A Tale of Two Cities, AI Redux
It might seem like AI is taking over the corporate world. A deeper look shows something a little more middling.
It was the most transformative of times, it was the least transformative of times. There’s a lot of hype in the business world of how AI is driving incredible change within their companies right now - layoffs where AI is the main culprit, consternation of entry-level jobs being replaced by AI, and the insane market dynamics behind these proclamations.
But outside of the tech industry, there are actually very few companies using artificial intelligence in novel and transformative ways. Artificial intelligence as a whole has been around for decades, but generative artificial intelligence at the level we see today is novel and demands a corporate repositioning and reality check.
The Most Transformative of Times
We alluded on a recent post that Prof G had a bold claim: AI has driven 60% of market returns since 2023. Mathematically, he may be correct, but digging in the story gets a but more nuanced (as almost any story does behind the headline.)
Yes, the tech giants continue to soar to new heights - mostly off of their cloud businesses. These cloud businesses are enabling companies that do not have the on-premises computing power to run models to quickly tap into the compute power required to move towards an agentic workforce.
Even those companies who may have stubbornly avoided Box or Sharepoint for file services and maintained on-prem servers cannot scale up as quickly as the major cloud players and - if they desire the same AI capabilities as their peers - will need to tap into the cloud. This, and the insane compute and power load that only resource-hoovering data centers (which are popping up in communities least able to resist their polluting ways), has been driving the tech sector’s stock gains.
So, yes, the large tech players are raking it in thanks to the need for other companies to tap into existing AI infrastructure. The question is, how long can this party last? The models are beginning to hit a plateau and the warnings around the energy needs for the kinds of data centers required to maintain this pace are growing. (This analysis won’t even mention the jump scare the industry got from DeepSeek and the prospect that maybe the volume of cloud services filling tech company coffers aren’t actually needed at the scale we’ve seen.)
Even if we can build the power generation required, there’s the pesky task of moving said power from generation to endpoint. The American power grid is woefully out of date, and thanks to hurdles created by regulation and bureaucracy, playing catch-up on an accelerated timeline like Europe would require the US government to, well, do something. No smart money is taking that bet.
So while the tech companies are riding an AI scalability wave, the question is for how long given the absolute sandbar that is power generation and transmission in a United States with embarrassingly poor infrastructure.
The Least Transformative of Times
If the medium-term future may look uncertain at best for AI cloud providers, what about their customers? They must be doing some super cool things with AI that is driving top-line growth, right?
Walk into almost any CFO’s office and tell them you want to start an AI project. After telling you no (it’s reflex, they can’t help it), they will say, “Fine, quantify the top-line growth this company will see from it.”
Dear reader, it has been this author’s job to help justify media budgets for major corporations for over a decade. The relationship between marketing dollars and top-line revenue is well established, and with the right tools and people, teams can draw causation points from A to B. In fact, when you see a company whose market value is anything over its book value, a very large part of that overage is thanks to marketing.
One of the only times we truly have a difficult time doing this is when an entirely brand new tactic or approach enters the marketplace and we want to put money into it to test. Without proven causal relationships, the honest answer to “How much money will this get me?” is sometimes “I don’t know.”
AI’s current state represents this concept on steroids. Another uphill battle is the short-termism that plagues most public companies thanks to SEC requirements for quarterly financial reporting. AI is expensive, and change management is slow. Companies will not see returns within three months of applying a solid, long-term AI strategy. For many shortsighted corporations, this is an unacceptable use of capex for the organization.
But Matt you ignorant slut, you’re probably saying, the McKinseys and the BCGs and the Accentures of the world have positioned themselves as cutting-edge AI practitioners who can accelerate your corporation into the brave new world of artificial intelligence today.
It might come as a surprise to some that those companies that have spent big bucks with the consulting firms to help create and deploy their AI strategies are having a bit of buyer’s remorse, as reported by The Wall Street Journal last week. I knew the article was going to be quite the journey by the fourth paragraph:
Clients quickly encountered a mismatch between the pitch and what consultants could actually deliver. They found that consultants, who often had no more expertise on AI than they did internally, struggled to deploy use cases that created real business value.
We often make fun of the consulting profession as expert Googlers, which may or may not be fair. But some of the quotes from clients in this Journal article are not helping the profession beat those allegations.
A litany of high-profile clients willing to put their name behind some blistering quotes cannot bode well for the consulting industry’s push for AI enablement:
“We love our partners, but oftentimes they’re learning on our dime.”
- Dave Williams, Chief Information and Digital Officer at Merck
“If I were to go hire a consultant to help me figure out how to use Gemini CLI or Claude Code, you’re going to find a partner at one of the Big Four has no more or less experience than a kid in college who tried to use it.”
- Greg Meyers, Chief Digital and Technology Officer at Bristol-Meyers Squibb
“They overpromised. We discovered that they really also had no idea how to do these things.”
- Magesh Sarma, Chief Information and Strategy Officer at AmeriSave Mortgage
“I can’t tell you how many times I heard, ‘Man, they came in, they charged us $20 million and what I feel like we got was a very long report on where AI is going without any real practical application.’”
- Pat Petitti, CEO of Catalant, a platform for freelance consultants.
The Straight Dope
Corporate AI usage needs a reckoning. To be sure, there are plenty of companies out there doing it quite well - I was lucky enough to attend a Section AI chat with Howard Hughes Inc that highlighted some good use cases for AI within large, complex organizations (have I mentioned this newsletter needs sponsors?)
But the more the press highlights AI usage, and the more I learn firsthand by talking with peers across industries, it’s clear there’s a long way to go. And that’s okay!
Generative AI is not only a new technology, but it’s also one that causes massive anxiety amongst the workforce thanks to the constant drumbeat of job replacement that those who are heavily invested in the technology continue to perpetuate. This makes it that much harder to drive uptake amongst a skeptical population who could actually benefit from it, if it weren’t being used as a cudgel to potentially put them in the bread line.
Don’t forget, if you believed the hype train of autonomous vehicles from the mid-aughts, we should currently all be a few years into a reality where most trips are taken in Level 5 autonomous for-hire taxis as opposed to the stupid old way of owning a car with a proper gas and brake (and clutch, if you’re cool.) The fact is, AI itself is complex, and rolling it out through large organizations is inherently difficult and will take time, effort, and money.
Those looking to do it quickly are going to be disappointed. Those hiring firms claiming they can do it for you quickly are going to be even more disappointed (as the above quotes clearly demonstrate.) Rome was not built in a day. And it still wouldn’t have been even if they had AI.
Grab Bag Sections
WTF Complaining: While I have a lot of problems with you people, it feels silly to complain about a crappy commute or a small airplane bathroom given macro events of the past two weeks and beyond. Put simply, America is heading down a path that is going to be extremely hard to return from and that has already set us back years (at the least) on the global stage.
The post-war launching of the Marshall Plan, the beginning of a Pax Americana, and the policies that launched an ascendant United States even further onto the global stage as she helped the globe to rebuild from the rubble of a horrific world war was a mere 80 years ago - I still have family who are alive that remember that time. The world today - and our country’s place in it - must look terribly strange from that vantage point. Losing global hegemony is one thing; willfully ceding it is another altogether.
Album of the Week: If we get into the wayback machine, we had an album of the week about Murphy Lee from Nelly’s St. Lunatics. My memory told me that he and Obie Trice, from Eminem’s D12, dropped the same week, but it turns out it was a few weeks apart. I remember it that way because I didn’t have the money to buy both albums, and I couldn’t pull a Game move. Lucky for this newsletter, I eventually saved up enough to cop Cheers, the debut album from Detroit’s Obie Trice.
While Eminem is famous for his bars, he’s also an accomplished producer, and he EP’d the album and mixed most of it. It’s a solid debut from Obie, who had been simmering in the Detroit underground for years before putting out Cheers. Because it was released during the incredible beef years that the Shady crew had with Ja Rule and Murder Inc., and Benzino and The Source (which somehow survives to this day), there are some solid diss tracks on the album.
“Shit Hits the Fan,” “We All Die One Day,” and “Outro” all heavily feature subliminals and direct hits at the Detroit collective’s enemies. But this is not a mixtape-level CD; songs like “Average Man,” “Don’t Come Down,” and “Look in My Eyes,” feature this newsletter’s sweetspot of artist vulnerability combined with rhyming ability.
Finally, there are no club bangers on this album (even if the gimmicky “Got Some Teeth” might try to corner the bizarre radio play category.) That said, solid features from Nate Dogg (RIP), Eminem, Lloyd Banks, and 50 Cent helped drive the album to be certified platinum in the US. So take an off-season jobbie nooner and devote an afternoon to a solid, made-in-America Detroit rap album.
Quote of the Week: “If destruction be our lot, we must ourselves be its author and finisher. As a nation of freemen, we must live through all time, or die by suicide.” - Abraham Lincoln, 1838
AI Usage in This Post
Other than the images, no AI was used creating this post. That said, one interesting thing occurred while trying to make the final image of a boss not quite understanding why his employee may be hesistant to use the company’s developed AI tools. We all know there are limits on AI image creation, particularly around violence and sexual content. I encountered a new one: “threats of unemployment or intimidation.”
It seems the AI has become sentient enough to become self-conscious about its potential role in job elimination. Go figure.
See you in two weeks!