Tellus Power CEO, Mike Calise, on localized energy, physical AI, and what it will take to modernize America’s failing grid
The conversation around artificial intelligence and energy has largely centered on data centers and the demand they place on a grid that was never designed to handle it. But there’s a quieter infrastructure story unfolding at the edge — one that involves microgrids, bidirectional charging, and the growing fleet of autonomous vehicles beginning to populate American roads.
Tellus Power is building at that intersection. The company specializes in localized energy systems designed to put power where it’s needed, when it’s needed, without waiting on a centralized grid that the American Society of Civil Engineers recently graded a D+. With U.S. electricity demand projected to grow 25% by 2030 while actual generation increased just 2.5% last year, the gap between supply and need is becoming impossible to ignore.
The EV Report connected with Tellus Power CEO, Mike Calise, to talk through what that gap means for EV infrastructure, how autonomous vehicles are reshaping energy planning, and why the future of the American grid may depend less on massive transmission projects and more on thousands of smaller, smarter systems working in tandem.
Given the massive energy strain that Artificial Intelligence has on the overall grid, how has AI impacted your business as you work to localize energy with microgrids as well as the EV industry and partners?
The ICF forecasts that electricity demand in the US is expected to grow 25% by 2030, meanwhile, the actual electricity that the US generates only increased by 2.5% last year. That’s not going to cut it.
As we look to strengthen grid infrastructure, one of the key problems is that everyone is drawing to the same proverbial “well” for energy. Whether it’s a data center, a grocery store, or a home, they all rely on a single grid to provide electricity.
AI is accelerating the need for high-availability, high-density power at the edge, particularly from data centers and the electrification ecosystem around them. That shift has directly impacted Tellus Power’s business because it’s changed what our customers need, and therefore what we’re building.
First, we’re leaning into power management architectures designed for data centers and other large loads, prioritizing resilience, power quality, and speed of deployment over waiting years for traditional grid upgrades.
Second, it’s pushed us to accelerate our roadmap toward V2X and bidirectional power capabilities, because EV infrastructure can no longer be treated as a passive load. In an AI-driven energy environment, large loads are increasingly expected to be flexible, dispatchable, and resilience-supporting.
Third, AI has changed how we partner. We’re forming deeper partnerships across distributed energy resources, so we can deliver integrated systems that reduce peak demand, improve reliability, and relieve pressure on the centralized grid.
In short, AI is driving a power problem, but it’s also driving a market shift. We’re responding by building the systems and partnerships that make localized, resilient energy practical at scale.
Walk us through how you go about localizing energy. How are these microgrids self-sustaining and not adding more stress to the overall power grid?
Localizing energy means shifting a meaningful portion of generation, storage, and control closer to where electricity is consumed, rather than relying entirely on a centralized grid.
Microgrids achieve this by integrating distributed energy resources, like solar and battery storage, into a coordinated system. These systems can operate independently when needed or remain grid-connected while prioritizing local supply.
What makes microgrids self-sustaining is not a single energy source, but how the system is designed. Local generation covers a baseline of demand, storage absorbs variability and provides resilience, and control systems manage when to draw from or supply power to the grid.
Because of that design, microgrids typically reduce stress on the broader grid. They lower peak demand, smooth load volatility, and limit the need for constant upstream power delivery, especially during high-stress periods. In some cases, they can even support the grid by exporting power when conditions allow.
Microgrids don’t compete with the grid, they complement it by absorbing volatility locally and making overall demand more predictable and manageable.
When people think of AI, typically, they think of names like ChatGPT or Gemini. How important is it to highlight and expand the AI narrative to show the growing surge of physical AI, like autonomous vehicles?
The reason why the AI narrative is dominated by the ChatGPTs and Geminis of the world is because more people are using them — they are easily accessible to the public, who can simply open an app on their phones and use AI.
But that’s only part of the story. What’s equally important, and often less understood, is the rapid rise of physical AI, where intelligence is embedded directly into machines that move, transport, store, and deliver things in the real world.
Unlike digital AI, physical AI doesn’t just consume compute, it consumes energy, space, and infrastructure. Autonomous vehicles, warehouse robotics, delivery systems, and industrial automation all depend on reliable, localized power to operate safely and continuously.
We’re already seeing this shift. Companies are deploying AI-driven robotics in logistics and manufacturing, and autonomous vehicle platforms from firms like Waymo and Zoox are scaling real-world operations. To put Waymo’s success into perspective, the company announced it reached 250,000 paid robotaxi rides per week in April. In December, recent data from Tiger Global Investment firm showed the company jumped to 450,000 weekly paid rides.
As these systems become more common, the constraint isn’t the intelligence, it’s whether cities and operators have the physical infrastructure to support them. That’s why expanding the AI narrative matters. Physical AI turns AI from a software conversation into an infrastructure conversation. To support it at scale, energy has to be available at the edge, systems need resilience, and grids need to be designed for flexibility rather than just centralized supply.
In order to power the next wave of physical AI, Tellus Power is working to ensure energy is available at the edge with microgrids and V2X technologies, so autonomous vehicles are able to keep running. It’s also crucial that cities have this technology in place so they’re able to welcome more autonomous technologies without worrying about if they have the infrastructure to charge them.
Soon, physical AI will move beyond just the delivery robots and autonomous rideshare services. Elon Musk is already predicting that Tesla’s humanoid robots will be some of the world’s best surgeons in just three years. While the possibilities of AI are endless, it’s critical that we work to get energy to the edge to support these physical AI innovations out in the field, and as we do that, it’s crucial that we revamp the overall grid to support the LLMs.
2025 was a big year for autonomous vehicles as Waymo expanded in states like Georgia, Texas, and Florida. While EV sales ebb and flow, autonomous vehicles are increasing at a rapid pace. When you’re placing and distributing microgrids, what’s more important, EV adoption in the area or the presence of these autonomous vehicles?
In practice, it’s less about choosing between EV adoption or autonomous vehicles, and more about understanding how energy is actually being used in a given area. Oftentimes, the availability of autonomous rideshare services and EV adoption rates go hand-in-hand. Companies typically won’t expand autonomous services in states and cities with particularly low EV adoption rates.
Autonomous vehicles tend to be far more energy-intensive and predictable than consumer EVs. They operate longer hours, charge more frequently, and often require higher uptime and faster turnaround. From an infrastructure standpoint, that makes them a stronger signal for where localized power and charging capacity are needed.
However, autonomous deployment rarely happens in a vacuum. Companies typically expand into markets that already have supportive EV policies, charging infrastructure, and grid readiness. So while EV adoption creates the baseline conditions, autonomous vehicles often become the primary driver of sustained load.
When placing and distributing microgrids, what matters most is utilization density and reliability requirements, not just vehicle counts. Fleets, depots, and autonomous service zones create concentrated demand that benefits from localized generation, storage, and resilience, especially as these systems scale. Localized energy reduces demand charges, improves uptime economics for fleets, and shortens deployment timelines compared to traditional grid upgrades.
EV sales will always move in cycles, but autonomous systems represent a more structural shift. They turn transportation into a continuous, infrastructure-dependent service, which is why energy planning increasingly follows where autonomy is being deployed, not just where EVs are being sold.
In terms of increasing the number of microgrids and EV charging stations, it’s been a rollercoaster. A district judge ruled the $5B in NEVI funds had to be released, now Congress is debating whether to cut $500M from those funds. How do you gameplan for infrastructure projects when there’s so many changes?
Recently, the fed allowed states to take more control and streamline infrastructure deployments according to individual state needs. District Judge Tana Lin’s decision to release the $5B in NEVI funding was a major step forward for the health of local power grids, especially at a time when autonomous rideshare availability is right at the inflection point. But with only a small percentage of that funding actually awarded to projects so far, and Congress now debating a potential $500M reduction, uncertainty remains.
Despite that volatility, states like Ohio, Pennsylvania, and California have approved FY2026 plans and continue issuing documentation for 150kW+ corridor infrastructure, signaling that momentum hasn’t stalled. California in particular has removed significant red tape, streamlining certain approvals to as little as five business days, even as most authorities having jurisdiction still lack standardized processes. Those policy decisions in the nation’s largest EV market tend to ripple outward, shaping deployment strategies in other states.
Given that environment, companies can’t afford to rely solely on government funding to drive projects forward. The focus has shifted toward maximizing the utilization of available capital, while equipment manufacturers continue supplying hardware into private and NEVI-aligned deployments. Industry leaders are closely monitoring these headwinds and tailwinds, adjusting timelines and partnerships to ensure projects can move sooner rather than later — even as the funding landscape continues to evolve.
While last year was positive for the expansion of autonomous vehicles, it also shed light on the need for American grid infrastructure. 1.9 million customers were affected in California wildfires and nearly 450,000 homes in Pennsylvania lost power due to severe wind. How can building out smaller microgrids help prevent these massive power outages?
With the examples of California and Pennsylvania, both would fall under the category of an “unplanned outage event.” These happen when weather systems severely damage a power grid or local power lines. While these outages cause a great inconvenience to the public, they are typically hard to predict and are therefore difficult to ever entirely prevent.
What’s changing now is the frequency of these severe weather events. The US experienced twice as many weather-related outages between 2014 and 2023 as it did between 2000 and 2009. There are two main reasons for that surge in outages. Firstly, dramatic weather events are becoming more commonplace, and secondly, the grid infrastructure we have in place needs significant repairs because they keep getting beaten up by these storm systems.
Back in October 2023, the US Department of Energy released data showing 70% of transmission lines were either 25 years old or approaching the end of their typical 50-80-year lifecycle. While we can’t snap a finger and make weather events go away, we can and need to prioritize infrastructure projects that build new essential transmission lines that can last another 50 to 80 years.
While building out new power lines will help protect Americans from unplanned power outages, microgrids can be especially helpful when managing overall grid strain during periods of high electricity use — think summers in California when everyone is using their ACs.
In an instance like this, utility providers are often forced to throttle the flow of electricity in certain parts of the grid to prevent sparking, fires, or transformer overload that can quickly accumulate into down-system failures. If the utility can predict the load demand correctly and the public participates enough to mitigate a critical load event, then the hope is that no outages will occur.
Over time, infrastructure repair and futureproofing will be the key to unlocking our fully electric future. Microgrids are part and parcel of this future-facing effort: by integrating a diverse and flexible array of energy products like batteries, bi-directional chargers, solar panels, etc., into the pre-existing grid, Tellus Power is reinforcing the outdated areas and offering more outlets for load shedding events. These smaller devices interacting at the microgrid level offer the most flexible option for modernizing our grid.
Microgrids essentially act as back-up generators. If you’re supplementing the grid with microgrid infrastructure, it allows you to have that backup generation that better localizes energy demands and needs based on the type of need. In cases of emergency, it can be utilized to supplement the energy.
The American Society of Civil Engineers gave the US energy sector a D+ grade — down from C- in 2021, saying the grid is aging and fragile. How does the US turn that grade around in 2026?
The U.S. can reverse its failing grid grade by transitioning from an antiquated “command and control” system which features stronger power grids accompanied by a network of decentralized, bidirectional microgrids. By aggregating standalone energy resources into microgrids, we create virtual power plants that provide inherent resilience without the need for costly new transmission lines, substations, or traditional peaker plants.
This shift modernizes the grid by bringing energy to the edge, placing power exactly where and when it is needed, a requirement driven by the expansion of EV charging, autonomous transportation, and a growing robotic workforce.
The current grid is limited by its unidirectional design, which often results in valuable energy, such as solar power, being wasted or “thrown in the ground” due to a lack of storage and localized distribution. By implementing bidirectional technology, we can maximize asset utilization, allowing energy to flow where it is most needed or back into the grid. Ultimately, this approach replaces outdated infrastructure with cleaner, smarter technology that is specifically designed to handle the high-utilization demands of a modern, mobile, and robotic economy.
By utilizing DC-to-DC technology, Tellus can bring energy where you need it, when you need it, and it harnesses massive amounts of energy to give you the exact electron profile you need for your specific use case.
What is the biggest hurdle for America as we look to revamp our grid infrastructure?
The primary hurdle for revamping America’s infrastructure is a widening “supply-demand gap,” where the requirement for electricity is growing incongruently with the grid’s capacity to deliver it. Currently, the U.S. is struggling to match supply with immediate demand, but the long-term challenge is a compound annual growth rate for electrons that far outpaces generation and transmission buildouts.
While the ICF forecasted U.S. electricity demand could surge by as much as 25% by 2030, driven by AI data centers and electrification, actual generation only increased by approximately 2.5% last year. This disparity necessitates a shift toward extreme utilization and efficiency, as the traditional model of centralized supply cannot scale fast enough to meet the compounding year-over-year need for localized, high-capacity power at the edge.
The real challenge isn’t technology. It’s deployment velocity. Permitting timelines, interconnection queues, and capital allocation cycles are still aligned with a 20th-century grid model, while AI, autonomy, and electrification are scaling on a software timeline. Infrastructure must begin moving at digital speed if we expect to close the supply-demand gap.
Where do you see the grid infrastructure five years from now?
In five years, grid infrastructure should look significantly different because the landscape of products needing energy “at the edge” shows rapidly growing demand.
Autonomous vehicles, smart fleets, humanoid robotics, and automated distribution robotics are surging in adoption. Companies like Amazon are expanding their use of in-house logistics and last-mile delivery vehicles, increasing the need for more charging stations and microgrid infrastructure. Uber announced plans to expand these AV capabilities to serve as delivery robots. Recent data shows the global automated guided vehicle market is projected to grow from $2.68 billion in 2025 to $4.72 billion by 2032. Cities are going to need thousands of more microgrids in order to keep up with the increasing adoption rates of autonomous vehicles.
At Tellus Power, we’re looking to drastically increase the amount of microgrids with DC fast charging, stationary storage, and vehicle-to-grid technology currently available in markets across the US. The industry is moving towards making DC standalone EV chargers an absolute minimum requirement, and making microgrids with V2G charging capabilities an industry standard. In the not so distant future, there will be two energy systems working in tandem with each other. The centralized grid will be able to handle the energy demand from AI data centers, and decentralized microgrids will power AI “at the edge,” that ultimately when aggregated are the building blocks for virtual power plants (VPPs).
With the proper build-out of microgrid infrastructure and VPPs, the US will be better equipped to handle more EVs, as well as other electrified systems requiring energy at the edge.
The throughline of this conversation is urgency. Whether it’s aging transmission lines, NEVI funding battles, or a power grid earning failing grades, the infrastructure challenge facing the U.S. isn’t a distant problem — it’s arriving at the same pace as the autonomous vehicles and AI systems that will depend on it.
Tellus Power’s thesis — that localized, bidirectional microgrids are a faster, more flexible path to grid resilience than waiting on centralized upgrades — is gaining traction in a policy and market environment that is, at the very least, forcing the issue. How quickly deployment can match the rhetoric will be the defining question for the industry over the next several years.
For more on grid infrastructure, EV charging, and autonomous vehicle expansion, visit Tellus Power North America at telluspowernorthamerica.com.
Sign up for our popular daily email to catch all the latest EV news!
