The AI Bubble, The 85% Drawdown, and The Myth of Experience
In this issue:
The “AI bubble” narrative misses the real opportunity
Education matters more than ever in the age of AI
Investor’s POV: Why embracing 85% drawdowns can lead to outperformance
1. The “AI bubble” narrative misses the real opportunity
Woodford Views podcast by W4.0, Episode: The AI Bubble Lie: Why Investors Are Getting This Completely Wrong (Nov. 14, 2025)
TLDR:
The "AI is a bubble" narrative focuses on valuation multiples while ignoring that AI is creating entirely new profit pools that don't exist yet, i.e., it's not just redistributing existing profits
Unlike past tech bubbles (dot-com, crypto), AI has immediate revenue impact: companies deploying AI are seeing measurable productivity gains and cost reductions today, not "someday"
The real question isn't whether AI is overhyped, but where the value accrues: infrastructure layer (Nvidia, Microsoft), application layer (OpenAI, Anthropic), or end users (every company using AI to cut costs)
Every tech cycle generates bubble fears. Dot-com in 1999. Social media in 2011. Crypto in 2021. Now AI in 2025. The pattern is familiar: explosive growth, soaring valuations, breathless media coverage, and inevitable pushback from skeptics warning about irrational exuberance.
But lumping AI into the "bubble" category misses something fundamental: AI is different because it's generating measurable economic value today, not promising value someday.
The difference from past bubbles
The dot-com bubble was about speculation on future business models that didn't exist yet. Pets.com, Webvan, and hundreds of other companies burned billions building infrastructure for markets that weren't ready.
The technology was real, but the business models were imaginary. When the bubble popped, most companies died because they had no path to profitability.
AI is the opposite. Companies implementing AI today are seeing immediate returns:
customer service costs down 40%,
software developers writing code 50% faster,
back-office operations automated entirely.
These aren't projections or promises – they're happening now. That creates a different dynamic.
When a technology delivers immediate cost savings or revenue increases, companies will keep investing in it regardless of stock market sentiment. This creates a self-reinforcing cycle that's fundamentally different from speculative bubbles.
Where the value actually accrues
The key question for investors isn't "Is AI real?" but "Who captures the value?" The answer has massive implications for portfolio positioning.
Right now, infrastructure providers are winning: Nvidia's chips power AI training and inference. Microsoft's Azure and Amazon's AWS provide the compute. These companies have clear revenue models and rising profits. The market's willingness to pay high multiples for these businesses makes sense – they're selling picks and shovels to every company participating in the AI gold rush.
But infrastructure advantages are temporary. Nvidia's dominance will eventually face competition from AMD, custom chips from hyperscalers, and potentially breakthrough architectures that don't require as much compute. The infrastructure layer typically gets commoditized over time.
The application layer – companies building AI products for specific use cases – faces different challenges. OpenAI, Anthropic, and similar companies have massive compute costs and uncertain business models. Can they generate enough revenue from API calls or subscription products to justify their valuations? History suggests most won't, but a few will become massive businesses.
The hidden winners
The real value might accrue to companies that use AI rather than companies that build AI. Consider a law firm that deploys AI to automate document review, cutting costs by 60% while maintaining service levels. Or a manufacturer using AI for predictive maintenance, reducing downtime by 40%. Or a hedge fund using AI to process alternative data, gaining trading edges unavailable to human analysts.
These companies won't be valued as "AI companies" – they'll just be more profitable versions of existing businesses. But the cumulative impact could be enormous. If AI allows the average S&P 500 company to reduce costs by 15-20% over five years, that flows directly to earnings, which drives stock prices regardless of what multiple the market assigns to those earnings.
For investors and financial advisors allocating capital, this suggests a barbell approach: own some pure-play AI infrastructure for exposure to the theme, but also own traditional businesses with strong management teams that will effectively deploy AI to improve margins.
The timing question
Even if AI delivers everything promised, timing matters. Nvidia trading at 40x earnings might be cheap if earnings quadruple over five years. Or it might be expensive if competition emerges and margins compress. The valuation debate is really a timing debate: how fast will the transformation happen, and who benefits most?
The mistake is thinking you need to answer that question with precision. You don't. You just need to recognize that significant AI adoption is happening, it's creating real economic value, and portfolios need exposure to capture some of that value. The exact allocation and specific companies matter less than having any exposure at all.
Consider a parallel: cloud computing in 2010-2015. Skeptics called it overhyped. Valuations seemed stretched. But Amazon Web Services, Microsoft Azure, and others were gaining real revenue and market share. The companies building in the cloud (Shopify, Zoom, Snowflake) or migrating to it (traditional enterprises) all benefited. You didn't need to pick the exact winners – you just needed to recognize the trend was real.
The Bottom Line
The "AI bubble" narrative is lazy thinking that mistakes high valuations for lack of fundamental value. AI is generating measurable economic value today, which makes it fundamentally different from speculative bubbles.
For asset allocators, the question isn't whether to have AI exposure, but how much and where. Some combination of infrastructure providers (Nvidia, Microsoft), beneficiary businesses improving margins through AI adoption, and diversified exposure through index funds provides balanced participation without requiring perfect foresight about which specific companies will dominate.
2. Education matters more than ever in the age of AI (+2 more insights from Tom Sosnoff)
In The Money by Zerodha podcast, Episode: Tom Sosnoff: Inside the Mind of a Trading Legend (Nov. 16, 2025)
Three insights from Tom Sosnoff, who sold 2 investing platforms for $1.8B…
Experience doesn’t give you an edge
"Nobody could watch more ticks in the S&Ps than I have over the last 40 years," Sosnoff admits. "So that should give me an edge, but it doesn't." This isn't false modesty – it's the foundation of his entire philosophy. After four decades, Sosnoff realized that experience doesn't translate to predictive ability. He has no edge over anyone else in knowing what happens next.
This insight drives everything. If you can't predict markets, you need a different approach. Sosnoff's answer: probability-based trading grounded in mathematics, not gut feeling. Instead of asking "What will happen?", he asks "What's likely to happen, and how can I position to profit from probabilities over time?"
Position size solves tail risk
When asked how he handles tail events – those sudden market shocks like the August 2024 volatility spike – Sosnoff's answer is disarmingly simple: position sizing. "I have learned to deal with things by keeping my position size in check," he explains. "I don't think there's a better way to do it."
This matters when managing your - or your client’s - portfolio.
The industry obsesses over complex hedging strategies, exotic derivatives, and sophisticated risk models.
Sosnoff suggests that surviving outlier events comes down to something far more basic: don't bet so big that one bad move ruins you.
Trade small, trade often, and let probabilities work over hundreds of decisions rather than betting everything on one or two big calls.
Why education matters more than ever
Sosnoff maintains that university education remains invaluable. "I can't stress how important university education is," he says. "A 10-year-old can use ChatGPT to find out everything about anything. But that doesn't make it actionable."
The distinction matters for wealth management professionals. Information is abundant and free. What clients pay for – and what distinguishes top advisors – is the judgment that comes from years of experience, the networks built over time, and the maturity to execute when others panic. Technology amplifies these advantages rather than replacing them.
3. Investor’s POV: Why embracing 85% drawdowns can lead to outperformance
I recently sat down with Joe Frankenfield, managing partner of Saga Partners, whose concentrated portfolio has delivered exceptional returns, despite experiencing an 85% drawdown in 2022!
KP: Your portfolio typically only has five or six positions. That's incredibly concentrated, isn’t it?
It’s concentrated compared to most professional portfolios, but that’s not the goal. I’m not optimizing for diversification; I’m optimizing for understanding.
The amount any portfolio should allocate to a single idea should be based on the set of ideas one has and the relative attractiveness of each of the ideas. The number of holdings should reflect how many good explanations you have about real opportunities, not an arbitrary diversification rule.
To make a good investment you must:
Understand the business well enough to explain how it solves real problems and creates value, and
The market must be underestimating that capability
Genuinely good ideas that meet both criteria are rare.
In an ideal world, I’d own dozens of companies if I could find that many that I both understand deeply and think are mispriced by the market. But reality rarely offers that many opportunities and it is difficult for any single person to discover numerous opportunities that meet both criteria.
Most people who own 20 or 30 stocks, let alone 100 or more don’t actually understand them; they’re diversifying ignorance, not knowledge. If that is one’s strategy, investing in an index fund like the S&P 500 probably makes more sense.
KP: Without much diversification, any single drawdown can be painful. How do you prepare yourself for that?
By expecting it as an inevitable feature of investing in stocks.
Most investors, even those who call themselves long-term, are still stock traders at heart. They’re buying today hoping prices go up soon so they can sell. But if you view yourself as a business owner, not a stock owner, the analysis changes completely.
Imagine you could never sell your shares, then the only return you’d receive would be the dividends the business produces over its lifetime. All of sudden that creates a huge barrier to being willing to own any stock because you are never allowed to sell it. That forces you to think differently: about customers, competitive advantage, innovation, and resilience, not about next quarter’s Fed interest rate move or the earnings report.
When you think like that, volatility becomes trivial. Every great business experiences massive drawdowns.
On average, all stocks fluctuate 50–80% within a given year. That is the market’s way of testing ideas. In 2022, all seven of our holdings fell by more than 50% around the same time, and our largest one, Carvana, fell 99%!
That was extreme. But rather than panic, I kept asking: Has my understanding changed? Has the company’s knowledge-creation capacity diminished?
If not, then the market is offering extraordinary opportunity, not a fatal blow. Volatility is not evidence of error; it’s the test that exposes error. If your explanations are good, they survive those tests.
KP: So you’d say embracing volatility is part of the process?
Exactly. Volatility is the price of outperformance. You can’t avoid it because reality will always challenge explanations at some point. Drawdowns are simply moments when the market temporarily disagrees with your understanding. If you’re right, the market will correct itself over time.
The goal isn’t to avoid volatility, it’s to understand what you own so thoroughly that volatility becomes irrelevant. For most stocks, I can not reach that level of understanding so I do not feel comfortable owning them regardless of the price they might sell for in the market. You can’t outsource that thinking. If you don’t understand the business yourself, you won’t survive the inevitable fluctuations.
KP: What's your process for building that level of conviction?
I don’t think in terms of conviction. Conviction is an emotion. Understanding is knowledge.
Investing isn’t about assigning probabilities to unknowable outcomes, that’s pseudo-precision. It’s about creating the best explanation you can today of a company’s problem-solving potential, then relentlessly testing and correcting it as new evidence arises.
The portfolio evolves through error correction. When I discover a better explanation, or realize my current one is flawed, I reallocate. Long-term holdings persist not because of conviction, but because the underlying explanation continues to withstand attempts at refutation.
KP: How has your thinking evolved over time, especially after experiencing such a large drawdown in 2022?
The longer I do this, the more I realize how difficult it is to find truly undervalued opportunities. Managing other people’s money adds another layer of complexity, because their expectations and timelines can vary. Fortunately, my investors are long-term and understand Saga’s philosophy.
After 2022, many people asked what I “learned” expecting me to say I’d diversify more or manage risk differently. But that’s not the lesson. The businesses we own today are pretty much the same ones we owned before the drawdown and their results have largely been what I would have expected given the way the unpredictable macroenvironment has played out.
The real takeaway was to raise the bar even higher on what qualifies as deep understanding. I now ask myself: If this stock dropped 99% tomorrow, how would I feel? If that thought terrifies me, it means I don’t understand it well enough, or it’s not the kind of business I should own.
Another crucial point that the last five years has reinforced in my portfolio management: never be a forced seller. That’s why we don’t use margin, short, or invest in options. You must be able to play out your hand, no matter what happens to the stock quote on any given day.
Volatility isn’t the enemy. Misunderstanding is. The lesson of 2022 is simple: You can’t outsource thinking.
The lesson is: If you don't understand what you own, you won't survive the volatility. And volatility is an inescapable part of the system.
Disclaimer
This newsletter is for informational and educational purposes only and should not be b2construed as personalized financial, tax, legal, or investment advice. The strategies and opinions discussed may not be suitable for your individual circumstances. Always consult a qualified financial advisor, tax professional, or attorney before making any decisions that could affect your finances. While we strive for accuracy, we make no representations or warranties about the completeness or reliability of the information provided. Past performance is not indicative of future results. All investments involve risk, including the possible loss of principal. The publisher, authors, and affiliated parties expressly disclaim any liability for actions taken or not taken based on the contents of this publication.
