AI and the Data Center
In 2020, we met with a tiny Canadian tech company that had a compelling story
They used a combination of artificial intelligence (A.I.) and smart monitors to save companies money on their electric bills. A lot of money.
When we saw them, they had two major clients – Starbucks and Bank of America. The company used smart thermostats to control the temperature of the offices. The savings were shocking.
Sadly, the company died of managerial stupidity. But the idea was sound. Letting an A.I. manage utilities to save money is a great idea. And other companies caught on.
Alaska Airlines used A.I. to optimize about 25% of its flights in 2023. By shaving off a few minutes here and there, it cut over 680 hours of flight time. That saved more than half a million gallons of fuel. At an average price of $6.50 per gallon, that’s a few million dollars’ worth of savings…on just 25% of its flights.
This is a trend poised to explode. The adoption of A.I. into the workplace will grow exponentially over the next couple of years. Especially as companies see how much money they can save. However, it comes with a cost.
A.I. and Energy – A Mixed Blessing
The landscape of energy is going through a huge transformation. Historically, electricity came from a huge central power plant. Today, with the proliferation of wind farms and solar farms, there are many more power plants. That complicates the balance of energy because the supply must be balanced with demand.
And here’s the kicker, as we discussed last week. According to analysts at Morningstar, Global power consumption from data centers was roughly 460 Terawatt hours (TWh) in 2022 and could double by 2026 to more than 1,000 TWh. That’s equal to the power demands of Japan.
That’s an astonishing growth rate of 16.8% per year.
And a lot of that will happen here in the U.S. According to S&P Global Market Intelligence, Northern Virginia was number one for the 2023 total power capacity of the top ten largest data center markets globally. Beijing was the second largest market with a little below 2,000 MW of capacity. That means U.S. electrical utilities will see a massive growth in power demand.
As we discussed here previously, that will translate into copper demand as well. Trafigura believes A.I. will add an additional one million metric tons of demand by 2030. This is in addition to all the other demand drivers. And it comes at a time when supply cannot keep up.
We can already see the impact in demand for copper. Supply can’t keep up and copper prices are now at all-time highs:
This is in part due to the demands from the data centers used to support A.I.
This is a trend we’re following at Mangrove Investor. There will be massive investment opportunity in the peripherals around A.I. The downstream impacts of metal demand, electrical demand, etc. will offer fantastic opportunities for investors.
For the Good,
The Mangrove Investor Team
Numbers You Need to Know
2,552
4.3 trillion cubic meters
293,000 tonnes
Copper consumption for data centers is estimated to jump to 238,000 tonnes in 2030. (Elements)
What’s New in Sustainable Investing
Labelling rules could boost sustainable investing interest
New labelling rules will help combat greenwashing and increase interest in sustainable investing, a finance expert has predicted. (FT Advisor)
Best Investment Banks and Sustainable Finance Awards Ceremony 2024
The awards ceremony brought together attendees from across the globe at the iconic Gherkin on St. Mary’s Axe in the heart of the City of London. (Global Finance)
Video Of The Week
Virginia Is the World’s Data Center Hub – at What Cost?
Virginia may be for lovers but it is also for Data Centers. Virginia is home to hundreds of data centers which support a majority of the world’s internet traffic