024 to 2025 grew revenue 20% year over year.
Earnings for the full year are forecasted at $4.20 a share. With the stock at $87, that is only a 20 PE ratio. Trailing twelve month price to free cash flow is just 22 also.
This year, Nextracker bought two companies.They purchased Bentek, an industry pioneer and manufacturer of electrical infrastructure used in all types of solar power plants. Additionally, in May they acquired 100% of the interest in OnSight, a supplier of autonomous inspection robots and fire detection systems purpose-built for solar plants.
Even after these purchases, Nextracker still has $743 million in cash. They have also been creating a lot of shareholder value. Shareholder equity is ballooning. It grew from negative -$3 billion in 2023 to $1.8 billion as of the latest quarter.
Array Tech ARRY
Array Tech ARRY also benefits from the grid expansion. They make similar sun tracking systems like Nextracker.They say that three decades of field-tested design improvements have resulted in the ARRAY DuraTrack being the most durable, reliable tracking system. They serve utility-scale developers which fits perfectly for this theme.
I expect they will have solid growth going forward, however, Array had a sharp decline in revenue in 2024. Profitability has been inconsistent too. This has given them a much lower valuation vs Nextracker. Array's PE ratio, based on this years forecasted earnings of $.67 a share is just a 13 multiple.
Canadian Solar CSIQ
Canadian Solar CSIQ, in their last conference call, mentions having USA and Spain data center products being ready in the next few quarters. CSIQ's market cap is currently just under $1 billion. If they start getting hyperscaler contracts, those could really move the needle for them given the market cap is only $977 million.
They can grow sales and earnings as they showed from 2021 to 2023. The valuation has gotten historically low. It is trading so cheap now for just .16 times sales and 1.00 times enterprise value.
Final Thoughts
In the current U.S. equity market, there are a lot of industries sporting very high to outrageous valuation multiples based solely on future potential. The solar industry's prospects are more than just a promise of growth, it is a clear reality based on the trend and data. Nextracker and Array Tech will be a big part of the grid and data center expansion. Based on their valuation, they are very compelling stocks for the long-term. I like Nextracker the most because of the strong fundamentals. Array is more of a risk to reward play if they can improve the bottom line, there is a lot of promise.
