Electric vehicle adoption has seen stellar growth in the last several years. The International Energy Agency (IEA) notes that public EV charging ports in the US doubled between 2020 and 2024 alone, reaching almost 200,000 thanks to the rapid expansion of DC fast chargers and OCPP-based software. In addition, over two million more ports will be required to support projected demand by 2030. What’s it going to take to scale EV charging technology to meet this challenge head on, and in a way that puts customer needs front and center? To answer this, we spoke with Epic Charging Co-Founder and Chief Technology Officer Michael Fridshtand, who brings 15 years of experience in software development, AI, VR, blockchain and big data.
From your perspective, what are the biggest technical challenges in building an EV charging software platform that can scale reliably across different charger manufacturers, site types and use cases?
At first, we were just working with a couple manufacturers, and it was pretty easy to integrate our software because they all followed the OCPP protocol standard. Everything was documented, so everything operated fine. But once we started working with Chinese, Taiwanese and other foreign brands, it became much more challenging because everyone interpreted the protocols differently. That made it pretty tough to scale, especially with the older legacy chargers, because it took time to plan and to create a scalable architecture. But now we have a strict process. How we do things. How we onboard. How we implement changes if needed. Scheduling calls with the engineers. And then the more chargers you onboard, eventually the less differentiation you see. We’ve already implemented everything you could possibly imagine. Today, we probably have hundreds of different types of chargers on our platform, so we’ve seen it all.
Epic Charging works across fleets, multifamily and large commercial accounts, so how do you think about platform architecture in a way that supports very different operational needs without creating unnecessary complexity?
My background is actually building large scale systems with big data. Here, the scale is not that big. We’re talking 20,000 chargers compared to, say, WhatsApp Messenger where you have a billion users across the world, which makes it way more complex. In Epic’s case, the architecture was built from day one to be scalable – to easily go from supporting 100 chargers to 100,000 chargers across a variety of use cases. And it's actually not as complex as you might think. Our platform is built around microservice architecture, which is like having a bunch of small boxes with different functionalities, and they all interact with each other. If the platform grows bigger, we can then split it into even smaller parts, which helps to support the complexity of the whole system. But if you don’t plan it this way from the start, and then you face issues where you can't scale anymore, you need to redo it all from scratch.
With interoperability being such a major theme in EV charging, how do you think about OCPP, hardware variability and integrations when building a platform that needs to remain flexible but also dependable in the field?
We do interoperability tests for every charger that we onboard. And we've built essentially a profile for every charger. So if a charger interacts a bit differently with the OCPP – and it doesn't matter which charger we're working with – we have a small layer of profiling that allows us to adapt to the charger design. For example, with one charger, we found if we sent a new message before we replied to a previous one, it created issues with the firmware and we would need to reboot it. So we created a small profile that said, okay, if a message comes to the charger, reply to it before sending the next one. But for every charger, it will be something different.
As the market matures, what technical capabilities do you believe will separate leading EV charging software platforms from others over the next few years?
We actually had an internal discussion recently about technical supremacy and how we could be leading there.
So our advantage will not solely be a technical one. First, we will need to implement AI inside every department in our company, from the CEO to the technical personnel. We need to show how it will help every aspect of their jobs. It's a shift in the mindset of the engineers, the software developers, the management team. Instead of writing code every day, you manage agentic AI agent to do it. Second, we’re working directly with the hardware, and we have all the certifications, like CTEP, which takes time to get. So we have an advantage on the business side, too, over pure software companies or the kid in his dorm room.
How has your thinking evolved over time when it comes to balancing speed of product development with long-term platform reliability and maintainability?
Remember the old adage, “quality, speed or price, choose any two?” Well, now you don't need to choose. Having all three is actually possible. You can decrease the cost, improve the speed and improve the quality, all with AI. We might have an engineering team of 10 people, while the closest competitor has 30, and we already did more than they did without AI. But now with AI, we can increase our overall performance two or three times over without having to expand the team. So in terms of the cost, you save money using AI. As to the quality, the AI will review every piece of code the software developer writes and will suggest changes to increase operability and efficiency. We recently had an issue that two developers tried to solve for two working days. Finally, they went to Claude, and the second thing it suggested was actually the solution. It took just 15 minutes using AI. And that happens every day.
You obviously see AI as a major part of Epic’s future, but where do you see it creating the most value for customers?
Downtime makes customers unhappy. Every minute a charger is not charging, it's a missed opportunity to earn money. So when it comes to reliability of the chargers, AI can be really helpful. It can help solve issues, not just on the software development side, but on the support side. You can use it to analyze the charging logs. Or integrate it into your support stack to reply to customers faster. It not only helps us as a business, it helps customers solve their issues faster. I’d say our ultimate goal is to improve the customer experience, to give them more reliable chargers, more reliable networks, more reliable sessions. And AI helps us do that.
Click here to learn more about the Epic Charging team and the experience they bring to the world of EV charging solutions.
Click here to learn more about the Epic Charging team and the experience they bring to the world of EV charging solutions.