The Thirsty Llama: Meta’s New AI and the Water Crisis
Hold on to your hats, people. Meta’s new AI model, Llama 2, has just been released and it’s making waves. Not the good kind, mind you. We’re talking about the kind of waves you get when you’re guzzling down water like it’s going out of style.
The Water Footprint of Llama 2
Shaolei Ren, an associate professor at UC Riverside, has been keeping an eye on this. He’s the guy who looks at how much water is used when data centers power up to train these new AI models. And let’s just say, he’s not thrilled.
According to Ren, training Llama 2 has a water footprint of 10.9 million liters. That’s 2.8 million liters if you exclude hydropower. To put that into perspective, the average human is recommended to drink about 3 liters of water per day. Llama 2, on the other hand, is drinking enough water to hydrate a small country.
Meta’s Silence and the Impact on Data Centers
Now, Meta didn’t exactly announce how much water they used to train this AI model. But they did disclose power consumption. Ren took that data, looked at how efficient Meta’s data centers are when it comes to using energy and water, and came up with his estimate. And it’s not pretty.
In fact, it’s almost double the water footprint of Meta’s previous AI model, Llama 1. And let’s remember, data centers already use a lot of energy. With the AI boom, this consumption is expected to skyrocket.
Ren warns, “If the power usage increases, everything else like carbon and water footprints will also increase.” Meta, for their part, hasn’t responded to Ren’s estimates.
The Future of Meta’s Data Centers
In 2021, Meta’s data centers used just over 5 million cubic meters of water. That’s about 1.33 billion gallons. And they’re planning to build and expand a data center in Arizona, a state that’s running out of water.
The Impending Disaster
So, here’s the deal. We’ve got a new AI model that’s thirsty as hell, a tech giant that’s planning to build in a state that’s drying up, and a world that’s grappling with a water crisis. It’s a recipe for disaster, and it’s time we start paying attention.
