From Google to Enabling the Future Grid - with Astrid Atkinson, CEO of Camus Energy
On tech development lessons from Google, moving from Big Tech exec to climate founder, and building infrastructure for distributed energy,
Welcome to Down to Zero where we (Florian & Shaneez) speak with climate tech leaders about how to build a successful climate-tech company. Follow along to learn their playbooks and how you can apply them to further your career in climate-tech.
Thank you also to Aktsa Efendy for his help on this piece!
The rise of distributed energy resources is making it much harder for utilities to manage the grid and get energy from where it's produced to where it's needed. Camus Energy is tackling this challenge through their grid orchestration platform that provides real-time situational awareness, device control, and market integration.
We sat down with Astrid Atkinson, co-founder and CEO of Camus Energy and previously a 15-year infrastructure veteran at Google where she led a 300-person team building infrastructure for Google's major products. Camus Energy recently surpassed $25M in Series A funding, co-led by Congruent Ventures and marketplace investor Wave Capital.
Read on for highlights from our conversation to dive into the inception of Camus Energy, the problems they are solving, and why her previous experience at Google has been critical to their approach.
From Google exec to founding Camus Energy
What inspired you to make the leap from Google to start Camus?
I had been at Google for about 15 years and was leading a 300-person team building infrastructure for Google's major products. But I didn't want to spend my whole career there. In parallel, I had gotten very concerned about climate change and interested in making a future career transition into that space, especially after my son was born.
I was researching the climate and energy space for 5-6 years and got very interested in challenges around how we manage the grid. There were parallels between managing supply and demand for energy on the grid with the technology approaches we developed at Google for managing the physical internet network. When my colleague Cody, who I had worked with for 15 years at Google, reached out about new roles, that was the catalyst. I was like, I have an idea for a company, let's talk.
The problem Camus Energy is solving
Can you talk more about the problems grid operators are facing and how Camus is helping?
The grid has a critical role to play in decarbonization - we need to first electrify everything and second decarbonize the grid. That means the grid needs to take on significant additional loads beyond what it does today. It's a physical network with physical network constraints.
As we put more work on the grid, we need a lot more decarbonized generating capacity from renewables and other sources. But we also need to figure out how to move that energy across the network efficiently. Just building our way out of it is prohibitively expensive - estimated to cost over $21 trillion by 2050. We're also moving from a one-way grid where centralized power gets consumed by end users, to one with decentralized generation and a lot of activity on the local network - EVs, batteries, solar, etc. We need a management model that can handle that.
That's where Camus comes in. We provide software for utilities to understand that complex local energy environment through data integration first. Then, we also help to optimize and control these resources to get the most out of the network and enable building more efficiently. Our deep tech experience allows us to build the "brain" to optimize across all the data.
Where do you see DERMS falling short compared to what you’re building at Camus Energy?
DERMS typically manage distributed energy resources like solar panels or batteries. They don't have good visibility into network conditions for the assets they're managing. That makes it difficult for them to be the “brain” that optimizes on a grid-level.
For instance, you want to spread out EV charging in your neighborhood to preserve transformer capacity, or you might want to curtail solar resources in times of overproduction. DERMS systems struggle to make these system level decisions about how those assets should behave, because they lack data-driven visibility into the state of the grid.
What role does AI play today and in the future for what you’re building at Camus Energy?
For large-scale, high reliability (you want zero failures) automation like for the grid, you wouldn't normally start with AI. AI is fundamentally taking in data and then producing probability-based rather than certainty-based outcomes. You want your results to be really predictable, transparent, reproducible and somewhat robust to variability and input data, all of which can be issues for AI technologies.
That’s why we’re using machine-learning and heuristics-based control to help us forecast behavior across many datasets for activity behind an individual electricity meter or fill in gaps in time series data.
But the day will come when we use AI to manage the grid. When complex, multi-variable input data is required, model-based solutions might produce potentially much better solutions than heuristics-based outcomes would.
Applying Google-scale engineering to the grid
In the past, you've talked about your theory of generalized infrastructure development. What is it and how do you apply it at Camus Energy??
At Google, a lot of my work was on building shared infrastructure services for large-scale products. We had to develop an approach to build generalized solutions that could solve a common problem for multiple use cases and scale massively. Many infrastructure projects failed because they were too specific, or were developed in isolation without enough input from the teams that would use them.
What worked was identifying a problem area with potential for a generalized solution, partnering early with 2-3 customer teams to develop an initial solution that solves their specific needs, and then immediately taking it to a large-scale customer to adapt it for scale before the architecture is set.
We're taking a similar approach at Camus. We're working closely with a set of early utility partners, including one very large customer, to make sure we're building a solution that actually solves their challenges around visibility, control, and optimization of distributed energy resources at scale.
How do you navigate the long sales cycles that come with selling to large utilities who are reluctant to adopt new technology?
Managing the grid is fundamentally an operations job focused on maintaining system stability. One of the easiest ways to do that is to not change things.
That makes it particularly difficult to convince someone to adopt technologies that were not really built with a deep understanding of the system challenges that they're meant to address. Instead, you need to bring solutions that are solving problems that they have. And then you need to be able to solve problems in a way that is deeply reflective of the problem space as they're experiencing it. For me, having worked in infrastructure at Google was a plus that helped me to do that for infrastructure operators at utilities.
Advice to climate tech founders
What's your advice for other founders embarking on a tough tech challenge?
1, Really understand the problem you're solving and make sure you're developing solutions closely with customers and partners.
2, Think hard about your initial team - you need people with deep expertise across the stack, from domain expertise to the ability to build extremely scalable and reliable software systems. Don’t hesitate to make leadership changes soon when you realize that you need expertise that you don’t have on the team.
3, There is no time that's too soon to start conversations with venture capital investors if you want to build a VC-backed company. Their advice was critical to building a company that was fundable and also built relationships that helped us. If you’re unfamiliar with venture capital, I would recommend Venture Deals and Secrets of Sand Hill Road to understand the language and motivations of folks in the venture space.
4, Learn how to build and manage a board early. If you have an opportunity to get involved in even a nonprofit board, that's pretty helpful for understanding how they operate.
5, Lastly, don't be afraid to tackle hard problems. The grid is critical infrastructure and transforming it is complex and high stakes. But it's also one of the most important levers we have to address climate change. With the right approaches, we can evolve the grid to enable a cleaner, more resilient, and more equitable energy future.