Venture capitalists and startup founders are fiercely debating where we should focus our climate tech investments over the next five to 10 years.

Specifically, my fellow investors and I debate whether we should invest only in hardware energy and climate solutions—physical products like Long-Duration Energy Storage (LDES) and more efficient solar panels—or if we should also invest in software and artificial intelligence solutions, like carbon intelligence, data and AI platforms that generate insights about the energy grid and climate hazards, especially with the rise of generative AI.

One school of climate investors views climate software as a waste of time and investment money especially because it is so difficult to measure the tangible impacts of software solutions for reducing greenhouse gas emissions.

It’s true, software’s impact can be hard to measure. Even if a company successfully reduces its carbon footprint, it is near impossible to tease out the percentage of reduced emissions attributable to the software versus the actions taken and technologies deployed to remove or mitigate emissions.

What’s more, be it an emissions tracker or a machine learning program or a social media platform, software requires energy to run. Data centers (the collections of physical servers that make our virtual world possible) are responsible for about 1% of the world’s electricity use, according to the International Energy Agency.

But another group of climate investors, including myself, thinks software will play a critical role in decarbonizing industry as well as supporting renewable energy and the new power grid.

Before moving into climate and energy, I worked on innovative medical devices. The first startup I worked for built insulin pumps for people with diabetes. We developed a software solution that could monitor glucose blood levels and the amount of insulin injected, allowing us to then iterate the physical pump to be more accurate. The stakes were high—injecting the right amount of insulin kept patients healthy and alive—and I saw how software and hardware needed to work together to solve the problem.

I believe software solutions are similarly critical to the physical energy transition even though the scale is much larger in this space. As our energy systems become more decentralized and markets become more uncertain, we must build a new, resilient energy model that can adapt to real-time constraints, like variable supplies of solar and wind power, as well as economic and supply chain fluctuations.

Hardware solutions like LDES, hydrogen systems, carbon removal technology and more all require orchestration, management and optimization cues from software.

Software could even accelerate the transition itself.

Advances in software solutions can drive the development and deployment of hardware technology solutions. During the computing revolution in the early 2000s, for example, the development of new software pushed hardware development forward.

Today, the cycle continues with advances in AI and machine learning set to accelerate the chip and computing industries at an even higher pace. In the energy space, new lines of code are much faster and easier to deploy than scaling new renewable energy infrastructure.

Software can also speed up scientific breakthroughs.

Climate tech includes areas that are evolving from scientific breakthroughs in the lab, technologies like new solvents for direct carbon capture applications and nuclear fusion. Those technologies in some cases are still in research and require great time and capital to mature.

When we talk about accelerating the energy transition, we actually talk about how to accelerate some of those research areas to help commercialize them. Investing in artificial intelligence could allow us to learn faster and make strategic decisions more quickly. Machine learning models could accelerate the research phase and advance new solutions to the commercial stage more quickly. We might end up with 20-year academic research journeys trimmed down to just four years.

In these uncertain times, we need to think creatively about building more resilient and smarter energy systems and accelerating scientific breakthroughs. If we invest only in hardware, we will lose our ability to accelerate innovation and to build flexible systems that can adapt to change and provide resilient energy to the planet.