From Rest of World:

The UAE’s new 1-gigawatt Stargate AI campus, backed by OpenAI and Abu Dhabi’s G42, is only the beginning of a massive infrastructure build-out across the region. Saudi Arabia has separately announced plans for 2,200 megawatts of new data center capacity, attracting tech giants including Google, Microsoft, and Amazon Web Services to establish regional hubs.

While these investments promise to transform the Gulf into a digital gateway linking Europe, Asia, and Africa, they come with an environmental cost that could undermine the very growth they seek to enable. The water demands of AI-intensive data centers may force governments to choose between their climate commitments and their technological ambitions.

I guess I understand the diplomatic phrasing, my dude, but there has been no forcing involved in this set of decisions, no forcing whatsoever. Those “climate commitments” were always already thinner than the paper that bore the signatures.

Weary cynicism aside, I felt the urge to quote this piece due to a bizarre sense of echo from my childhood. VCTB veterans may know that I lived in Saudi Arabia for a couple of years as a kid in the mid-Eighties. My old man worked in computers back then, specifically as a manager of what was then called a “troubleshooting” team for Amdahl, which was at the time one of IBM’s few viable rivals in the mainframe business.

As I understand it, the Saudis needed mainframes mostly for the oil business, but I suppose perhaps for other purposes as well. At the time, a computer capable of doing serious long-run processing tasks—like the payroll run for a big firm, for example—was a bunch of filing-cabinet-sized boxes in a medium-sized office room, linked together by cables that I vaguely recall as being as thick as my six-year-old wrist, a tangled nest of dusty black snakes hidden beneath the square tiles the covered what was known as the “floor void”.

Again, as I understand it, while the Saudis were not exactly short on money, they couldn’t afford (and probably didn’t need) the absolute cutting edge of mainframe technology at the time, which I suppose would have been a Cray or a top-notch IBM set-up. Instead, they were buying systems a few generations behind the leading edge. And this, I presume, is why Amdahl was out there in the Gulf, hanging on at the trailing edge of the business: Big Blue was focussed on the big-ticket players, and a fading rival could still pick up some gigs on the fringes of its interests.

IBM was definitely out there, though. I recall my father and his colleagues discussing that firm (for which both he and my mother had worked, during an earlier phase of computing as an industry) in much the same way I might have discussed a particularly domineering classmate with my peers at the international school I was bussed to each morning: a mixture of awe and contempt, layered over a core of fear and resentment of the unarguable boss of the playground.

Setting aside questions of cost, the major limit on computing power at that time was actually waste heat, as it still is today. Size was to some extent a function of this problem: much of what was in those mainframe cabinets was open space between the circuit-boards, and the computer room couldn’t be too tightly populated with cabinets because you needed to let the ambient heat dissipate somehow. This, I presume, must have been particularly tricky in a country where the temperatures were routinely above 30°C for over half of the year.

(As a rural-raised British kid, Saudi was my first encounter with air conditioning, which I never much liked. I have also long had a theory that my predilection for hot, dry weather was somehow programmed into my skinny little body during this time; this theory is clearly unsupportable by any scientific theory of which I am aware, but I nonetheless believe it on such a basic level that the science doesn’t matter at all.)

Like any challenger firm, Amdahl had its own differentiator, though I have no idea whether it really made much of a difference over their offering cheaper, older systems; perhaps a scholar of computing history could enlighten me. The thing was, back in those days a powerful IBM mainframe was connected not only to power and data cables down in the floor void, but to the building’s plumbing, also; just as it is with a modern data center, those systems relied on flowing water as a heat-sink.

And this was the Amdahl difference, as I recall it—and I recall it only because of a marketing tag-line which may not even have been official: I remember it as printed on a T-shirt worn by one of my old man’s troubleshooters, who would sometimes come to eat barbecue and drink contraband bootleg booze at our compound.

“Save water: buy an air-cooled computer.”

We left Saudi in, I guess, 1984? 85? (I really need to sit down and timeline my own childhood; keeping track of all the different schools and homes is getting harder with distance.) Again, the state of cutting-edge computing in that country is left to the expertise of someone who was older than six or seven at the time. I think my old man was still with Amdahl until the 1990 crash, though I may be misremembering that, too; he was still in computers, that’s for sure, and skeptical of the promises of desktop PCs, even as he made sure to always have the cutting edge of such in his office.

(Unusually for someone of my generation, my first computer was not a Commodore or a Spectrum, but rather the vast, rattling box-tower that housed my father’s hand-me-down 8086. He fitted it out with a hard-disk drive before passing it on to me. “Twenty megabytes!” I remember thinking. “I’ll never fill that up!”)

The last decade of the 20th century spelled the end of mainframes, and of my father’s career, and much else besides. It’s a commonplace to talk about how much has changed, whether socially or technologically. Perhaps it’s my own middle-agedness—I’m around the same age now that my father was when the industry that raised him out of working-class anonymity in a fading navy town decided to toss him aside as surplus to requirements—but it increasingly feels to me that we’ve actually been in something more like stasis since some hard-to-define date in the mid-Eighties. Yes, of course, there’s the Number Go Up and the Bigger Better Faster More… but that’s kinda my point.

We may well have computers in our pockets that could, at a guess, out-process the entire installed capacity of the Gulf states four decades previous, but we are still, figuratively speaking, still running the same damned programs.