By Tony Grayson, President & GM of Northstar Enterprise + Defense | Former U.S. Navy Nuclear Submarine Commander | Stockdale Award Recipient | Veterans Chair, Infrastructure Masons

Published: December 1, 2025 | Updated: January 14, 2026 | Verified: January 14, 2026

TL;DR:

The AI bubble isn't about technology—it's about psychology. When the primary argument for investment is "everyone else is investing," you're watching evolutionary biology, not rational analysis. Seventy-five percent of people will deny their own senses to fit in (Asch experiments). Fund managers face Career Risk: it's safer to lose money with the herd than profit alone. The WPPSS nuclear disaster proves what happens when linear extrapolation meets an S-curve. I'm bullish on AI technology. I'm bearish on the financing behavior around it.

"We're not seeing collective wisdom. We're seeing collective wiring." — Tony Grayson, Former U.S. Navy Nuclear Submarine Commander & Fortune 500 Tech Executive

30-Second Summary

• The Pattern: AI stocks account for 75-80% of S&P 500 returns since ChatGPT launch

• The Valuation: S&P 500 at 23x forward earnings—highest since dot-com bubble

• The Concentration: 5 companies hold 30% of S&P 500—greatest in 50 years

• The ROI Problem: MIT reports 95% of organizations get zero return on GenAI investment

• The Historical Parallel: WPPSS built cooling towers for energy nobody bought—we're pouring concrete for AI clusters on the same logic

• The Bottom Line: The technology is revolutionary. The behavior around it is biology masquerading as strategy.

Commander's Intent

Purpose: Equip investors, operators, and allocators with a clear-eyed framework for recognizing when synchronized investment reflects psychology rather than analysis.

End State: Readers understand the biological drivers of herd behavior (dopamine, FOMO, Career Risk), can identify S-curve traps in technology adoption, and have specific tools to counter their own cognitive biases.

Key Tasks:

1. Explain the evolutionary psychology behind synchronized investment behavior

2. Connect the WPPSS nuclear disaster to current AI infrastructure patterns

3. Provide the "Sober Playbook"—specific techniques to counter herd mentality

4. Distinguish between technology conviction and behavior analysis

The Psychological Drivers

Before diving into the analysis, here's a quick primer on the psychological concepts that drive bubble behavior, and I think are applicable to AI Bubble 2025. If you're already familiar with behavioral economics, skip to the analysis.

Herd Mentality (High Impact)

The evolutionary instinct to follow the group. For 200,000 years, humans who stuck with the tribe survived; outliers got eaten. This wiring doesn't distinguish between physical danger and portfolio risk. When billions flow into AI, your brain reads "safety in numbers"—even when the numbers are wrong.

Career Risk (High Impact)

The professional calculation that it's safer to fail conventionally than succeed unconventionally. A fund manager who loses 30% doing what everyone else did keeps their job. A fund manager who loses 10% doing something different gets fired. This creates systematic pressure toward synchronized positioning.

FOMO and Dopamine (Medium Impact)

Anticipation of gains floods the brain with dopamine, activating the same neural pathways as addictive substances. This chemically suppresses risk assessment. You're not weak for feeling FOMO—you're human. But biology isn't strategy.

Key Concepts

The S-Curve Trap

Technologies follow an S-curve: slow start, explosive middle, flattening top. Bubbles form when investors price the explosive middle as if it continues forever. The trap is extrapolating recent growth rates into perpetuity.

The Asch Conformity Experiments

Classic psychology studies showing 75% of people will give obviously wrong answers to match group consensus. When shown obviously different line lengths, three-quarters of subjects gave wrong answers to match planted confederates. This is the biological foundation of herd behavior.

"It's safer for a fund manager to lose money doing what everyone else is doing than to miss out on gains by standing alone." — Tony Grayson

The Abilene Paradox

A group phenomenon where everyone collectively decides on an action that no individual actually wants—because no one wants to be the dissenter. Investment committees often exhibit this pattern: private doubt, public consensus.

Normalcy Bias

The tendency to interpret warning signs through the lens of past experience. "We've seen volatility before, and it worked out." This is what killed the Challenger crew—engineers had documented O-ring problems, but the system had always held. Until it didn't.

AI Bubble 2025

The biggest argument against AI Bubble 2025 is that "everyone is doubling down." History and psychology tell us that is exactly when we should be most worried.

Where I land: Having spent years commanding nuclear submarines and later building cloud infrastructure for global tech giants, I’ve learned that the most dangerous variable in any complex system isn't the hardware—it's the human operator. The innovation is real, but the behavior surrounding it has decoupled from reality. We’re mistaking biological impulses for strategic conviction.

The defense for current valuations always relies on social proof: "The smartest people in the room are doubling down." This argument mistakes volume for value. The fact that everyone is doubling down doesn't prove the valuations are rational. Ironically, it is the definitive psychological signal that we are in a bubble fueled by AI Bubble Herd Mentality.

When we see massive, synchronized investment waves, we’re not seeing collective wisdom. We’re seeing collective wiring. This creates a powerful Cognitive Trap Investment that leadership must build systems to counter.

The Human Operating System: The Failure of Linear Thinking

We like to believe that rational actors drive markets; in reality, they are driven by dopamine, cortisol, and ancient evolutionary impulses.

1. The Evolutionary Imperative: Herd Mentality

The most powerful force in finance isn't interest rates; it's the deep-seated human need to belong to the group. The fear of being the outlier translates into Career Risk. It is professionally safer for a fund manager to lose money doing what everyone else is doing than to miss out on gains by standing alone.

• Evidence: The classic Asch Conformity Experiments demonstrated that nearly 75% of people will deny their own senses to fit in with a group. Today's investors, when they see billions pouring into the AI Bubble, are not calculating fundamentals; they are agreeing with the group, terrified of being the outlier.

2. The Neurochemistry of FOMO and Greed

The anticipation of quick gains floods the brain's reward system with dopamine. This neurochemical response is not a simple character flaw; it is a biological phenomenon.

• Evidence: Neuroeconomics research confirms that the anticipation of financial gain activates the same neural pathways as addictive substances. Read more on behavioral science and investing. This dopamine rush chemically suppresses our brain’s capacity to assess downside risk. When the AI Bubble Herd Mentality takes over, we are chemically encouraged to ignore the cliff edge.

"FOMO isn't a character flaw. It's neurochemistry. Your brain is drugging itself into confidence." — Tony Grayson

The WPPSS Warning: Mistaking the S-Curve for a Straight Line

A primary driver of AI Bubble psychology is Recency Bias—the instinctive belief that explosive past performance equals infinite future performance.

"They built cooling towers for energy nobody bought. Today, we're pouring concrete for AI clusters based on the same linear extrapolation." — Tony Grayson, who served 21 years in the Navy's nuclear program

The S-Curve Trap

Humans are terrible at intuitive statistics. We look at the immediate past (the last 12 to 24 months of explosive growth), and we instinctively extrapolate that line straight up and to the right forever.

• Historical Data: History proves that growth is almost always an S-curve—it explodes, then flattens. Bubbles occur when investors price an S-curve as if it were a straight exponential line.

We are currently mistaking the "Inflection Point" for permanent exponential growth—the classic trap.

The Nuclear Construction Analog

Forget the Dot-Com bubble; the most relevant analog is the Nuclear Construction Boom of the 1970s. Utilities assumed post-war electricity demand would double every decade and projected it linearly forever. They borrowed billions to build massive fleets of reactors.

• The Flawed Behavior: The most famous example is the Washington Public Power Supply System (WPPSS). The thesis was sound ("the world needs more energy"), but the behavior was flawed. The demand curve was an S-curve, not a straight line. When efficiency improved and demand flattened in the 1980s, the project collapsed into the largest municipal bond default in history.

• The Warning: They built cooling towers for energy that nobody bought. Today, we are pouring concrete for gigawatt-scale AI clusters based on the exact same linear extrapolation. We are building the WPPSS of the 21st century, confusing a temporary spike in training demand for infinite utility. This high-risk behavior is tied to the Nvidia Vendor Financing Trap.

The Sober Playbook: Leading Against Your Own Biology

"The most dangerous variable in any complex system isn't the hardware—it's the human operator." — Tony Grayson

If you are a leader, the answer is to recognize your own biology and build systems to counter the AI Bubble Herd Mentality:

• Stress-Test the S-Curve: Ask your team: "If adoption flattens next year, does this investment still make sense?" If the answer is no, you are gambling on a straight line in a curved world. This rigorous stress-testing is essential, as discussed in Why Business Continuity Plans Fail.

• Create Safety for Dissent: Assign a "designated dissenter" in every meeting to break the Abilene Paradox and the illusion of unanimity. This is an application of Contextual Intelligence vs. Servant Leadership.

• Separate Technology from Psychology: Be bullish on the innovation, but bearish on the behavior.

The choice is always the same: A collection of scars or a collection of what-ifs. When the primary argument for investment is the sheer volume of investment already occurring, the thundering sound is the AI Bubble Herd Mentality, driven by biology, running full speed in one direction.

"Be bullish on the innovation. Be bearish on the behavior." — Tony Grayson

Frequently Asked Questions: AI Bubble Psychology & Investment Risk

What is the AI Bubble Herd Mentality?

AI Bubble Herd Mentality is the phenomenon where massive, synchronized investment in the AI sector is driven primarily by psychological pressures (FOMO, Career Risk) and social proof, rather than purely rational financial models. It reflects the deep-seated human need to conform to the group. When the primary argument for investment is the sheer volume of investment already occurring, that's the definitive psychological signal of bubble behavior.

What is the Cognitive Trap Investment?

The Cognitive Trap Investment refers to behavioral biases—specifically Recency Bias and Linear Extrapolation—that cause leaders to mistake the temporary explosive growth phase of an S-curve for permanent, infinite exponential growth, leading to asset mispricing and bubble formation. Humans are terrible at intuitive statistics and instinctively project recent explosive growth as a straight line forever, when growth is almost always an S-curve that flattens.

How does the WPPSS project relate to the AI Bubble?

The WPPSS (Washington Public Power Supply System) project of the 1970s is a key historical analog. WPPSS assumed infinite energy demand and built billions in capacity based on linear extrapolation. When demand flattened in the 1980s, the project collapsed into a $2.25 billion default in 1983—the largest municipal bond default in U.S. history. They built cooling towers for energy nobody bought. Today's AI infrastructure investment risks the same mistake: overbuilding based on linearly extrapolating training demand.

What is the antidote to Herd Mentality in investment decisions?

The antidote is to build organizational systems that prioritize logic over impulse. This includes: stress-testing assumptions against S-curve flattening ("If adoption flattens next year, does this investment still make sense?"), creating safety for dissent by assigning a designated dissenter in every meeting to break the Abilene Paradox, and separating technology from psychology—being bullish on the innovation while bearish on the behavior. More on building systems to counter herd behavior in Why Business Continuity Plans Fail.

What is the S-Curve Trap in AI investing?

The S-Curve Trap occurs when investors look at explosive recent growth and extrapolate it straight up forever, when history proves growth is almost always an S-curve that explodes, then flattens. Bubbles occur when investors price an S-curve as if it were a straight exponential line. We're currently at the inflection point, mistaking it for permanent exponential growth—the classic trap of bubble psychology.

What are the Asch Conformity Experiments and how do they apply to AI investing?

The Asch Conformity Experiments demonstrated that nearly 75% of people will deny their own senses to fit in with a group. Today's investors, when they see billions pouring into AI, are not calculating fundamentals—they are agreeing with the group, terrified of being the outlier. This evolutionary imperative to belong creates Career Risk: it is professionally safer for a fund manager to lose money doing what everyone else is doing than to miss out on gains by standing alone.

How do dopamine and neurochemistry drive FOMO investing?

The anticipation of quick gains floods the brain's reward system with dopamine. Neuroeconomics research confirms that the anticipation of financial gain activates the same neural pathways as those activated by addictive substances. This dopamine rush chemically suppresses the brain's capacity to assess downside risk. When AI Bubble Herd Mentality takes over, investors are chemically encouraged to ignore the cliff edge—it's not a character flaw, it's biology.

What is Career Risk and how does it fuel bubbles?

Career Risk is the fear of being the professional outlier. It is safer for a fund manager to lose money doing what everyone else is doing than to miss out on gains by standing alone. When clients ask "Where is your AI investment?" portfolio managers seek something with AI in its name to show, often without serious thinking about what the business will look like. They're buying a name because clients want something to point to as their AI investment. This creates self-reinforcing cycles described in Contextual Intelligence vs. Servant Leadership.

What happened with WPPSS and why did it default?

WPPSS (nicknamed "Whoops") planned to build five nuclear power plants in the Pacific Northwest. Initial cost estimates of $4.1 billion ballooned to $24 billion. When energy demand flattened rather than growing exponentially as projected, the project collapsed. In July 1983, WPPSS defaulted on $2.25 billion in bonds—the largest municipal bond default in U.S. history. Only one of the five plants was ever completed. The cooling towers at Satsop, Washington still stand as monuments to linear extrapolation failure.

Is there currently an AI bubble in 2025?

Experts remain divided. Howard Marks of Oaktree Capital notes valuations are "high but not crazy" and hasn't detected "critical mass of mania." However, warning signs include: AI-related stocks accounting for 75-80% of S&P 500 returns since ChatGPT launched, S&P 500 trading at 23 times forward earnings (highest since dot-com), 30% of S&P 500 held by just five companies (greatest concentration in 50 years), and MIT reporting 95% of organizations getting zero return on $30-40 billion in GenAI investment.

What is the Abilene Paradox and how does it affect AI investment decisions?

The Abilene Paradox occurs when a group collectively decides on a course of action that is counter to the preferences of many individuals in the group, because no one wants to be the dissenter. In AI investment committees, everyone may privately doubt the valuations but no one speaks up because they assume everyone else believes. Creating a safe space for dissent (assigning a designated dissenter in every meeting) breaks this illusion of unanimity.

What lessons from nuclear construction apply to AI infrastructure?

The 1970s Nuclear Construction Boom offers critical lessons: utilities assumed that post-war electricity demand would double every decade and projected it to continue linearly forever. When efficiency improved and demand flattened in the 1980s, projects collapsed. Today's AI infrastructure faces similar risks: we're pouring concrete for gigawatt-scale AI clusters based on linear extrapolation of training demand. The technology may be real (nuclear was real, AI is real), but the behavior surrounding it has decoupled from reality. More on this risk in Nvidia Vendor Financing Trap.

Watch: The Psychology Behind Market Bubbles

For a further visual analysis of the neurochemistry driving these decisions, watch this video on behavioral finance:

Related Articles from The Control Room

Nuclear for Data Centers: Why the Gen IV SMR Timeline is 2035 — The engineering reality behind SMR delays and what it means for AI power demand

AI Training vs. Inference: The $300 Billion Shift Nobody's Pricing — Where AI infrastructure demand is actually heading as the market shifts from training to inference

The $3 Trillion AI Supercycle: A Submarine Commander's Risk Assessment — Why JLL's bullish report misses critical AI infrastructure financing risks

Fermi Stock Crash Explained: The First Casualty of AI's Nuclear Fantasy — What FRMI's 33% drop reveals about AI infrastructure investor psychology

Sources

Asch Conformity Experiments — Simply Psychology, original research on 75% conformity rate

Howard Marks: Is It a Bubble? — Oaktree Capital, October 2025

Yale Insights: This Is How the AI Bubble Bursts — Yale School of Management, October 2025 (MIT 95% zero return study)

Federal Reserve Financial Stability Report — Board of Governors, November 2025

NPR: Here's Why Concerns About an AI Bubble Are Bigger Than Ever — Bobby Allyn, November 2025

CNBC: Are We In An AI Bubble? — October 2025

PBS NewsHour: What's Next for AI — Cade Metz interview, December 2025

CFO Brew: The AI Bubble, Explained — NYU Prof. Aswath Damodaran, November 2025

Neuroeconomics of Financial Decision Making — National Institutes of Health (.gov)

Wikipedia: Energy Northwest (WPPSS) — $2.25B default history

Wikipedia: Abilene Paradox — Group dynamics phenomenon

Wikipedia: S-curve (Sigmoid function) — Technology adoption curve

___________________________

Tony Grayson is a recognized Top 10 Data Center Influencer, a successful entrepreneur, and the President & General Manager of Northstar Enterprise + Defense.

A former U.S. Navy Submarine Commander and recipient of the prestigious VADM Stockdale Award, Tony is a leading authority on the convergence of nuclear energy, AI infrastructure, and national defense. His career is defined by building at scale: he led global infrastructure strategy as a Senior Vice President for AWS, Meta, and Oracle before founding and selling a top-10 modular data center company.

Today, he leads strategy and execution for critical defense programs and AI infrastructure, building AI factories and cloud regions that survive contact with reality.

Read more at: tonygraysonvet.com