This is the place Atlantic Ocean Intelligence, a limited-run series in which our writers help you wrap your mind around artificial intelligence and a new machine age. Sign up here.
In last week's newsletter, I described artificial intelligence as data hungry. But the technology is also pretty thirsty, relying on data centers that not only require a lot of energy, but water to cool themselves.
Karen Howe, a contributing author Atlantic Ocean, recently visited one such data center in Goodyear, Arizona. Microsoft owns the facility, which could eventually use an estimated 56 million gallons of drinking water each year—“the equivalent of the amount used by 670 Goodyear families,” notes Karen. No one is in danger of going thirsty, but as Kern writes, “water supplies in the region are quite limited, and the more work is done by data centers, the less water there is to, say, new houses.” water supply.
I followed up with Karen to ask about the growing demands of AI on our environment. It's still a matter of debate whether the technology is really worth its huge costs, even as tech companies commit more and more resources to it. How should we think about all this? “Companies are building data centers faster than ever in the race to build generative AI, but there's little accounting for their impact on the environment,” Karen told me. “There is a narrow window in which the public must ask: Is this what we want? Once the facilities are built, it will be much more difficult to reverse that decision.
– Damon Barris, Senior Editor
AI is taking water from the desert.
By Karen Howe
On a sweltering day this past September, I made the risky decision to try to make the rounds of some data centers. The ones I chose sit between a regional airport and some farm fields in Goodyear, Arizona, a half-hour drive from downtown Phoenix. When my Uber pulled up next to the unmarked buildings, the temperature was 97 degrees Fahrenheit. The air crackled with a latent energy, and there was some kind of pulsating noise coming from the power lines above my head, or perhaps from the buildings themselves. With no shelter from the blinding sunlight, I began to lose my sense of being real.
Microsoft announced its plans for this location, and two others not too far away, in 2019—a week after the company disclosed an initial $1 billion investment in OpenAI, the buzzy startup that later I will release ChatGPT. From that point on, OpenAI began training its models exclusively on Microsoft's servers. Any query for the OpenAI product will be directed to Microsoft's cloud computing network, Azure. To meet this demand, Microsoft is adding data centers at a staggering rate, spending more than $10 billion on cloud computing capacity each quarter. One semiconductor analyst called it “the greatest piece of infrastructure construction humanity has ever seen.”
I went to Arizona to see it for myself. The Goodyear site stretched further along the road than my eyes could see. A black fence and patches of desert vegetation lined its perimeter. I grabbed my phone and two bottles of water and started walking the length of it. Microsoft bought 279 acres for the location, according to city documents. For now, the plot consists of two prefabricated buildings, thick and squat, with vents and pipes visible on the sides. A third building is under construction, and seven more are to be built. Each will be outfitted with rows of servers and computers that must be kept below a certain temperature. The complex is partially designated for use by OpenAI, according to a person familiar with the project. (Both Microsoft and OpenAI declined to comment on the claim.) And Microsoft plans to absorb its excess heat through constant airflow and the evaporation of potable water as needed. Consumption of the latter is expected to reach more than 50 million gallons per year.
Read the full article.
What to read next?
P.S
Earlier this week, President Joe Biden signed legislation that could result in a ban on TikTok if the app is not spun off from its Chinese parent company. As Charlie Warzel writes. Atlantic Oceanit'll be a more complicated process than it sounds — especially when it comes to the app's powerful AI algorithms.
– Damon