We may receive a commission on purchases made from links.
Even Amazon is getting in on the agentic AI game, with a recent announcement highlighting how the company plans to deploy agents built just for Amazon Web Services (AWS). Amazon outlined all of its new AI tools during a conference in December, highlighting not only its “frontier agents” — which the company touts as being able to take on more complex tasks for longer periods — but also its Nova Forge service, which allows other companies to train their own privatized instances of the Amazon Nova model, using their own data to create said frontier agents.
The new announcements came as part of the company’s yearly re:Invent conference, and AWS Chief Executive Matt Garman said the new products help highlight Amazon’s ongoing commitment to artificial intelligence, something that has reportedly come into question due to Amazon’s slower response, despite the rate of AI adoption throughout the enterprise industry.
Despite being criticized for lagging behind a bit, these new changes could be the boost Amazon’s AI offerings need to help it catch up to competitors. None of this strictly applies to consumers, though, as the AWS agents are more geared toward developers and companies alike. However, improved AWS AI integrations could help determine what your online experiences look like going forward, depending on how companies put them to use.
Of course, none of these “advancements” come freely, and while Amazon appears to be making big strides, the company’s push will carry a much larger, longer-term cost in the requirements attached its new datacenters.
New data centers could require more power and water
The idea of using AI to simplify how businesses do things sounds like a good idea — especially if it helps make things easier on the customer, too. However, there’s two major issues with expanding AI options like the new Nova Forge and these AWS agents: power and water usage.
There’s already been much talk around just how much electricity AI platforms like ChatGPT use, and we’ve even heard reports that power bills might increase thanks to how much energy AI is sucking up. All of this has led to many tech companies, like Google, investing in nuclear power efforts, though it could be some time before the fruit of those labors becomes evident.
If you’re worried about how much energy data centers require to power existing AI tools, then you’ll likely be concerned to hear that the company is expanding its AI offerings. While Amazon suggests its datacenters use less water than the clothing industry, and that its data centers don’t drive up electricity bills, there’s evidence tied to suggestions that AI will make your power bill go up.
A good investment, but for who?
Amazon has been very open about where it plans to invest as it grows its AI data center coverage, and one particular investment will see the company putting an additional $15 billion toward building data centers in Northern Indiana, where the project is expected to add 2.4 gigawatts of data center capacity in the region. This is on top of the previous $31.3 billion that the company has invested in that state alone.
Alongside this, though, the company is also working with NIPSCO to create what it calls a “first-of its kind framework specifically developed with local residents and small business in mind.” Through this effort, Amazon will pay fees to use any existing power lines, while also covering the upfront costs to build new power plants, erect new power lines, and purchase the equipment needed to deliver power to data centers.
That project is set up to add “up to 3 gigawatts” of energy capacity to the already existing grid, though with Amazon using at least 2.4 gigawatts for its data center, it’s unclear how the remainder will be distributed. Ultimately, this is good news for those that have been waiting for Amazon to show more of its might in the AI race.
For critics and skeptics, though, it’s a warning: As more AI datacenters are built, it’s likely we’ll see expanding energy and water usage as big companies like Amazon tap into existing grids to keep things running smoothly.
