About AI and Data Centers

Alaskans are rightfully alarmed about the sudden push for artificial intelligence (AI) and the data centers that drive it.

AI data centers hurriedly constructed elsewhere have a terrible track record of air, water, and noise pollution, consuming vast quantities of power and water at the expense of the local population and provide few jobs beyond their brief construction period.

Our state does not have sufficient electrical generation capacity or internet infrastructure to support near-term construction or operation of hyperscale data centers. Longer range concepts for such centers are either opaque to the public, half-baked, or both. I cite the coal-powered Terra project proposed for West Su as an example.

I also call out AIDEA, which without consulting local communities or Mat-Su Borough at all, applied to DNR for a free conveyance of nearly 20,000 acres of prime recreational, salmon spawning and hunting grounds between the City of Houston, Meadow Lakes, and Willow. AIDEA representatives say that no applications for data centers have been entertained at this early juncture. However, AIDEA’s proposal leaves the land open to ANY kind of industry. Planned or not, data centers were cited four times.

Across the political spectrum, in council meeting after council meeting Alaskans are loudly saying they do NOT want AI data centers in or near their communities.

Like any emerging industry, it will be up to the data center companies to prove to us they will be good neighbors to our communities and to operate within the boundaries we require of them, such as acceptable energy use plans and being held to high standards for the protection of the air, water, soil, wildlife, and people around them.

I advocate for a minimum 18-month Alaska-wide moratorium on new data centers designed or intended for AI systems or capable of consuming more than 50 megawatts. Until we fully understand what is coming at us and have legislated accordingly for our own enduring interests, I’ll oppose large data centers in our state.

A Deeper Dive into Artificial Intelligence and Hyperscale Data Centers

There is no standard definition of the term "data center." In various forms they have existed for decades to enable the technologies we use every day like cell phones, cloud storage, and processing your latest online order.

However, the data centers that are creating public alarm are different. They are designed to support AI, which is a rapidly advancing type of software that is the brain of machine learning. To operate that software at commercial scale, companies like OpenAI, Anthropic, and Meta require large, purpose-built data centers, generally able to consume 100 megawatts or more of electricity. These are referred to as "hyperscale" data centers.

The new data centers of this kind, and their hunger for ever more advanced silicon, are why it took less than four years for GPU chip maker NVIDIA to grow from a market capitalization of $300B to become the largest company in the world, currently valued at about $5.25 Trillion.

The adverse effects on communities where hyperscale data centers have been built are becoming well-known: pollution, noise, heat, high volume water use, massive electrical draw, and few long-term local jobs after construction.

Yet an even deeper concern is the rapid advancement of the AI software itself. Technologists widely believe that AI is nearing the point where it can teach and improve itself beyond human capability and control, a dire situation termed "technological singularity." If AI is left unregulated, this will almost certainly happen in the next few years. Economic and social upheaval far faster and more impactful than the industrial revolution would inevitably follow. The thousands of jobs already lost in the tech sector are only a small foreshadowing of what a future with unrestrained, unregulated AI in the hands of a few oligarchs will look like.

This is not just theoretical sci-fi stuff. OpenAI recently made news by losing containment of supposedly sandboxed AI agents that then hacked into other companies. Further, the human willingness to abuse this powerful technology is ever-present. The People’s Republic of China is far down the path of the surveillance state envisioned in Orwell’s prescient novel “1984.” Our own nation’s capacity to follow suit already exists. Flock cameras tied to AI data centers – just as one example – are popping up by the thousands. In the reckless bombing of an Iranian girl's school, both human error and careless use of Anthropic’s Claude AI model were publicly implicated; which factor is more to blame may never be answered. 

The world’s deepest pockets with largely unstated motives are attempting to strong-arm data centers into communities across our nation with rushed, opaque land swaps, NDAs, vague promises, and minimum possible public notice. We can make guesses about what they are up to. But in truth we, the public, don’t know, except that without question it is oligarchic behavior incompatible with our representative democracy. 

Regulation through law is how we stop overreach and prevent harms. The duopoly of vast wealth and purchased political power has been trying to turn "regulation" into a dirty word for a long time. You know which side of the road to drive on and have safety glass in your windshield because we regulated the horseless carriage as it advanced. What rational person would say nuclear energy needs no regulation? We even regulate handheld laser pointers because they can dazzle or blind people, most notably pilots in flight.

Because the AI genie is out of a bottle that can never be re-corked, we collectively must both speak out and act. At every level of government from local ordinance to international treaties, it is imperative to our way of life that we limit and control this increasingly powerful technology.