For decades, the electrical grid earned its best reviews by being invisible. When power flows reliably, most people don’t think about transmission lines, interconnection queues, or how utilities balance supply and demand minute by minute.
That anonymity has been fading. High-profile crises — including fires in California and freezes in Texas — pushed the grid into public view. In 2025, it moved from background infrastructure to front-page concern as worries about demand, supply, pricing, and pressure on natural resources converged. At the same time, a wave of startups has been pitching software as a practical way to stretch existing capacity and make the system more resilient.
Electricity rates are up 13% in this U.S. this year, driven in part by an AI boom that is showing up in unexpected places, from repurposing supersonic jet engines for data center duty to efforts aimed at beaming solar power down from space. The surge in power consumption is not expected to ease: data center electricity use is projected to nearly triple in the coming decade.
That growth forecast is intensifying consumer frustration over pricing and drawing criticism from environmental groups, some of which have called for a nationwide moratorium on new projects. Utilities, long accustomed to operating out of the spotlight, are being forced to accelerate upgrades and plan new power plants to meet rising loads — all while a fear lingers in the background: if an AI bubble bursts, expensive infrastructure could become stranded.
This mix of urgency and uncertainty is exactly where software is being positioned as a faster, less capital-intensive lever to pull.
Why software is becoming a grid strategy, not just a tool
The modern grid is no longer a one-way system where centralized power plants push electricity outward to passive customers. Wind and solar farms vary with weather, batteries can both consume and supply electricity, and large new loads like data centers can arrive quickly and concentrate demand in specific regions.
In that environment, building more hardware — new power plants, new transmission lines, new substations — is often necessary, but it can be slow, politically difficult, and expensive. Those investments are also long-lived, which makes them risky when demand projections are uncertain. Utilities can be cautious for good reason: reliability is non-negotiable, and mistakes can cascade.
Software promises a complementary approach. If it can help utilities locate spare capacity, better coordinate distributed energy resources, or speed up the process of connecting new generation, it could reduce the amount of new infrastructure needed immediately — or at least buy time while long-term projects move forward.
Startups hunting for spare grid capacity
One of the most direct software pitches is that the grid already has underused capacity — it’s just hard to find, verify, and use quickly. Two startups highlighted in recent reporting illustrate how that search is becoming a product.
Gridcare: combining infrastructure data with local constraints
Gridcare has compiled a broad set of inputs — including data on transmission and distribution lines, fiber-optic connections, extreme weather, and even community sentiment — to identify locations that might support new large electricity users. The idea is not simply to point to a spot on a map, but to provide utilities with enough evidence to be confident the local grid can handle it. The company says it has already identified several sites that had been overlooked.
Yottar: matching known capacity with mid-size demand
Yottar takes a related but more targeted approach. It looks for places where capacity is already known to exist, then matches those locations to the needs of medium-size users. In the current data center boom, the ability to connect faster can be a competitive advantage, and software that reduces the friction of site selection and grid access becomes more valuable.
Both approaches reflect a broader shift: instead of treating the grid as a fixed constraint that simply forces projects to wait, startups are trying to make grid capability more legible and searchable — and, ideally, faster to act on.
Virtual power plants: turning scattered batteries into dispatchable capacity
Another prominent software trend is the creation of “virtual power plants.” The concept is straightforward: rather than relying solely on big, centralized power plants, a company aggregates and coordinates many smaller energy assets — especially batteries — and operates them as if they were a single power source.
When demand spikes or supply drops, these aggregated batteries can discharge to support the grid. When supply is abundant or prices are low, they can recharge. The “virtual” part is the software layer that monitors, forecasts, and dispatches the fleet.
Base Power: a Texas fleet built through home battery leasing
Base Power is building a virtual power plant in Texas by leasing batteries to homeowners at relatively low prices. For customers, the battery is a backup power resource during outages. For the company, the aggregated fleet can be used to help prevent outages by selling that combined capacity back to the grid when it’s most needed.
Terralayr: bundling distributed storage on the German grid
Terralayr is pursuing a similar virtual power plant model in Germany, but with a key difference: it doesn’t sell batteries itself. Instead, Terralayr uses software to bundle distributed storage assets that are already installed on the German grid.
These models try to solve a problem that is becoming more common: the hardware is increasingly distributed, but grid value often depends on coordination. Without orchestration, batteries can sit idle or respond in ways that don’t align with what the wider system needs.
Coordinating wind, solar, and storage so they “idle less”
Beyond batteries alone, several startups are focused on the software layers required to integrate and coordinate a wide mix of distributed energy resources. Recent reporting points to companies such as Texture, Uplight, and Camus, which are working on systems designed to help wind, solar, and batteries operate together more effectively.
The goal is to increase utilization and grid contribution. In practice, that can mean:
- Better forecasting of supply and demand so the right resources are available at the right time.
- Smarter dispatch so assets respond in ways that reduce strain on local circuits and substations.
- More coordination across devices and sites to reduce waste and improve reliability.
These software layers matter because renewables and storage don’t behave like traditional power plants. Their output can vary quickly, and their value often depends on how precisely they can be scheduled and controlled.
Modernizing outdated processes with AI
Some of the biggest opportunities for software may lie in modernizing the grid’s more outdated workflows — the parts of the system where paperwork, legacy tools, and slow processes turn into real-world delays.
Nvidia and EPRI: industry-specific models for efficiency and resiliency
Nvidia has partnered with EPRI, a power industry R&D organization, to develop industry-specific models intended to improve grid efficiency and resiliency. While details and results will take time to emerge, the direction signals that major technology players see the grid as a domain where specialized AI could have practical operational impact.
Google and PJM: using AI to work through interconnection backlogs
Google is working with the grid operator PJM to use AI to help sift through a backlog of connection requests from new sources of electricity. Interconnection queues have become a critical bottleneck in many regions: even when new generation is ready to be built, getting approval and securing the necessary studies can take a long time.
If software can meaningfully reduce processing time or help prioritize projects more effectively, it could accelerate the arrival of new electricity supply without immediately requiring massive new physical buildouts.
Why utilities may adopt software faster than new infrastructure
Even with mounting pressure, utilities are typically cautious adopters of new technology because reliability is paramount. But they also face constraints when it comes to building: new infrastructure is expensive, takes years to complete, and is designed to last for decades. Ratepayers and regulators often push back when large projects threaten affordability.
Software changes the equation. It is generally cheaper than physical construction, faster to deploy, and easier to iterate. If software vendors can meet the grid’s reliability requirements, they may find a more receptive market — especially as utilities try to manage rising demand without committing prematurely to expensive, long-lived assets in an uncertain climate.
None of these improvements will appear overnight. Still, the groundwork being laid now suggests 2026 could be the year these approaches begin to take hold at larger scale.
The bigger picture: more electricity is coming, with or without hype
Even if parts of the AI boom cool, broad electrification trends are likely to keep pushing demand higher. Transportation, heating, and other segments of the economy are shifting toward electricity. Meanwhile, the volume of planned data centers continues to reshape regional load forecasts and grid planning.
That means the grid will still need refurbishment and expansion. But software can help ensure that upgrades happen in a smarter order: finding overlooked capacity, coordinating distributed resources, and reducing administrative bottlenecks that slow down new connections.
Conclusion
The electrical grid is being asked to do more, faster, under greater scrutiny. With electricity rates up 13% in this U.S. this year and data center power use projected to nearly triple in the coming decade, software is emerging as a pragmatic way to unlock capacity, orchestrate distributed resources, and modernize operations while larger infrastructure projects catch up.
Related Articles
- Climate Tech in 2026: Investors See Data Centers, Power Demand, and Grid Resilience Driving the Next Wave
- Why Social Media Follower Counts Matter Less in 2025 as Algorithms, AI “Slop,” and Niche Communities Take Over
- Meta Acquires Manus: Why the Fast-Growing AI Agent Startup Matters for Facebook, Instagram, and WhatsApp
Based on reporting originally published by TechCrunch. See the sources section below.
