The A.I Bubble Driven By Legacy Software Firms

Availability of electric power will force the reckoning!

The world has been sold a story.

The story goes that the only way to make your application, or AI agent, fast and scalable is to move it and all your data to the cloud.

In pursuit of this story, the world is frantically spending trillions of dollars to build more, and bigger data centers.

There is only one problem – the story is not true.

Placing your application, or AI agent, in a cloud data center and bringing all your data to that cloud is expensive and inefficient – and there is a sustainable alternative.

Fractal Computing enables re-implementing the software so that it is much smaller -- copying the smaller software to all of the data locations.

Re-implementing in days - not years.

Fractal Computing coordinates the software copies so they behave as a single system.

No cloud, no data center required.

Sustainable computing is here. Currently in production with billion dollar corporations.

Fractal Computing exposes the false story and breaks the cycle of ever-growing, unnecessary energy demands.

Building massive, nuclear powered data centers for A.I. is one of those phrases that will go down with putting a colony on Mars in our lifetime - as a joke in a decade.

Sounds great but the physics don’t work.

You are living through a Bubble - the A.I. Bubble.

It isn’t that A.I. won’t happen - it’s that A.I. in some form is 40 years old - it’s BEEN happening for a long time.

The difference now is LLMs - large language models.

We aren’t going to get into them here - Google them if you like.

The issue for our readers is the drive for massive new data centers is not an A.I. thing - it is a LLM thing - and LLMs are not necessary.

There are now a growing number of LLM alternatives - both here and in Europe - some having to do with integer work versus floating point - it gets geeky too fast for our audience and for the most part, it will sort itself out over the next couple of years.

You will learn from poking around that LLMs are not necessary - that will be accepted wisdom within the year - there are a growing number of alternatives.

What is important here - data centers are not needed to deliver A.I. dominance.

In fact, as we will note in an upcoming post, adopting the data center coast to coast model (they won’t go coast to coast - many will end up in Virginia, sadly), that model may place America BEHIND adversaries in A.I. and other advanced computing development.

Anyway, you are living through a BUBBLE. Let’s review some prior tech bubbles for the younger readers.

If you lived through the Netscape days in the 1990s - when a browser company - it’s just a browser and now it’s gone - drove Wall Street madness - you saw this craziness before.

Netscape stock was shooting up dollars a minute the week it hit the stock exchange - everything was Netscape. It was just a browser, not remembered by anyone under 30 today. Like Bob Hope and Bing Crosby.

How about the Year 2000 bubble?

Because computers did not have the code to move 1999 to 2000 in their year-end calculations, planes might fall from the sky.

Companies spent billions, under great pressure - to modify those date calcs. Boards of directors had Year 2000 committees to make sure their company did not explode due to a date calculation.

Nothing much happened - but a lot of software companies got rich. The last day of December 1999 the big story was that nothing happened.

American companies got smoked by the legacy software and consulting industry!

Remember when a preposterous company who was nothing but an ISP (you don’t probably know who your ISP is right now!) - called AOL - America OnLine - merged with Time Warner becoming one of the most talked about mergers in history?

Remember that nonsense!

The whole thing blew up right afterward and AOL is now remembered in 1990s flicks for its ever present CDs - mailed by the gazillions - showing up as wall art.

Tech bubbles have patterns - let’s review them here.

Tech bubbles are always driven by tech companies trying to shorten sales cycles, go public, use manufactured madness to generate energy disproportionate to their product capabilities.

Think Netscape.

Another characteristic - in a bubble, everyone, and I mean everyone is in on it. Every day someone gains clicks - readers years ago - by ratcheting up the ante one more time.

The tech press leads with greater predictions every month.

Technical knuckleheads like Fox’s Charles Payne - who know zip about anything more than selling a book that will insure you underperform the S&P - tout how important data centers are - everywhere - now!

Cloud computing is one of the current bubbles now retreating toward its statistical mean - remember there is not cloud - it’s just someone else’s data center.

If you are old enough, you will remember “timesharing.” It had a short life in the late 70s and died in the early 80s. Timesharing was using someone else’s computer to do stuff - precisely the same as the “cloud today.”

Companies roared to “cloud computing” a few years ago - now that baby is settling down as the bills come in and CEOs and CFOs are discovering that running your data in two places costs twice as much.

Who knew?

Bubbles always end. Some like Year 2000 just go away. Netscape and word processors like Wang, for example.

Others linger on - but in a dwarfed, reduced form that parallels their true, basement market valuation - DevOps, virtualization and other existing technologies live on.

A.I. madness is happening today on very different levels.

First there is the CHAT GPT stuff which to A.I. deep practitioners is a toy.

Comedian “Professor” Irwin Corey decades ago presaged the state of much of these A.I. results in play today - watch this video:

https://youtube.com/shorts/s6mGapqLloE?feature=shared

Take about every document you can find, put them in an indexed database - throw a ton of power at it and you can ask any question about any subject and get an answer.

The result sounds great - someone did it last week with some of our written material and it was hilarious - you hear your ideas written back from someone with no edges.

That’s what we would sound like if we just paid more attention in English class!

A.I. answers are smooth - sound nice to the ear - no shocks or bumps in the message - because A.I. does not think, it does not question, it returns text via complex indexing.

Those answers are insightful - because they are answers someone put into a document long ago - so you think you just applied artificial intelligence.

You didn’t do anything more intelligent than using an optimized index. Probably open source.

The issue at hand is data centers - $500 billion worth - or not - to achieve A.I. dominance.

Their argument - FOR endless energy-consuming data centers is building a BUBBLE.

As we noted in last week’s Black Swan post - which went viral - the big data center guys are screwed because we can show there are no applications needing a big data center.

We and others in our world, can take any application, A.I. or not, that today requires a data center, and we can run it on a 4-inch by 4-inch computer - usually in a few days - always in less than a quarter.

We thought we would entertain you with an example of an application that cannot be done, with any conventional technology, on tiny computers. So let’s do it here:

https://rumble.com/v6pusnc-who-funds-indivisible-they-are-astroturfing-d.o.g.e..html

Scroll around in the video - it may not be your thing. But here is what this application does.

It tracks 2.4 million NGOs including every dollar they receive from other NGOs, every dollar they give to other NGOs, performs recursive analysis showing all forward paths of that money and reverse paths.

It tracks every employee, every director, address, phone, employee - going back 7 years.

Everything you see in this video - all 2.4 million of those nasty NGOs - all of it runs on a 4-inch by 4-inch computer. That’s like your cell phone on top of your spouse’s cell phone - not a data center.

The power used is that of a desk lamp - less perhaps.

It runs - as you can see - in real time - every answer instantly generated both graphically and textually.

There is no conventional technology on the planet today - that can do this level of compute on such a small hardware footprint - at such speeds.

You career-ended SQL cubicle residents, with your VMware logo XXXL shirts and Microsoft diplomas stapled to the cubicle’s cloth wall - before you ever send us a challenging note again - show us a video of you doing this with any conventional technology.

Otherwise, skip the “…I’m an engineer and they really can’t show…….”

We just did!

Don’t email me that I am not providing deep technical info here - this is a Substack, not product documentation.

We have scores of videos of live systems, we demonstrate billion record databases in real time - running from a tiny computer all the time - to qualified prospects.

We show these videos on free websites and have done so for 4 years.

A.I. isn’t new nor is it going away. A.I. has been around for a generation.

A.I. isn’t the bubble - the bubble is software companies, consulting companies, tech press - scaring everyone that America will lose the A.I. race with China because we do not have big data centers on every corner.

Our Chinese pals - remember DeepSeek - proved THEY do not need big data centers.

Why do we?

Fractal and our growing family of uncles and partners are taking it one step further.

No application, A.I. or legacy needs a data center. Period.

We prove it every day with real world systems moved to 4-inch by 4-inch computers - running, for instance, the entire Federal Election Commission database - which takes the government a $40 million data center.

We do the same thing - faster, consuming almost no energy on a 4-inch by 4-inch computer - again, that is the Black Swan Event.

This week one of the biggies - one of the monster data center software companies called us - because they cannot - with all that horsepower - process the data for an NGO tracking problem at quantum speed.

We can, we do, we did. They still cannot.

They use conventional tech and if they cannot do it, nobody can.

In an upcoming piece, we are making the case that the legacy software industry is pushing for the adoption of the big data center model - because their 1980s tech needs it.

Adopting their big data center model is harmful to children and other living things - as they said in the 60s.

We will address that issue shortly

FractalComputing Substack is a newsletter about the journey of taking a massively disruptive technology to market. We envision a book about our journey so each post is a way to capture some fun events.

Subscribe at FractalComputing.Substack.com

Fractal Website: Fractal-Computing.com

Fractal Utility Site: TheFractalUtility.com

Fractal Government Site: TheFractalGovernment.com

Jay@FractalWeb.App

Portions of our revenue are given to animal rescue charities.