AI Will Change Everything. And Yet, Most Investors Will Lose Their Money.
Read time: 6 minutes
Welcome to AI-Empowered Leaders. In this weekly email, I share actionable advice on AI adoption, use cases & strategic thinking from my experience as AI Trainer, Leadership Coach, and C-Level Advisor.
From the Trenches
"AI is a bubble."
"AI will take over the world."
AI this. AI that.
Few topics are hotter than AI these days.
I just read another article about the AI bubble.
This one was actually good. Sharp, funny, well-argued. It said OpenAI is burning money with no plan, that almost nobody pays for AI, that the whole thing is parasitic. At many points I thought: maybe he's right.
Then I did what I tell every client to do. I went and checked the numbers myself.
Half of them didn't say what the article claimed. And the half that did told a very different story than the one I'd just been sold.
The Briefing
Here's what's actually on the table right now.
The bear case has teeth. OpenAI's 2025 audited financials leaked in June:
- $13.07 billion in revenue against roughly $34 billion in costs, an operating loss of about $20.9 billion.
- Moody's found roughly $662 billion in data center lease commitments that are signed but sit off balance sheet.
- Morgan Stanley and J.P. Morgan estimate the tech sector needs to issue around $1.5 trillion in new debt over three years to fund this.
The market has noticed.
Across 49 AI and AI-infrastructure stocks in the US, Taiwan and South Korea, the median company is now 40.7% below its all-time high. Thirteen are down at least 50%.
What doomer articles skip:
In the same quarter:
- Microsoft's Azure grew 43%
- Nvidia's data center revenue grew 92%
- Anthropic went from a $9 billion revenue run rate at the end of 2025 to $47 billion by May 2026
- More than 1,000 companies now spend over $1 million a year with them.
- Germany, ifo found 54.5% of companies using AI in May 2026, up from 40.9% a year earlier.
Bottom line: Stock prices falling. Usage exploding. Both true, at the same time.
The Real Story
1) "Is AI a bubble?" is four questions pretending to be one.
This is the mistake almost every article makes, in both directions.
There are four separate questions here:
1) Is the technology still getting better?
2) Does it create real value in real work?
3) Can the companies selling it make money?
4) Are the current valuations justified?
Those four resolve independently. They can point in opposite directions for years.
Most people answer one and assume they've answered all four.
The doomers answer question three and conclude the technology is fake.
The evangelists answer question one and conclude your stock portfolio is safe.
Neither follows.
2) The most likely outcome is that both camps are right.
Look at the railways.
Real technology. Genuinely transformed the economy. Permanently changed how goods and people moved. And a near-total wipeout for the people who financed it.
Same pattern with canals. Same with fiber optics in 1999. The infrastructure got built, the technology stuck, the productivity arrived, and the original investors lost almost everything.
The pattern is clear:
"Revolutionary technology" and "catastrophic bubble" are not alternatives. Historically they're the same event, seen from two different balance sheets.
I think in probabilities (Critical Thinking 101):
- roughly 70% chance of a serious financial correction in the next two years.
- Roughly 25% chance it turns into a genuine macro event, because AI capex is now propping up a meaningful share of US GDP growth.
- Around 80% chance that AI is permanently embedded in knowledge work by 2030 regardless of which of those happens.
Those are answers to different questions.
3) The number that should actually worry you has nothing to do with OpenAI.
McKinsey's latest survey: 88% of organizations now use AI in at least one function. Only 39% report any measurable impact on enterprise profit. And only about 6% attribute more than 5% of their EBIT to AI.
Now that's a gap.
Three years in, near-universal adoption, and 94% of companies can't point to the money.
I don't think it's a tech failure. Stanford's AI Index measured 14 to 15% productivity gains in customer support and 26% in software development.
The value is real and measurable at the task level. It just evaporates before it reaches the P&L.
I see this in every engagement.
Copilot rolled out to four thousand people. Nobody redesigned a single process. The tool sits on top of a workflow built for a world without it, and people conclude AI is overhyped.
Electric motors took factories about forty years to pay off. This is because factories were still laid out around a central steam shaft, and nobody wanted to move the walls.
You're in the wall-moving phase. That's the whole game.
THE PLAYBOOK
1. Stop asking whether AI is a bubble. Ask which of the four questions you're exposed to.
If you're an investor, question four is your problem. If you run a company, question two is the only one that pays your bills. Most leaders worry about the wrong one.
2. Audit your gap between adoption and value.
Pick your biggest AI deployment. Ask: what line in our budget or our P&L moved? If the honest answer is "none," you have a workflow problem, not a tool problem.
3. Assume your vendor might not survive. Build so it doesn't matter.
Keep your prompts, your processes and your team's skills portable. If your AI capability lives inside one vendor's proprietary format, you've made a bet you didn't intend to make.
4. Buy capability now, while it's cheap.
GPT-4 class performance cost around $20 per million tokens in late 2022. In early 2026 it's around $0.40. Investors are currently subsidizing your access to frontier AI. That's a transfer to you. Use it while it lasts.
5. Redesign one process, not ten.
Take a single workflow. Pull it apart. Rebuild it assuming AI is available at every step. That one exercise teaches you more than any pilot program.
THE MONDAY TEST
This week, try this: name one process in your company where AI has genuinely changed how the work gets done, not just how fast someone types.
If you can't name one, you're in the 94%. That's fixable, and it has nothing to do with what happens to Nvidia's share price.
My Honest Take
The "OpenAI's ads are failing by 90%" headline you may have seen is misleading. It's not measured underperformance but a forecast gap: eMarketer thinks the entire US chatbot ad market will be worth $5.41 billion by 2030, while OpenAI is targeting $100 billion on its own. That's still damning for OpenAI's business plan. It says nothing about whether AI works for you.
Whenever you’re ready, here’s how I can help you win with AI:
1) AI Business Advisory
Spot, plan & launch AI use cases that save hours and unlock new value.
2) AI Enablement
Take your team on a journey from AI beginners to critical-thinking power-users—working securely across tools, saving costs, and driving results.
I’ve already trained and coached 3,000+ leaders who are saving hours and performing at a higher level. Your team could be next.
Have questions? Hit reply to this email and I'll help out!
Talk soon,
Alex