If you believe Meta, its forthcoming Albertan data centre will be, at worst, unnoticeable to people across this province:

We pay the full costs of our data centers’ energy use so consumers aren’t negatively impacted, and fund new and upgraded infrastructure. We worked closely with Greenlight Limited Partnership, Altalink, Capitol Power, and the Alberta Electric System Operator to plan for and meet our energy needs years in advance of this data center coming online.

Sure sounds like the commitments of a stand-up corporate citizen. However, if you listen to the power companies, they tell a somewhat different story.

Rory White, the National Observer:

In its latest earnings call, [Capital Power’s] CEO Avik Dey predicted a “return to higher pricing” telling analysts that he couldn’t rule out prices of “$80 or $90” per megawatt hour of electricity by early 2028. By 2029, rival company TransAlta is predicting an average of $100 — more than triple the average price this year — a high not seen since the province’s 2021-2023 energy crisis.

Driving this increase is the province’s reliance on natural gas power, which cannot be built fast enough to accommodate data centre demand, according to Will Noel, a senior electricity analyst at the Pembina Institute. “In Canada and internationally, we’re seeing this huge crunch in gas turbine supply shortages,” he said. The province has also backed itself into a corner in terms of its options for alternatives. “Alberta has over the past couple of years really stifled the growth in its wind and solar.”

White reports this would increase the electricity costs to a typical household by “hundreds of dollars more per year”. However, as White caveats, these prices have precedent as recently as three years ago — and that undersells it. In 2022, electricity costs were as much as five times above current rates. Even without adjusting for inflation, we are currently paying less than we did from 2018–2020, and a tripling of the current rate would have a nominal cost similar to that of 2008. That is not to say this is good, but I think it is worthwhile seeing the “tripling” figure within the context of recent fluctuations.

⌥ Permalink

Karl Bode:

One lie that companies have been telling local municipalities is that if they greenlight a massive local AI data center, it will immediately bring a flood of savvy innovators to your podunk-ass town.

The promotional materials for Kevin O’Leary’s still hypothetical “data centre park” imagine a futuristic campus full of bright young minds doing complicated A.I. stuff on-site in north-central Alberta. But why would they be there — fifty kilometres from the nearest city and 500 kilometres from Edmonton, the nearest major city? Why would they not be in Vancouver, or Silicon Valley, or anywhere else with an internet connection?

Also:

These companies aren’t coincidentally aiming construction at states and municipalities that are too broken and corrupted to put up meaningful regulatory opposition. […]

That is one reason why we are seeing a bunch of these proposals here. And these companies benefit from a lack of transparency.

⌥ Permalink

Nathaniel Rich, author of the novel “Cloudthief”, in a non-fiction retelling for New York Times Magazine of a 2007 heist of a London data centre by Terry Ellis and others:

The fixer — Ellis called him Ray and won’t reveal his name — met him in North London near Hampstead Heath for coffee and cakes. When it came time to discuss business, to avoid being overheard, they strolled into the park.

Ray had brought Ellis a few jobs before. But this job, he warned, was of an entirely different order. As Ellis claims in “The Art of Robbery,” a self-published memoir written after his release from prison, he eventually learned that Ray had been contacted by a consultant employed by “some influential bankers from America.” The bankers “were involved in prime mortgages” and had “circumnavigated” certain regulations. Damning evidence of these circumnavigations could be found in banking files held in the King’s Cross area in a giant building known as a data center.

This is a dramatic story, and one I think should be read with a heavy dose of skepticism. It seems that most of the criminal details have been shared by Ellis. For a start, the claim that some bankers ostensibly contracted with “Ray” is just a little too perfect for a recession-era tale. These bankers are pretty much universally loathed, and this justification makes this theft seem more palatable than a simple financial motive. For example, there was a similar data centre theft in October 2006, which would be unrelated to the lending crisis in the following years.

Another problem is that Rich says crimes like these are covered-up in part by a data centre operator because they are loathe to “admit to flaws in its security, [which] would only encourage additional attacks and scare away its clients”. Therefore, the lack of evidence for the specific circumstances of this crime is supposed to be a buttress for its likelihood, not a weakness, which is not reassuring.

The story of the theft was, as far as I can tell, broken by Here is the City, then a gossipy financial news site:

The data center itself is thought to be used by a number of companies, including JPMorgan, which is believed to have told staff that some of its systems could be off-line for parts of the day today as a result of the theft. Fortunately the thieves are thought to have got away with just the computer hardware, and not any sensitive information which may also have been stored at the facility.

Tom Espiner, of ZDNet, a few days later:

Reports circulating on the Internet last week that JPMorgan, a customer of Verizon Business, had been affected by the burglary were incorrect, according to a source at the investment bank. There has been no loss of service or data, said the source.

On the one hand, of course all these parties tried to cover this up. The reading-between-the-lines story implied by these early reports and Rich’s telling is that some banking higher-ups, perhaps from JPMorgan, wanted to cover up some crimes, and denying any meaningful effect is just more cover-up. But little of this is substantiated by contemporary or current reporting — which is, of course, the whole problem with using a lack of evidence as the foundation for a story.

Rich, in the Times:

“The banks knew they were sending mortgages to people who couldn’t pay back,” he says today. “That’s what broke the whole system. That was the big con.” Ellis remains convinced that the bankers who paid for the Verizon job wanted to destroy evidence of their involvement in fraudulent subprime mortgages — the inside information that Ellis received about the data center, he believes, “would have had to come from the top” — but he can’t prove it. He never saw what was on the servers.

“Our job was to get the motherboards,” he says. “We were paid quite handsomely. Whatever happened after that was none of our concern.”

In contemporaneous reports, the Metropolitan Police noted the theft of motherboards and processors. But if these bankers wanted to cover up their fraudulent practices, surely the hard drives would have been the target, right? In Rich’s version, entire servers were taken, so perhaps this is just a misunderstanding.

This story smells fishy. I believe the theft happened, of course, and Ellis’ involvement, but I am not as convinced this had anything to do with covering up some white collar crime. (By the way, the Guardian in 2018 published an interview with Ellis about the interesting prison where he was transferred and which led to his rehabilitation.)

The heist element is only about half of Rich’s story; much of it is a discussion about data centre secrecy:

The public fogginess about data centers is not an accident. It is the product of a willful strategy by the world’s largest tech corporations, whose business models rest on the public assumption that the internet, and all the data it holds, is as immaterial as air — or as a cloud, to borrow the metaphor commonly used to describe the sum of information stored on servers. As the digital-media scholar Tung-Hui Hu writes in “A Prehistory of the Cloud,” the cloud “hides its physical location by design.”

[…]

It was a lot easier to defend data when people didn’t know it existed. The more people learn about data centers, the more they hate them. […]

If you read a website like this one, you were probably aware that data centres were commonplace twenty or more years ago. Like the one near King’s Cross, some were hidden in plain sight, while others were purpose-built facilities that look like hangars stuffed with servers. But the A.I. boom has meant rapid increases in the speed, scale, and quantity of data centres. People quickly learned not only of their existence, but how much pressure they put on local resources. Tech companies, it seemed, were caught by surprise; and as someone who spends a lot of time immersed in this world, so was I.

Much of the consternation I have seen in more general audiences has been about data centres in general. People simply were not aware that Amazon has warehouses full of products, and other warehouses full of computers. As Rich writes, this is deliberate, for business secrecy reasons, security, and environmental costs. But, also, I think some of that unawareness is because of just how boring it is. If nobody wants to know hidden information, is it really a secret? It only became one when the information these companies were hiding had real-life effects.

It does seem that public awareness is putting pressure on corporations to improve data centres and make them more efficient. But that is not a standard. New data centres are powered by petroleum with a pinky promise of renewable offsets. In some regressive regions, like Alberta, new power plants for data centres must be powered by methane gas. In a further complication, Meta’s proposed data centre is scheduled to be completed before the power plant is ready, meaning it will be dependent on existing grid power for perhaps years. Meta’s is just one of the data centres proposed for Alberta. Another one, a gigawatt cluster, would also require a dedicated gas-fired power plant, while Kevin O’Leary’s questionable project is supposed to require over three times the combined power of those other two.

For years, the tech industry told us we did not need to have much concern for how digital products and services worked, and many of us did not bother to find out. But it turns out the demands of our email and Netflix subscription were comparatively easy to hide. At the very least, what we ought to demand from projects with the scale and ambition of these data centres is open disclosure of their power consumption, water use, and emissions.

But we ought to demand more than the bare minimum. Transparency does as much good as a big banner reading we are destroying the planet but we are also creating a lot of value for shareholders. When a single data centre is projected to use about as much power as the entire city of Calgary is currently — Enmax says 1,260 megawatts as of writing — we should have a say in whether that makes sense. A.I. remains a thing that is happening to us rather than with or for us. It is built on assuming consent and asking forgiveness, which has more-or-less worked for the industry and gave it way too much confidence. Tech companies could have spent decades being better corporate citizens. Data centres are just one part, but they are representative of the difference between the stories told by tech companies and the things we can actually know.