AI is reshaping nearly every industry, and nowhere is the opportunity more transformative than in energy. Fast-tracking FOAK (First-of-a-Kind) manufacturing, particularly in the solar industry, could redefine how we scale climate solutions globally.
From my perspective, the rise of AI has two critical impacts on the energy industry: the surge in energy demand; and the acceleration of technology development, specifically in deep-tech and perovskite solar technology.
Let's look closer at energy demand. After two decades of flat US electricity consumption, the curve has broken upward. And while forecasts vary, one thing is clear: AI is generating unprecedented new demand.
This surge is generating lots of interest in natural gas and even coal. But there are significant capacity constraints to natural gas growth, which have already roughly tripled the costs of new gas plants in the last three years. While demand may slow the closure of coal plants, the economics of coal makes new builds very unlikely.
Here's where the economics and timelines become compelling: it takes roughly two years to deploy utility-scale solar, wind, storage—or microgrids. Natural gas needs three years, at a higher price. Nuclear? A lot of companies are bullish on nuclear. Yet in the last 16 years, the US built one nuclear power plant. Construction took half a decade longer than planned and cost billions of dollars over budget.
And I know what you’re going to ask next: what about Small Modular Reactors (SMRs)? Designed to be smaller (50-300 MW vs 1000+ MW for traditional plants), SMRs are theoretically better suited for dedicated data center power, with reduced construction time and costs. Yet the reality isn’t as bright near-term—even optimistic projections put the first commercial SMR deployments at the early 2030s.
So if the goal is more cheap, fast, and reliable power, solar is a winner! And high-efficiency US-made perovskite tandem solar delivers even more: higher performance, smaller footprint, faster time to power, and lower cost of energy. These characteristics are extremely appealing to energy-hungry AI companies.
There's also tremendous potential to re-power current solar plants with higher efficiency panels, which could scale power generation much faster as there’s no need for new grid connections or lengthy approval processes.
Unlocking the Advantages: AI Accelerating Deep Tech Energy Innovation
The Cambrian explosion of new AI applications triggered by ChatGPT’s launch at the end of 2022 has opened the door to transformational improvements for many industries, from software development to medicine to law.
As AI continues to improve, many foresee energy technology as one of its next big frontiers. At Swift, we can see several clear categories emerging.
First, there’s the broad impact of automating and accelerating general operational processes that are still heavily reliant on human effort. This is already happening through the application of generative AI systems in digital spaces. Perhaps soon, this could advance to humanoid robots in physical ones.
Beyond that, we can break things down into a few key areas:
1. Optimizing Energy Generation, Distribution, and Consumption
AI has the potential to analyze massive amounts of data in real time, helping us make smarter decisions about when and how energy is produced, distributed, and used. There are multiple success stories: Nest optimizes end-user consumption. Google DeepMind and newcomer Emerald AI are using machine-learning systems to slash data-center cooling loads and turn AI compute into flexible grid assets. Duke Energy’s AI-driven self-healing grid tech reroutes power and trims outage time.
2. Geographical Exploration and Site Selection
Companies like Fervo Energy, Terabase Energy, and KoBold Metals are already using AI to find new natural resources or identify the best sites for all kinds of power plants, speeding up what has traditionally been a slow and manual process.
3. Reducing Failures Through Predictive Maintenance
With advanced analytics, AI can spot patterns that indicate equipment is likely to fail—whether that’s on the grid or in manufacturing processes—allowing us to prevent costly outages and downtime before they happen. Tractian or MaintainX are great platforms for this, which we currently use at Swift Solar.
4. Accelerating Technology Development
Perhaps most exciting is the role AI can play in research and development. From simulating new materials to optimizing designs for solar cells and batteries, AI could dramatically speed up the pace of innovation in the energy sector.
As a deep tech company, we’re deeply interested in R&D. This is where AI could make a truly outsized impact—helping us accelerate our learning rate, streamline experimentation, and push the boundaries of what’s possible in solar technology. In fact, we’re already using AI tools today that demonstrate the great potential on the horizon.
The prerequisite to an AI-first approach in technology development is the data, hardware, and operational infrastructure. We made the bet early on and built this foundation with a talented engineering and R&D operations team. With a slogan of “collect as much data as possible,” we have intelligently ingested data from every possible source in our operations for the past 5+ years.
As a result, we have a huge, well-structured database that jumpstarted usage of AI models without the need for much data cleaning—a task that can take a lot of effort and time for machine learning engineers.
Furthermore, we developed a tailored experiment-planning-and-execution platform, which standardizes our experimentation workflow with sample tracking. The planned experiments feed into a ticketing system, which guides our technicians through execution with our process and characterization equipment. This has introduced a level of “automation” to our lab work, without having to build a full robotic lab with little flexibility or wait for the arrival of general-purpose humanoid robots.
These infrastructure components, in combination with the massively enhanced AI developments of recent years, have enabled a very powerful system for accelerated R&D at Swift Solar. We have implemented an R&D Copilot that automates even more of our experiment-planning-and-execution platform, and we’re building toward a full self-driving fab.
We’re also developing AI to predict experiment outcomes early on in the process flow. Imagine a solar cell stability test that could take months before conclusions can be drawn. We are leveraging our huge dataset—among the biggest in the perovskite industry—to train prediction models. The latest developments of multimodal models, often used in large language models (LLMs), allows us to fuse data from images, spectra, time-series and scalar parameters to make even more precise predictions.
Finally, LLM-enabled coding agents have made our lean software development team more productive, and made it possible for the team to build internal agentic systems that help with R&D knowledge management, either from our database or from language-based sources (think internal research reports or specific literature).
In the longer-term, we could see AI enable breakthroughs in material discovery—identifying new compounds, substances, and formulations that solve challenges in developing technology or unlock improvements in existing systems. AlphaFold, for example, has put pharmaceutical development on the verge of a revolution. While energy has not yet had its AlphaFold moment, there are a few players that are pushing the boundaries, including Orbital Materials with their Orb system, Microsoft with MatterGen, and DeepMind's GNoME—and we are watching their work with great interest.
The Long-Term Vision: Building Tomorrow's Energy Infrastructure
We're still some time away from realizing all of this potential, but a virtuous cycle is already beginning: AI creates unprecedented demand for clean energy, which drives market opportunities for companies commercializing breakthrough technologies like perovskites, which in turn use AI to accelerate their own development and manufacturing capabilities. Swift Solar isn't just a player in this cycle—with our data advantage, technical expertise, and concrete AI implementation roadmap, we're positioned to lead it.
