What AI Won't Do, Part II

Wednesday, June 17th, 2026

Bill Bonner, from Youghal, Ireland

---

Reuters:

Bank of Japan raises interest rates to 31-year high

The Bank of Japan raised interest rates to a 31-year high on Tuesday, marking another landmark step in normalizing monetary policy as it focused on taming price pressures from the energy shock caused by the Iran war.

Now here is a thing worth turning over. It puts two dots at wide angles...and pulling away from each other. Artificial Intelligence — now coming under the scrutiny of political quacks who couldn’t tell a transistor from a turnip — is supposedly destined to make us all “very rich.” So, the price of money ought to be sinking, not climbing. The logic is plain; the IPOs of the hyperscalers alone are forecast to shower an additional $4 trillion upon the moneyed classes. They will have a lot more money to lend.

Yet the rates rise. All over the world, central banks are raising rates, not lowering them. And with the US inflation reading now higher than the Fed Funds rate, odds are that no editorial, no policy white-paper, no breathless dispatch from a man with stock options will alter the verdict. Cycles, patterns, and History Herself — that grim and unsentimental dowager — always speak last, and they are not in the habit of asking permission. For four decades money grew cheaper. Now it grows more dear. That, we believe, is the primary trend. AI may be magic or mischief, but the tide does not consult the barnacles.

So how to connect the two dots?

Consider SpaceX, which last year flung $12.7 billion into AI — thrice what it spent on its — and lost roughly half of it. For all the trumpeting, all the press-agentry, all the worshipful prose, this magnificent enterprise has not produced a single honest penny of new wealth. It has consumed billions in real hours and real material. They are not tucked away on some shelf awaiting a rainy day. They are gone — vanished, combusted, irrecoverable as last year’s snow.

To add to the wealth of the world, a company must make a thing it can sell to a willing and able buyer for more than it cost to make it. This elementary act of commerce — known to every fruit peddler and shoeshine boy since the Phoenicians — SpaceX has not yet performed, and may never perform. So far, measured honestly, it is a destroyer of wealth, not a creator of it. And so is AI.

Skanda Amarnath puts the big techs’ wager at some $2 trillion in software and machinery. In the last year alone the hyperscalers are said to have invested $400 billion in AI — a figure expected to swell to $700 billion this year. By what miracle of arithmetic do they ever get that money back?

Capital is not manna. It does not fall free from heaven. It must be reckoned with. And here the outlays are so monstrous that the depreciation alone is likely to dwarf whatever income the ‘intelligent’ machinery throws off. Translated out of the accountant’s mumble: they will lose money, properly counted, and lose a lot of it.

Strange things do happen, of course. We would not be so reckless as to pronounce success impossible — only improbable, which in the long run amounts to nearly the same thing for the man holding the shares.

When the dot-com bubble burst in the year 2000, hundreds of companies went to their graves even though the internet was still alive and well. But Amazon, Facebook, now known as Meta, Google, now decked out as Alphabet, and Microsoft crawled from the wreckage and prospered. They sold real services to real people and pocketed real profits, and they managed the feat without any “policy framework” to lead them by the hand. Now they shovel those profits into a fresh bet — AI — and the question hangs in the air like cigar smoke: will it pay, or are they merely squandering shareholders’ money to keep themselves looking cool?

Nobody knows. Occasionally, an aging gentleman dyes his hair, purchases a scarlet roadster, and happily runs off with a girl half his years. More often, he succeeds only in looking like an old fool.

And perhaps a company grown fat on one technology can vault onto the next. But that is not the way it generally goes. The expert makers of horse drawn carriages did not bolt internal-combustion engines onto their hansoms and cabriolets; instead, they were swept off the road by upstarts named Ford, Dodge, and Olds. Nor, when the internet broke over the land, did the old data-grinders — IBM, Honeywell, Control Data — lead the charge. They were prisoners of their own yesterdays, manacled to the very thing that had enriched them. The new wave was surfed by Gates, Jobs, and Bezos, who had all come to the beach empty handed.

Elon, too, began with nothing, and now gallops at the head of every parade in sight — to Mars, to electric cars, to AI...and his “Boring Company” is busy burrowing beneath Las Vegas.

The hyperscalers are resolved not to be left at the post. Even the flush and the canny ‘whales’ are eager to join in. The Wall Street Journal:

Gina Rinehart, Australia’s richest person, has bought a more-than $1 billion stake in SpaceX, betting on founder Elon Musk and what she says is the potential for his rocket maker to shape industries for decades.

Ne sutor ultra crepidam — let the cobbler stick to his last. Mrs. R., it appears, has mislaid the maxim. The Journal goes on:

Her closely held company built a giant iron ore mine in Australia’s Pilbara, where Rinehart’s father, Lang Hancock, is widely credited with discovering a bounty of iron ore in the 1950s. Today, roughly half the world’s iron-ore exports come from that region.

She made her fortune the honest, gritty way — by selling Australia itself, by the ton. To leap from the certainties of the Pilbara’s open pits into the dark pool of rockets and thinking machines is a plunge into the genuinely unknown. And who can say — the dice are sometimes kind, even to those who throw them blind.

More to follow — on why, in a world where the price of money is climbing, AI may not make us ‘very rich’ after all.

Regards,

Bill Bonner