Synthetic Intelligence dominated headlines final yr and continues to take action this yr, albeit for diverse causes. Final yr’s focus was on its capabilities, as organizations and people started to grasp how AI may improve their private {and professional} lives.
Amid this pleasure, doomsday predictions about AI’s potential to obliterate humanity additionally emerged (thanks, James Cameron and Cyberdyne Methods).
There have been amusing but absurd developments, such because the viral phenomenon of turning components of the US into motion figures. Whereas a few of these developments had been entertaining, most weren’t. The underside line is that we’re witnessing unprecedented progress, utilization, and useful resource consumption similar to the Industrial Revolution — maybe we must always name it the Intelligence Revolution.
This yr, we already see the impression of hundreds of thousands of execs harnessing AI’s energy. AI shouldn’t be magic; it depends on tangible expertise and huge sources, together with servers, electrical energy, water, reminiscence, chips, and information facilities. When 50 million individuals faucet into huge computing energy to rework themselves into motion figures, sources turn into strained. That is our actuality.
As sources turn into scarce, costs inevitably rise. I want I had paid extra consideration in Economics class. In my time at UF, Eco101 was out there on VHS, making the college an early adopter of digital studying, as I used to be largely digital in 1992. Regardless, we’ll dive into the important sources, beginning with water. A large information heart can eat as much as 5 million gallons a day.
In response to the Lincoln Institute of Land Coverage, Texas information facilities consumed roughly 50 billion gallons of water final yr. At a smaller scale, a 20-40-query ChatGPT dialog consumes the equal of a 16-ounce bottle of water. The dimensions of water utilization, each small and huge, is staggering.
Subsequent, let’s talk about electrical energy.
Knowledge facilities in the US are estimated to eat 5-10% of the nation’s energy. Within the spirit of an economics lesson, let’s make clear what information facilities are: massive services housing rows of servers, information storage programs, and networking tools, together with energy and cooling programs important for his or her operation.
This infrastructure underpins firms offering digital companies. Each time you ship an e-mail, stream a present, save a photograph to “the cloud,” or interact with a chatbot, you’re interacting with a knowledge heart.
I anticipated Florida to be a number one supplier of information facilities (corresponding to for elections, soccer, and holidays), however it seems Texas (300+), Virginia (600+), and California (300+) are the highest states for information facilities within the U.S. The overall variety of information facilities within the U.S. is roughly 5,000, accounting for 40% of the worldwide market, whereas Florida has roughly 100 operational facilities (Knowledge Middle Map).
Observe that many firms declare to have a “information heart,” however merely having a server in a closet doesn’t qualify. Nevertheless, a big information heart could quickly be coming to Florida, doubtlessly boosting our rating (Developer Plans Knowledge Middle).
Transferring on to reminiscence and chips (to not be confused with the Chips TV Intro).
A reminiscence chip is an built-in circuit that accommodates hundreds of thousands of tiny transistors and capacitors and is used to retailer, retrieve, and handle information in digital units. These chips could be both RAM (short-term) or Flash/ROM (long-term) reminiscence, and they’re important for working software program, storing recordsdata, and enabling expertise.
Knowledge facilities require substantial reminiscence, significantly high-bandwidth reminiscence (HBM) for AI chips (Nvidia, AMD, Google), which is at present straining world provide. Costs for pc reminiscence are anticipated to rise by over 50% in early 2026, and we’re already witnessing this enhance.
HBM chips are way more complicated than the RAM utilized in shopper laptops and smartphones. Designed for high-bandwidth necessities, HBM is fabricated by a fancy course of that stacks 12 to 16 reminiscence layers on a single chip, forming a “dice.”
When an organization like Micron manufactures one little bit of HBM reminiscence, it should forgo producing extra typical reminiscence for different units.
As you store for brand spanking new PCs, be ready for value will increase of round 25%. In my 20+ years in expertise, I’ve solely seen a spike like this as soon as, as a result of manufacturing decreases throughout the pandemic. This case is totally different, and its results will probably be felt all year long (PCWorld).
Now let’s tackle uncommon earth metals, which appeared like an issue for 2024, however it hasn’t gone away; it’s simply turn into extra manageable. Uncommon earth components (REEs) are a bunch of 17 metallic components essential for high-tech functions, together with electrical autos, wind generators, and electronics. The main subject is that China accounts for greater than half of the world provide.
In the US, firms are in search of alternate options. As an illustration, Mosaic in Florida is exploring waste mining to reclaim REEs from leftover supplies from phosphate mining. They’re additionally investigating various makes use of, together with road-building supplies. Though this course of seems pricey, we hope it’s going to yield outcomes as we discover choices past importing.
In Tallahassee, I mentioned AI with Eduardo Gonzalez Loumiet, a accomplice at Ruvos, a world software program firm working information facilities worldwide. He acknowledged, “The query isn’t whether or not synthetic intelligence is coming; it’s already right here, and its use will solely proceed to speed up. As a area anchored by schooling, authorities, and an more and more progressive personal sector, we should be trustworthy about what that actuality calls for. Widespread AI adoption requires actual funding in infrastructure — energy, connectivity, information capability, and resilience — so our establishments can use these instruments responsibly and at scale.
“We already see this firsthand. At Ruvos, synthetic intelligence is being utilized in well being care not solely throughout Florida but additionally globally, supporting public well being, medical programs, and decision-making at unprecedented ranges. If Tallahassee desires to stay aggressive and efficient as AI turns into foundational to how work will get finished, we should make investments now within the infrastructure that helps each innovation and long-term sustainability.”
Thanks, Eddie, on your insights and your service as Chair of the Tallahassee Chamber of Commerce.
AI could result in short-term computational and enterprise challenges, however it’s not the apocalyptic state of affairs that John and Sarah Connor warned us about. Whereas they predicted AI may destroy the world, what we’re experiencing is a drain on our sources. We should always contemplate slowing down the usage of ChatGPT for motion determine creations and resist asking Co-Pilot to compile all of your 2025 emails right into a abstract, as these actions contribute to a domino impact. This state of affairs is definitely relevant to the digital mini-apocalypse we’re experiencing as we speak. A correction is feasible, and steadiness could be restored.
We — all of us — should work in direction of this, remembering Sarah Connor’s phrases:
“There isn’t a destiny, however what we make.”
We’d higher get to work to make sure we harness our human intelligence successfully on this digital intelligence revolution, leaving the motion figures to Hasbro.
