It’s worse. It’s also better. And it points to a specific outcome in gold.
I recently wrote that what we see now is a dot-com 2.0, and I received a thoughtful comment (thanks, Eric!) below the analysis saying that “The Difference between the Dot.com bubble / crash and this AI bubble is the AI companies make money as none of the dot com companies did, it’s a tough compare.”
That inspired me to dig some more, and it turned out to be a more complex and more interesting analysis than I had expected. Long story short, it’s better and worse than it was at the dot-com bubble, depending on how one looks at it. Overall, I’d say that we’re somewhere between the dot-com and 2008 in terms of analogies.
It still seems that we’re going to get a medium-term crash in equity, commodity, and precious metals prices, while the USD Index soars.
Here’s why:
Is This Dot-Com 2.0? What the AI Bubble Means for Gold Investors
The question on every investor's mind right now is whether we're witnessing a repeat of the dot-com bubble - and what that means for precious metals positioning. After conducting extensive research comparing the current AI boom to the late 1990s internet mania, I think the answer is more nuanced than most commentators suggest.
Yes, we're in a bubble. But it's a different kind of bubble with both better fundamentals and worse systemic risks.
Let me explain why this matters enormously for gold investors, and why the implications differ significantly from the 2000 crash.
The Numbers Tell a Mixed Story
The most straightforward comparison starts with the Buffett Indicator - Warren Buffett's preferred measure of whether the stock market is overvalued. This ratio compares total market capitalization to GDP, essentially asking whether stock prices have gotten ahead of the real economy.
The current Buffett Indicator sits at approximately 216-230% (depending on calculation methodology), according to multiple sources tracking this metric in real-time. To put this in perspective, the dot-com peak reached "only" 143%.
That's right - we're 61% worse than the dot-com bubble peak by this critical measure.
But here's where it gets interesting. The NASDAQ Composite's forward price-to-earnings ratio hit approximately 60x at the dot-com peak in March 2000. Today's "Magnificent Seven" AI companies (Microsoft, Alphabet, Amazon, Meta, Apple, Nvidia, Tesla) trade at roughly 27x two-year forward earnings - less than half the dot-com valuation extreme.
So how can the market be more overvalued overall while individual company valuations are more reasonable? The answer lies in market concentration and the profitability divide.
The Critical Difference: Profitability
This is where the AI boom fundamentally differs from dot-com mania.
During the dot-com peak, 86% of NASDAQ-100 technology IPO firms were unprofitable. Of 199 internet stocks tracked by Morgan Stanley in October 1999, the companies collectively generated $21 billion in revenue but posted $6.2 billion in losses - an industry-wide negative profit margin.
Companies like Pets.com spent $11.8 million on advertising to generate just $619,000 in revenue during a nine-month period. They lost $0.26 per dollar of sales before even counting marketing expenses. Webvan spent over $10 for every $1 earned. These weren't viable businesses stretching to profitability - they had fundamentally broken unit economics.
Today's AI leaders are among the most profitable companies in history. Nvidia generated $29.76 billion in net income on $60.92 billion in revenue for fiscal 2024, maintaining gross margins of 73-76% and operating margins around 52%. Microsoft posted $101.83 billion in net income for fiscal 2025. Google's net income exceeds $70 billion annually.
These companies aren't hoping to become profitable someday - they're printing money right now.
However, AI startups look eerily similar to their dot-com predecessors. OpenAI reportedly lost $5 billion in 2024 on $3.7 billion in revenue - a 135% loss-to-revenue ratio. The company projects it won't break even until 2029-2030, burning through potentially $115 billion in cumulative cash by then. Yet it recently raised capital at a $300 billion valuation, representing 75x its 2024 revenue or 23x its projected 2025 revenue.
For context, Amazon at its dot-com peak traded at roughly 18x revenue while losing money - and that was considered insane at the time.
The Concentration Risk Very Few Talk About
Here's what really concerns me from a systemic risk perspective: market concentration has reached levels that exceed even the dot-com era.
The top five companies now represent 30% of S&P 500 market capitalization - the highest concentration in 50 years. AI-related stocks have contributed approximately 75% of S&P 500 returns, 80% of earnings growth, and 90% of capital spending growth since ChatGPT's November 2022 launch.
During the dot-com bubble, internet stocks were more dispersed across hundreds of companies. When the bubble burst, many individual companies went to zero, but the losses were distributed. Today's structure concentrates enormous valuation in just a handful of names, meaning corrections can produce outsized market-wide impacts.
Yale School of Management's Jeffrey Sonnenfeld calculated that AI-related capital expenditures actually surpassed U.S. consumer spending as the primary driver of GDP growth in the first half of 2025, accounting for 1.1% of GDP growth. When a single sector drives that much economic growth through capital spending rather than productivity gains, the economy becomes structurally vulnerable to investment pullbacks.
The Debt Financing Wild Card
This is truly critical, and a reason as to why I think the eventual correction could be more severe than the dot-com crash despite better underlying fundamentals.
AI now accounts for 14% or more of investment-grade corporate debt, with companies issuing massive amounts of debt for data center construction and AI infrastructure. Microsoft has guided toward $80 billion in AI-enabled data center capital expenditures for fiscal 2026. Oracle is structuring complex deals with OpenAI involving hundreds of billions in computing commitments.
The dot-com bubble was primarily equity-funded. When it burst, investors lost money, but banks and credit markets remained stable. The recession lasted just eight months, GDP declined only 0.3%, and recovery began quickly because no credit crisis materialized.
The 2008 housing bubble, by contrast, involved massive debt financing - mortgages, CDOs, bank credit. When defaults cascaded through credit markets, it transformed a sector correction into a systemic financial crisis.
Current AI investment sits somewhere in between. The debt component is growing but not yet dominant. However, if this trend continues and companies can't generate returns to service these debts, we could see a 2008-style credit crisis rather than a manageable equity correction.
The IMF has issued warnings about this specifically, with Chief Economist Gita Gopinath suggesting potential global asset losses of $35 trillion if a severe correction occurs - seven times the $5 trillion destroyed in the dot-com crash.
What Does This Mean for Gold?
The AI bubble story is an interesting context, but it's not what's going to drive gold prices in the months ahead. The USD Index is.
Let me explain why this matters so much. Gold investors often assume that equity market stress automatically translates to gold strength. The logic seems obvious: stocks crash, investors panic, money flows to safe havens, gold rises. But this relationship is far more conditional than most people realize.
During the dot-com crash (2000-2002), gold rose modestly from approximately $280/oz to $310/oz while the NASDAQ fell 78%. That's a 10% gain over two years during one of the worst equity crashes in history. Not exactly the safe-haven surge many expected.
The critical detail that the above doesn’t tell? The precious metals market still declined heavily during the initial stage of the stock market slide.
Also, when the above happened, the precious metals market was after a multi-year bear market. The opposite is the case right now.
It’s much more like in 2008, where PMs were vulnerable to a sell-off. In particular, silver and miners fell hard back then.
In other words, it’s much more likely that PMs will decline along with stocks instead of being supported by them.
The USD Index’s impact goes well beyond the traditional link in this case.
How USD Strength Accelerates the Bubble Dynamic
This is where your USD strength projection from the Peak Chaos analysis becomes critically important. A rally in the USD Index from 100 to 110 would significantly accelerate the AI bubble correction timeline through multiple quantifiable transmission mechanisms.
The research is unambiguous on this point: USD strength acts as a systematic tightening mechanism that exposes leveraged positions and debt vulnerabilities across the financial system.
The Credit Tightening Channel
The research from the Federal Reserve and leading financial institutions has established clear quantitative relationships between USD strength and credit conditions. A 1% increase in the broad dollar index is associated with a 0.5% increase in BBB-rated corporate bond spreads over Treasury yields.
For context, a move from USD Index 100 to 110 represents a 10% appreciation. Using the empirically-established relationships, this would imply:
· Corporate bond spreads are widening by approximately 50 basis points across investment-grade debt
· Bank lending contracting by roughly 10% for every 2.5-point USD Index move (based on Federal Reserve research showing this correlation with C&I loan originations)
· Mutual fund outflows of 0.77% of assets under management for each standard deviation move in the dollar
Currently, AI companies account for 14% of investment-grade corporate debt. That $560 billion invested by the hyperscalers includes substantial debt financing. Oracle's complex deals with OpenAI, Microsoft's $80 billion in AI data center capex, and similar structures all depend on favorable credit market conditions.
If USD strength drives credit spreads from current 2.8% levels to 3.3-3.8%, it fundamentally changes the economics of AI infrastructure financing. Companies that looked marginally profitable at low borrowing costs suddenly face significantly negative unit economics.
The International Revenue Exposure
Most AI leaders generate 40-60% of revenue internationally. Microsoft derives approximately 48% of revenue from outside North America. Alphabet and Amazon show similar patterns. A 10% USD appreciation mechanically reduces the dollar value of international earnings by approximately 4-6% for these companies, all else equal.
This matters enormously when you're trying to justify capital expenditures requiring multi-year payback periods. If Oracle projects it needs $300 billion in computing revenue from OpenAI over five years to justify its data center investments, and the strong dollar reduces international revenue growth projections, the entire business case deteriorates.
The financial models underlying current AI valuations already assume aggressive revenue growth. A strong dollar makes those assumptions even more heroic.
The Dollar Funding Stress Mechanism
Research from the IMF and Bank for International Settlements demonstrates that USD strength widens cross-currency basis swaps, making dollar funding more expensive for international borrowers and creating stress in global funding markets.
Current cross-currency basis levels sit at -50 to -100 basis points for major currency pairs. During the March 2020 stress period, the EUR/USD basis reached -364 basis points. Even a move halfway to those extremes would signal significant dollar funding stress.
This creates secondary effects for AI companies through their customers and partners. If European or Asian enterprises face higher dollar funding costs, they reduce spending on cloud services and AI infrastructure - directly hitting the revenue assumptions underlying current AI company valuations.
The Risk-Off Amplification Effect
Perhaps most importantly, USD strength typically coincides with risk-off sentiment as investors flee to safety. Recent analysis from Investing.com notes that "the dollar tends to thrive in this kind of ambient anxiety. This is classic de-leveraging behavior: traders reduce exposure, cross-margin gets twitchy, and liquidity preferences migrate back toward the greenback."
The AI sector represents the most concentrated speculative position in modern market history. When risk-off dynamics accelerate, the AI trade faces disproportionate selling pressure. Bitcoin's 25% decline while the Magnificent Seven fell only 7% during recent volatility demonstrates how leverage unwinds, but the relatively modest Mag7 decline reflects that serious de-risking hasn't occurred yet.
A sustained USD rally from 100 to 110 would likely trigger exactly this de-leveraging cascade, as:
1. Margin requirements increase on leveraged positions
2. Institutional investors reduce exposure to concentrated bets
3. Mutual funds experience redemptions, forcing the selling
4. Credit concerns emerge, widening spreads further
5. International capital flows reverse toward dollars
This creates a self-reinforcing cycle where USD strength → credit tightening → equity weakness → more USD strength.
The Fed Policy Complication
Your analysis of Fed behavior during tariff shocks identified that the Fed will likely prioritize anchoring inflation expectations over short-term growth concerns. A strong dollar typically coincides with tighter financial conditions, which the Fed would view as doing some of their work for them.
This means that unlike the dot-com crash when the Fed could aggressively cut rates from 6.5% to 1.0%, a USD rally from 100 to 110 would likely see the Fed cutting more slowly or even pausing if they view the dollar strength as appropriate tightening.
Current rates around 4.5% with persistent inflation concerns provide limited ammunition. If the USD rallies to 110 while delivering 50 basis points of credit spread widening, the Fed might cut to 3.5-4.0% at most - nowhere near enough stimulus to offset the tightening from credit markets and currency strength.
This is the 2008 parallel that concerns me. When the housing bubble burst, the Fed's aggressive rate cuts were initially ineffective because credit markets had seized up. The transmission mechanism from Fed policy to the real economy broke down until quantitative easing restored it.
So, Is This Dot-Com 2.0?
My answer: it's worse in some ways, better in others, but ultimately different enough that direct comparisons are misleading. And critically, the USD dynamics create a more dangerous setup than existed in 2000.
Worse:
· 61% higher Buffett Indicator (230% vs 143%)
· Much greater market concentration (30% in the top five companies)
· Broader household exposure (60% vs 40-50% stock ownership)
· Emerging debt financing creating credit risk
· 7x larger potential wealth destruction ($35 trillion vs $5 trillion)
· Less Fed policy space to respond
· USD strength accelerates credit tightening through quantifiable mechanisms
· International revenue exposure amplifies the USD impact on earnings
Better:
· Dramatically superior profitability of market leaders
· Real products generating actual cash flows (not pure speculation)
· Technology with proven immediate utility (not just future promise)
· More regulatory awareness and monitoring
· Stronger corporate balance sheets overall
The USD Wild Card:
· Dot-com crash occurred with a relatively stable dollar (DXY 95-105 range)
· Current setup has dollar at extreme oversold (100) with strong rally potential (110+)
· USD strength historically triggers credit tightening (0.5% spread widening per 1% USD move)
· This transmission mechanism didn't exist in 2000 because debt financing was minimal
· Makes the current situation more dangerous through credit channel amplification
This split between profitable giants and cash-burning startups is key. The profitable AI leaders (Nvidia, Microsoft, Google, Amazon) have genuine, sustainable businesses. They can weather a correction and likely recover, similar to how Amazon eventually delivered 50x returns to investors who bought at the 2000 peak and held for 25 years.
The unprofitable AI startups burning cash at 100%+ of revenue with 25-500x revenue multiples look far more like Pets.com and Webvan - likely to fail catastrophically when capital markets tighten.
For gold investors, the critical question isn't whether the AI correction is coming - it clearly is, with 54% of fund managers already acknowledging bubble conditions. The question is what the USD does, because that's what will actually determine gold's price.
The AI-led stock market sell-off would exacerbate the USD-triggered declines in the precious metals, though.
Let me be clear about the scenarios and their gold implications:
Scenario A: USD rallies to 110, AI correction unfolds (75% probability)
· Gold DECLINES 10-20%
· Target: $2,500-2,850
· AI bubble burst doesn't help gold because USD strength dominates
Scenario B: USD remains weak, AI correction unfolds (15% probability)
· Gold RISES as a safe haven
· Target: $3,100-3,400
· Requires my USD analysis to be wrong
Scenario C: USD rallies, AI bubble continues (10% probability)
· Gold DECLINES 10-20%
· Target: $2,500-2,850
· The decline takes longer and is not as sharp as in scenario A
While gold declines, the same is likely for silver and mining stocks, but they are all likely to recover before stocks find their bottom – just like what we saw in 2008 and 2020. And then investors will recall that there are 100 reasons for silver to move higher and the white metals – along with the rest of the precious metals sector will move higher – likely much higher.
[Timing-related analysis follows in today’s Gold Trading Alert - the full version of this analysis]
Thank you for reading today’s free analysis.
Przemyslaw K. Radomski, CFA
Founder
Golden Meadow®