Blue Chip Consultants: Spin, Sizzle, and Fizzle with AI
October 14, 2025
This essay is the work of a dumb dinobaby. No smart software required.
Can one quantify the payoffs from AI? Not easily. So what’s the solution? How about a “free” as in “marketing collateral” report from the blue-chip consulting firm McKinsey & Co. (You know that outfit because it figured out how to put Eastern Kentucky, Indiana, and West Virginia on the map.)
I like company reports like “Upgrading Software Business Models to Thrive in the AI Era.” These combine the weird spirit of Ezra Pound with used car sales professionals and blend in a bit of “we know more” rhetoric. Based on my experience, this is a winning combination for many professionals. This document speaks to those in the business of selling software. Today software does not come in boxes or as part of the deal when one buys a giant mainframe. Nope, software is out there. In the cloud. Companies use cloud solutions because — as consultants explained years ago — an organization can fire most technical staff and shift to pay-as-you go services. That big room that held the mainframe can become a sublease. That’s efficiency.
This particular report is the work of four — count them — four people who can help your business. Just bring money and the right attitude. McKinsey is selective. That’s how it decided to enter the pharmaceutical consulting business. Here’s a statement the happy and cooperative group of like-minded consultants presented:
while global enterprise spending on AI applications has increased eightfold over the last year to close to $5 billion, it still only represents less than 1 percent of total software application spending.
Converting this consultant speak to my style of English, the four blue chippers are trying to say that AI is not living up to the hype. Why? A software company today is having a tough time proving that AI delivers. The lack of fungible proof in the form of profits means that something is not going according to plan. Remember: The plan is to increase the revenue from software infused with AI.
Options include the exciting taxi meter approach. This means that the customers of enterprise software doesn’t know how much something costs upfront. Invoices deliver the cost. Surprise is not popular among some bean counters. Amazon’s AWS is in the surprise business. So is Microsoft Azure. However, surprise is not a good approach for some customers.
Licensees of enterprise software with that AI goodness mixed in could balk at paying fees for computational processes outside the control of the software licensee. This is the excitement a first year calculus student experiences when the values of variables are mysterious or unknown. Once one wrestles the variables to the ground, then one learns that the curve never reaches the x axis. It’s infinite, sport.
Pricing AI is a killer. The China-linked folks at Deepseek and its fellow travelers are into the easy, fast, and cheap approach to smart software. One can argue whether the intellectual property is original. One cannot argue that cheap is a compelling feature of some AI solutions. Cue the song: Where or When with the lines:
It seems we stood and talked like this before
We looked at each other in the same way then
But I can’t remember where or QWEN…
The problem is that enterprise software with AI is tough to price. The enterprise software company’s engineering and development costs go up. Their actual operating costs rise. The enterprise software company has to provide fungible proof that the bundle delivers value to warrant a higher price. That’s hard. AI is everywhere, and quite a few services are free, cheap or, or do it yourself code.
McKinsey itself does not have an answer to the problem the report from four blue chip consultants has identified. The report itself is start evidence that explaining AI pricing, operational, and use case data is a work in progress. My view is that:
- AI hype painted a picture of wonderful, easily identifiable benefits. That picture is a bit like a AI generated video. It is momentarily engaging but not real.
- AI state of the art today is output with errors. Hey, that sounds special when one is relying on AI for a medical diagnosis for your child or grandchild or managing your retirement account.,
- AI is a utility function. Software utilities get bundled into software that does something for which the user or licensee is willing to pay. At this time, AI is a work in progress, a novelty, and a cloud of unknowing. At some point, the fog will clear, but it won’t happen as quickly as the AI furnaces burn cash.
- What to sell, to whom, and pricing are problems created by AI. Asking smart software what to do is probably not going to produce a useful answer when the enterprise market is in turmoil, wallowing in uncertainty, and increasingly resistant to “surprise” pricing models.
Net net: McKinsey itself has not figured out AI. The idea is that clients will hire blue chip consultants to figure out AI. Therefore, the more studies and analyses blue chip consultants conduct, the closer these outfits will come to an answer. That’s good for the consulting business. The enterprise software companies may hire the blue chip consultants to answer the money and value questions. The bad news is that the fate of AI in enterprise software developers is in the hands of the licensees. Based on the McKinsey report, these folks are going slow. The mismatch among these players may produce friction. That will be exciting.
Stephen E Arnold, October 14, 2025
Comments
Got something to say?
-
- Subscribe to Beyond Search
Feature archive
News archive
-
