While the good versus evil debate continues about artificial intelligence, one of its negative aspects is nearing a solution. That problem is the environmental impact of massive data centers that support AI. Traditional domestic power sources designed for our cities, are being tested when an AI data center moves in. The massive power requirements of these centers can overwhelm existing systems not to mention the impact on the local environement.
Nuclear power plants might be a consideration but they are massive systems that take a decade or more to build and they rely heavily on huge amounts of water for their cooling needs. These nuclear power systems have been in use since 1957 in the U.S.
A newly designed microreactor nuclear power system has been successfully tested, and it could provide the needed power without the drawbacks of traditional nuclear systems. The new designs are much safer, use Low-Enriched Uranium, are easy to build in a factory, can fit on a truck, and use near-zero water. These systems could be ready in the 2027-2030 timeframe.
While a single TRISCO-fueled microreactor is not large enough at 20-50 megawatts to power a large AI datacenter, they can be clustered to scale to the gigawatt needs of these power-hungry installations. This seems to be another area where AI may be replacing many menial jobs while creating other opportunities.
No comments:
Post a Comment