TL;DR: The AI boom looks like 1999 all over again. Survive it by building products that print money, save money, kill real pain, and speak plainly to decision makers.
The meeting that broke me happened last month. Brilliant founder, flawless deck. He opened with: “We're building an AI agent that uses cutting-edge LLMs to revolutionize enterprise workflow optimization.”
“Cool,” I said. “What problem does it actually solve?”
Silence. Then: “Well… it's AI.”
That's when it clicked. We're not just in an AI boom. We're in an AI bubble. And a lot of smart people are standing on a trapdoor they can't see.
The Moment I Knew We Were in Trouble
That founder is not alone. I can count on one hand how many AI pitches in 2025 start with a customer problem instead of model specs. If your opener is “We fine-tuned on 9B rows of proprietary data,” I'm already worried.
200 dot-com flashbacks later, it's obvious: we replaced value with vibes. The bubble isn't powered by technology. It's powered by pitch decks.
Red flag: If your entire differentiation fits inside a buzzword cloud—agents, reasoning, context windows—you're betting on investor FOMO, not customer pain.
Welcome to 1999 (Again)
Here’s the scoreboard we’re all ignoring:
- Global AI investment hit $252.3 billion in 2024 — a 13× jump since 2014.
- Tech giants reserved $320 billion for AI infrastructure in 2025.
- U.S. startups raised $104 billion in the first half of 2025, almost matching all of 2024.
- OpenAI is hunting a $500 billion valuation with less than 10% of users paying.
Tell me that doesn't smell like Pets.com. Even Sam Altman admits we're in bubble country, and he's the one selling tickets to the ride.
The ROI Reckoning
MIT dropped the stat that keeps me up at night: despite $30–40 billion in enterprise AI spend, 95% of generative AI pilots produce zero measurable ROI.
Read that again. Nineteen out of twenty projects failed to justify their cost. And yet the money keeps flowing. That's not innovation—that's speculation with better branding.
I see the bubbly logic everywhere:
- Marketing teams rebrand “use ChatGPT more” as an AI strategy.
- Startups tack “AI-first” onto slides to chase the next round.
- Enterprise pilots automate processes no one cared about.
Bubble math: Nvidia invests $100B into OpenAI, which spends it on Nvidia GPUs. Analysts call it revenue round-tripping. Telecom execs tried the same accounting trick in 2001 right before the crash.
The Infrastructure Nobody Will Use
Remember the dot-com era when we laid fiber that sat dark for a decade? We're reenacting it with AI supercomputers named “Stargate” and “Colossus.”
Bain & Company says we need $2 trillion in fresh AI-driven revenue by 2030 just to justify current infrastructure spend. That's not a roadmap. That's a prayer.
Today's motto: “Train it and they will pay.” History answered that with dot-com carnage.
Reality check: The infrastructure won’t save you. Only a business model that prints or protects cash does.
The Survivors' Playbook
When the dot-com bubble popped, the underlying tech survived. The nonsense didn’t. Amazon, Google, eBay—none of them won on tech alone. They won because they solved painful problems.
Make Them Money
Businesses don't buy AI. They buy revenue. Draw a straight line from feature to dollars or expect to be cancelled when budgets tighten.
Save Them Money
If the CFO can't see your impact on a P&L, you are a luxury. Luxuries die first when the music stops.
Solve Real Pain
Painkillers beat vitamins every time. Fraud, downtime, compliance—those pains get budgets forever.
Talk Like a Human
Lose the jargon. Say “Increase conversion by 23%,” not “We combine frontier models with agentic reasoning.”
Want a litmus test? Replace every AI word with a business outcome:
- “Add $4M pipeline in Q4.”
- “Cut support costs by $400K.”
- “Reduce compliance fines to near-zero.”
If the sentence still lands, your AI is worth something. If not, build again.
Rule 01 · Make Them Money
AI lead scoring, personalized outreach, dynamic pricing—these are not “features.” They're revenue lines. In a slowdown, sales leaders keep them. In a boom, they double down.
Rule 02 · Save Them Money
Automation that cuts real headcount, optimization that reduces waste, workflows that shrink cycle times—these are budget-proof. Walmart's AI supply chain optimization saves $75M annually. Try cutting that in a recession.
Rule 03 · Solve Problems That Hurt
Fraud detection. Predictive maintenance. Compliance automation. Nobody brags about them at demo day, yet they're the backbone of resilient AI revenue. Meanwhile, “AI for meeting agendas” dies the moment budgets wobble.
Rule 04 · Talk Like a Human
The best AI products are human + AI. The chatbot handles routine questions, the agent drafts, the expert signs off. Why? Because people trust people. Pitch the outcome, not the architecture diagram.
What Happens When the Music Stops?
If the bubble pops in 2026, here's the dot-com replay:
- Nasdaq drops 80%.
- Thousands of startups disappear overnight.
- Funding deserts the market for years.
But there's a flip side:
- The infrastructure stays.
- The real businesses emerge.
- The next two decades get built on what's left standing.
AI isn't the danger. Bubble math is. The question isn't “Will AI survive?” It's “Will you?”
The Fusion Formula
Contrast the flop pitch with one that closed:
“We built AI for insurance claims. 14 days down to 14 minutes. Labor costs down 60%. Three pilots, 8× ROI in year one. Every pilot converted to paid.”
No jargon. No vision slides. Just a spreadsheet that made a CFO smile. That's the fusion model: cutting-edge AI + boring business fundamentals.
- Revenue: Does it make them more money?
- Efficiency: Does it save hard dollars?
- Pain: Does it fix something they must fix?
- Clarity: Can you explain it in one sentence?
Check all four boxes and you sleep at night. Miss one and you should probably call it a science project.
The Choice
Ride the hype and hope to exit before the crash, or build something durable. The AI party will end. The AI revolution won't.
The survivors won't be the ones with the biggest models or the longest context windows. They'll be the ones who can answer the only question that matters:
Your move. Look at the thing you're shipping today—not the roadmap, not the vision. Would a business in a recession pay for it?
If you hesitated, you know what to do. Build the product CFOs defend when the bubble bursts.
Back to Real Wave Blog