Mind the Gap (and the GPU): The New Aristocracy of the Infinite Algorithm

Mind the Gap (and the GPU): The New Aristocracy of the Infinite Algorithm
Photo by Stefan Cosma / Unsplash

Gather 'round, children, and let me spin you a yarn. A tale not of dragons or valiant knights, but of algorithms and billionaires in regrettable hoodies. We're talking about a certain strain of "engineering prowess", a particularly virulent strain, mind you, that has, much like a ravenous beast let loose in a Victorian tea party, eaten its way through fair competition, leaving a trail of shattered dreams and rather stressed-out societies.

You see, for decades, we've admired the sheer audacity of human ingenuity. We marvelled at the beasts of the track: the Porsche 917/30, a leviathan of the Can-Am series that so thoroughly annihilated its rivals in the 70s, they practically packed up their toys and went home. The damn thing was so monstrously powerful, so utterly dominant, that the series itself, bless its brave but ultimately futile heart, simply withered and died. It was the engineering equivalent of turning up to a knife fight with a tactical nuclear weapon, a tad unsporting, wouldn't you say?

Then came the Williams FW14B in Formula 1, with its "active suspension," turning a mere car into a computer on wheels, gliding over bumps with an almost supernatural grace while lesser mortals bounced about like drunken sailors. Nigel Mansell, God rest his hairy-chested soul, won races by such vast margins that the FIA, in a rare fit of common sense, banned the lot. "Too good," they essentially said, "too clever by half, and frankly, rather boring for the spectators."

And let's not forget the Audi Quattro, a four-wheel-drive brute that stomped all over the two-wheel-drive rally cars of the early 80s like a particularly aggressive German boot. The message was clear: evolve or perish. It was an exciting, if terrifying, era, the Group B rallying monsters, cars so utterly unhinged that they frequently outstripped human reaction times, leading to a rather predictable and tragic end for the class.

These weren't just innovations; these were engineering nuclear bombs, detonating the very concept of a level playing field. And now, my dear readers, that same ethos has left the racetrack and taken up residence in our pockets, our homes, and, rather alarmingly, inside the very fabric of our collective sanity.

The Algorithm's Grasp: When Data Becomes the New Turbocharger

Cast your mind, if you dare, to the digital realm. Here, the "unlimited engineering" isn't about horsepower or downforce; it's about processing power, data accumulation, and the insidious, ever-optimising algorithm. Think of Google, for instance. A perfectly innocuous search engine, initially. But as it devoured more and more of the world's information, as every click and query fed its ravenous data maw, it created a "flywheel" of such gravitational pull that any would-be competitor was instantly crushed into digital dust. It’s the computational equivalent of the Porsche 917/30, so far ahead, so utterly dominant, that to even contemplate challenging it feels like a whimsical fool's errand. "Oh, you're building a new search engine? How quaint. Do you have 20 years of global data and enough server farms to rival a small country? No? Right then, pip pip."

Or consider the "social graph", a delightful term that sounds like something out of a particularly dull geometry lesson, but in reality, refers to the intricate web of connections curated by the likes of Meta. These platforms, with their billions of users, have engineered a network effect so pervasive, so utterly ingrained in our daily lives, that trying to launch a new social network is akin to trying to start a new ocean. "But our app has a lovely shade of blue!" chirps the plucky startup founder. Meanwhile, the Meta behemoth merely flexes its infrastructural muscles, reminding everyone that their "engineering" includes not just the pretty interface, but the global nervous system that underpins half the internet. These aren't just product advantages; they're engineering moats, dug deep and wide, filled with the tears of frustrated venture capitalists.

The Group B of the Mind: Our Collective Cognitive Breakdown

Now, here's where the dark humour starts to curdle a bit. Remember those Group B rally cars? Too fast for their own good, too powerful for human reflexes, leading to inevitable, catastrophic crashes. Well, my friends, we are now living through the Group B era of societal evolution. The turbo-charged development of AI and social media is outstripping our collective ability to adapt, to legislate, to even think properly about the consequences.

Our tech overlords, those visionary bro-CEOs in their designer sneakers, are building tools of unprecedented power, often with the giddy enthusiasm of a child given a loaded firearm. "Move fast and break things!" was once a quirky Silicon Valley mantra. Now, "things" include our democracies, our mental health, and the very concept of objective truth.

Social media algorithms, engineered with the precision of a Swiss watch (albeit one designed by a slightly sadistic genie), are constantly optimising for engagement, not for well-being. They've discovered our neural vulnerabilities, our dopamine triggers, and they exploit them with the efficiency of a finely tuned racing engine. The result? A collective cognitive overload, an epidemic of anxiety, and a global attention span that would make a goldfish look like a scholar. We are, in essence, driving a high-speed AI through a crowded marketplace, and rather than installing brakes, our esteemed engineers are simply adding more turbo. "More data! More engagement! Who cares if society is quietly fracturing like an old teacup? The numbers are up!"

The H100 GPU Gap: The New Aristocracy of Compute

If you thought the wealth gap was a bit of a pickle, wait until you get a load of the "compute gap." In the glory days of motorsport, the richest teams bought the best wind tunnels, the finest engineers, and the most exotic materials. In the age of AI, the currency of power is the H100 GPU, a piece of silicon so ludicrously expensive and so utterly essential for cutting-edge AI development that only a handful of global behemoths can afford them in meaningful quantities.

This isn't just about innovation; it's about consolidation. The "garage startup" myth of Silicon Valley, where two plucky lads could build the next big thing on a shoestring budget, is as dead as the dodo. Now, you need the financial firepower of a small nation to even train a competitive large language model. This "unlimited engineering" is creating a new aristocracy, an oligarchy of compute. We are ceding control of our future not to a diverse array of innovators, but to a handful of "Super Teams", Microsoft, Google, Meta, who alone possess the digital infrastructure to push the boundaries. What happens when our entire global economy, our understanding of reality, and our very jobs are dictated by the AI models trained exclusively by these few entities? One suspects the answer won't involve universal basic income and artisanal bread.

The Unending Race: Why This Pace is Detrimental to Humanity

Ultimately, the relentless, turbo-charged speed of current tech development is not merely disruptive; it’s genuinely detrimental. It’s a perpetual motion machine of innovation, divorced from the human capacity for absorption or ethical reflection. We are experiencing the societal equivalent of the Can-Am series collapsing under the weight of its own technological excess.

The pressure isn't just on the entrepreneurs scrambling to keep up; it's on us. The citizens, the consumers, the unwitting participants in this grand experiment. We are under constant pressure to adapt to ever-changing interfaces, to decipher increasingly sophisticated scams, to filter through an avalanche of algorithmically optimised misinformation. Our privacy is eroded with every update, our attention span fractured by every notification, and our mental well-being strained by the curated perfection of our digital neighbours.

There's no brake pedal in this race, no red flag from the FIA to say "Hold on, chaps, this is getting a bit much." The mantra remains: faster, bigger, more data, more compute. The "unlimited engineering" of tech isn't just creating better products; it's creating an inescapable digital environment that, much like those Group B rally cars, is becoming too powerful, too fast, and ultimately, too dangerous for the frail, analog beings attempting to navigate it.

So, as our tech overlords continue to build their magnificent, world-changing toys, perhaps we should ask ourselves: at what point does "unlimited engineering" stop being progress and start becoming a rather terrifying form of societal self-sabotage? I, for one, am rather looking forward to the inevitable system crash. Perhaps then we can all go outside and actually speak to each other, without an algorithm helpfully suggesting what we should be saying. A quaint thought, I know. Now, if you'll excuse me, I hear my phone vibrating with an urgent notification about a revolutionary new app that will completely transform my life. Again.

In the spirit of our "unlimited" masters, please consider donating to ensure this publication remains stubbornly analog and blissfully unoptimized by a billionaire's whim. Thank you!

Donate

References:

Besiroglu, T. et al. (2024) The Compute Divide in Machine Learning: A Threat to Academic Contribution and Scrutiny?. arXiv:2401.02452. Available at: https://arxiv.org/abs/2401.02452 (Accessed: 19 January 2026).

Azoulay, P., and Jones, J. L. (2024). Old moats for new models: Openness, control, and competition in generative AI. (NBER Working Paper No. 32474). National Bureau of Economic Research. Available at: https://www.nber.org/system/files/working_papers/w32474/w32474.pdf (Accessed: 19 January 2026).

Khanal, S., Zhang, H., and Taeihagh, A. (2024). Why and how is the power of Big Tech increasing in the policy process? The case of generative AI. Policy and Society, 44(1), 52-69. Available at: https://doi.org/10.1093/polsoc/puae012 (Accessed: 19 January 2026).

Sastry, G. et al. (2024). Computing power and the governance of artificial intelligence. arXiv:2402.08797. Available at: https://doi.org/10.48550/arxiv.2402.08797 (Accessed: 19 January 2026).

Read more