I write this with reluctance because I know that I will receive hundreds of emails correcting me on a few niggling little details.

But write on, I must.

“Write on, write on, write on.”

“Cost of Compute” refers to the $8 to $13 that every AI company has to spend on electricity and short-lived computer chips for every $1 that comes through the door.

Losing a dozen dollars for every dollar you touch isn’t a problem when investors are showering you with cash from a fire hose.

But it’s beginning to look like the well has run dry.

I did not want to defend where I got my information, so I went to the Goog and asked, “Oh Great Googness, why are people referring to the S&P 500 as the “S&P 10”?

Check this out, cub scout, straight from the AI of the Almighty Google:

“The S&P 500 is being referred to as the “S&P 10″ because a handful of massive technology-related companies dominate the index’s performance. Due to market-cap weighting, these top 10 stocks disproportionately influence the index’s total return, making the ‘broad market’ performance heavily reliant on these few, AI-exposed companies. More than $40 of every $100 invested in the S&P 500 is going into just 10 companies, creating a high level of concentration not seen in decades. In some recent periods, those top 10 stocks have accounted for nearly 90% of the entire index’s gains, indicating that the remaining 490+ stocks contribute very little to the overall upward movement.”

Allow me to highlight Three Big Problems.

Manufacturing companies, food companies, service companies, and all the other cash-hungry hopefuls that are the true wonders of the American economy have not been able to raise any money because way too many people have been dumping everything they’ve got into AI.
That money has now slowed down, which means that a lot of AI dependent companies are now being burned by their “burn rate,” a slang term for “precisely how fast they are losing money.”
AI is getting worse, not better, despite the fact that everyone is repeating like parrots, “AI is worse now than it will ever be. Hour by hour, AI will just keep getting better and better forever and ever.”

Okay, I can tell from the look of doubt that I see in your eyes that you need me to explain a little bit more about Problem Number 3.

They can’t raise prices fast enough to stop the bleeding, so most of the AI companies have reduced their Cost of Compute by 86%.

“But how?” you ask.

Here’s how. In the recent past, you could give your $200/mo AI some detailed instructions and it would go on a deep dive to bring you the golden nuggets of information that you requested. The 86% savings of Compute Cost is because they instructed the AI to just look in the cache for what they told someone else who asked a similar question. You get a recycled answer, and the AI company saves 86%.

I’ll wrap this up by giving you the storyteller’s definition of “inflection point.”

Bad storytellers say, “This happened, then This happened, then This happened, then This happened, then This happened.”

Real stories happen like this: “This happened, THEREFORE this happened, BUT then, This happened.”

“BUT then, This happened” is called “an Inflection Point.”

“So what’s going to happen next?”

Let’s wait and see.

Roy H. Williams