When a winter storm hit Texas in February, it has brought some of the coldest weather the state has seen in decades. The deep freeze led to a massive electricity generation failure, leaving millions of people without power for several days.

The ensuing power crisis revealed multiple problems with how the electric grid is operating in the state. Among these problems was that when a sudden power shortage happened, the grid wasn’t smart enough to send electricity where it was most needed. For example, in Houston and Austin, empty skyscrapers were lit up at night while people in the surrounding neighborhoods were left with no electricity at all.

As explained by Fast Company, in those cities, the downtown skyscrapers were all connected to a central circuit, but that was also the case for the hospitals, meaning that the power couldn’t be turned off without also shutting off the power for the hospitals.

In a bid to come up with a solution to this problem, Intel has recently developed a platform that can help the grid communicate with individual buildings in real-time. “It’s integrated with the grid and the building,” says Michael Bates, Global General Manager of Energy at Intel. “The grid, which is sensing the power needs across all of its territory, can turn things off and know exactly what is turning off, and keep other things on.” Essentially, this means that a hospital could stay on while other buildings in the vicinity could shut down temporarily.

The platform is also designed to work with renewable energy coming from multiple, intermittent sources rather than one centralized power source, such as a coal or gas plant. This makes the grid more resilient, enabling grid operators to get power from one location while others are down.

Such a system is also key to making renewable energy feasible since grid operators need to be able to rely on renewables in the same way they rely on coal plants that were always producing energy.

“Highly distributed, intermittent sources of energy need to, at some point, look and act like that baseload, or you’re never going to replace the coal plant,” says Bates. “If it’s too intermittent, the business model breaks down. With the application of real-time, AI decision making, you can consolidate all those distributed loads, and then smooth out the peaks and valleys by balancing it and make it deliver energy in the same way the coal plants are delivering energy.”