Are We Nearing Peak AI?

#writingcommunity #booksky #amwriting #writing Unfetterred Treacle

I’m not too worried about the whole “technological singularity” thing, you know, the one where AI gets so smart it takes over everything and changes the world forever. That might be something to think about someday far off, but it’s not what’s been on my mind lately.

What I’ve been wondering is this: what if we’re actually closer to the high-water mark of AI than we think?

Not because the tech is running out of steam, but because the world it depends on is starting to show cracks.

We keep hearing about AI dominating the future, spreading everywhere, transforming everything. But what if those predictions are a little overblown? What if the real limit on AI isn’t how good our technological capability is, but whether the world can keep the lights on?

The Cracks in the Foundation

AI doesn’t exist in a vacuum. It runs on semiconductors, rare earths, high-bandwidth data networks, satellites, and server farms big enough to make a coal plant blush. And all of that? It depends on a global system that’s feeling more fragile by the day.

Between trade wars, tariffs, and the general messiness of international politics, we’re looking at some serious pressure on the systems that keep the tech world spinning. The current tariff landscape alone is enough to give economists indigestion, and there’s talk we might hit a global recession before the year’s out. That kind of squeeze makes it harder to keep building and powering the AI infrastructure we’ve come to expect.

The End of Calm Seas?

Since World War II, the U.S. Navy took on the job of keeping the sea lanes open. That was part of the Bretton Woods system, make trade safe, keep the world humming. It worked. For decades, American warships helped maintain peace on the water, and that meant reliable shipping routes for everything from coffee beans to cobalt.

But times are changing. The Navy’s aging. Budgets are tight. Priorities are shifting. And as Peter Zeihan has pointed out more than once, we might be looking at the slow sunset of the Pax Americana—the peaceful(ish) global order that made the modern global economy possible.

No Navy watching the seas? That means more piracy, more regional conflict, and more uncertainty in global shipping. And guess what needs stable shipping? Chips. Data centers. The entire AI ecosystem.

AI Is a Global Puzzle

Most folks think of AI as software, but the truth is, it’s hardware-heavy. You need silicon wafers from Taiwan, cobalt from the Congo, copper wiring, rare earth minerals, and massive data centers connected by undersea fiber. And to build just one advanced chip, you’re looking at supply chains that involve thousands of companies across dozens of countries.

Did you know? The semiconductor supply chain touches over 9,000 companies around the globe. That’s a lot of moving parts—and a lot of chances for something to go wrong.

We’ve already seen some warning signs: chip shortages, trade restrictions, nations starting to hoard key resources. And the tech world? It’s starting to split into rival camps—each with its own standards, rules, and systems.

If that trend continues, scaling up AI the way we have might become less feasible. Not impossible, but a lot harder, and a lot more expensive.

Are We Already Leveling Off?

Here’s the other thing: even the tech itself might be slowing down a bit. Some of the latest research is showing diminishing returns when it comes to training ever-bigger language models. Larger computers doesn’t always equal better results. You still need good data, smart design, and some human common sense.

And when you factor in rising hardware costs and geopolitical roadblocks, you start to wonder if we’re approaching a natural plateau.

Zeihan and others think we’re entering an age of deglobalization, where countries focus more on self-reliance, less on cross-border collaboration. That’s not exactly the best setup for building world-spanning AI systems, not to mention the ability to create the top end chips that AI relies on.

So, What Happens Next?

Don’t get me wrong, AI isn’t going away. It’s still going to shape the future. But we might see a shift from “bigger and faster” to “practical and more efficient.” Think less moonshot, more toolbox.

Instead of big, global AI, we might get lots of regional ones, each with its own quirks and rules. Some open source. Some locked down. Some focused on innovation, others built for control.

We don’t need AI to be godlike. We just need it to be useful, and maybe a little less dependent on a world order that’s looking shakier by the minute.

So yeah, I’m not worried about the rise of the machines. I’m more interested in whether the global conveyor belt that feeds them will keep running.

What do you think? Are we nearing the top of the AI curve, or just catching our breath?