Wall Street’s Odor Of Decay

There is the odor of decay on Wall Street. It smells as though the AI brew may be getting ready to turn.

There are three main groups in the AI complex. There are the chipmakers — Nvidia (NVDA) leading the pack, followed by Broadcom (AVGO), Micron (MU), AMD (AMD), ASML (ASML) and TSMC (TSM). They are the pick and shovel sellers. Together, they are said to be worth $11 trillion. But they have only free cash flow of $236 billion. The theoretical investor, if he could buy the whole group and pocket the FCF, could expect to get paid back by the year 2073.

That assumes that no new tech comes along in the next half-a-century to compete with them…and no further capital spending (no new innovations requiring investment) are necessary. We might as well assume that the Baltimore Orioles will win the pennant each year for the next half a century too. But at least we see, once again, how hard it is to make money from an investment when it is over-priced at the get-go.

What sends the chipmakers into orbit is spending by another piece of the confabulation — the ‘hyperscalers’ — Microsoft (MSFT), Alphabet (GOOGL), Apple (AAPL), Oracle (ORCL) and Meta (META). These are companies that survived and grew fat after the last big boom/bust cycle in tech. They are worth, together, about $16 trillion. Now, instead of sharing out earnings with stockholders, they spend their money on capital investments (Capex) trying to stay on top of the latest tech innovations. Which just goes to show how hard it is to make money in new tech at all; there is always newer tech.

Their total Capex for the next twelve months is said to equal to nearly 3% of GDP. And there’s the trick. The money, flowing into the AI cluster, flatters sales and profits for them all. Capital goes into one company…and is taken as income by another. The casual investor sees the whole whoop-de-doo as a boom…and is desperate to get a ticket before the show leaves town.

Finally, there are the explorers and miners themselves…OpenAI, Anthropic, xAI among the most prominent. In turn, they contract with the chipmakers and hyperscalers…providing more ‘proof of concept’ and further exciting the AI bulls.

Behind them all is the hunch that if we were to reduce the cost of knowledge…thereby eliminating the salaries of the people who have it…we’d all be a lot better off.

But it’s all very well to make something cool. Then you have to sell it. Henry Ford doubled his auto factory wages…giving ordinary workers the money they needed to be automobile owners. In the AI revolution, however, instead of doubling their employees’ wages, AI is firing them. Forbes:

Companies Are Firing Workers To Fund AI That Isn’t Working Yet

In the Dot.com craze, investors believed internet traffic would double every 100 days. So, presumably, would sales. Success depended on getting as many ‘eyeballs’ as possible.

But now AI may actually pluck out the ‘eyeballs’ it needs to pay for itself. Eventually, as in all other big tech revolutions, the displaced workers will find new sources of revenue. Perhaps, as in ages past, they will rent themselves out as grooms, gardeners and chamber maids on the vast estates of the AI impresarios. But between the pink slip and the livery, it could be a rough transition.

And while the eyeball base may be shrinking, we are supposed to believe that the big AI ‘labs’ have a tollbooth on the road to tomorrow. This is not an impossible thing. The costs of entry into the business are so high that, like oil retailers, there are naturally only a few of them. Or, citing reasons of ‘national security,’ the feds might take charge of the turnpike themselves. The big tech bosses stood shoulder to shoulder with Donald Trump at his second inauguration. Could some public/private backroom deal be in the works, forcing AI traffic onto a few controlled highways — with regulated tollbooths? Maybe.

But every bubble eventually makes eye contact with its pin. The AI craze is facing all the prickly thorns that typically doom an investment mania. Most important, whatever trickle of profit the industry eventually produces will be dwarfed by the torrents of investment capital that have flooded in to get it.

After the high-water mark in 1999, Microsoft, a genuine success story, saw its stock price cut in half. For Intel (INTC) investors, the pain was much worse. The stock did not recover until this year — more than a quarter of a century after the bust. And JDS Uniphase investors, just to take a name from one of the tombstones, lost 97% of their money.

The pin is patient. But we doubt AI investors will do much better.