Understanding Economic Bubbles

My latest in Project Syndicate. I argue that relative to the dot.com era, the productivity upside is more muted this time, and the risks greater: 1. As the dot-com bubble burst, IT was already delivering a visible productivity boom; by contrast, today’s firm-level AI adoption is already slipping. 2. The dot-com boom left durable assets like fiber that still carry traffic, whereas AI depends on rapidly obsolescing chips and servers—creating a capex treadmill that needs constant reinvestment. 3. Unlike the surplus era of the late 1990s, today’s AI boom sits on top of large deficits, rising debt, and heavy interest bills, competing with clean energy, defense, and housing for scarce savings. 4. With a large existing debt stock and positive real rates, interest costs rise quickly, crowding out welfare and public services. If AI’s payoff is slow or modest, more resources flow to bondholders instead of Social Security, health care, and core services, especially in a downturn. 5. The dot-com bust mainly hit equity investors; today’s AI build-out is increasingly credit-financed, so if AI revenues disappoint, stress will show up in credit markets and on bank/insurer balance sheets, raising systemic risk.

It is (literally) the trillion-dollar question. Are we in an AI bubble? And is there an analogy to the dot-com bust? 𝗧𝗵𝗲𝘀𝗲 𝗮𝗿𝗲 𝘁𝗵𝗲 𝗔𝗜 𝗻𝘂𝗺𝗯𝗲𝗿𝘀: • Nvidia reached ~$5 trillion market cap in 2025, the fastest value creation for a listed company in history. •Nvidia now makes up ~8% of the S&P 500, the highest single-stock index weight since the 1970s. • AI-linked megacaps generated ~70–80% of U.S. equity gains in 2025. • Citi forecasts cumulative AI infrastructure capex by major tech firms above ~$2.8 trillion through 2029, with ~$1.4 trillion in the US alone. • Roughly half of recent VC tech funding has flowed into AI startups, concentrated in a few companies. 𝗔𝗻𝗱 𝘁𝗵𝗲𝘀𝗲 𝗮𝗿𝗲 𝘁𝗵𝗲 𝗱𝗼𝘁-𝗰𝗼𝗺 𝗻𝘂𝗺𝗯𝗲𝗿𝘀:        • Cisco’s market cap grew from ~$5 billion in 1993 to $555 billion by March 2000 - over 100x in seven years. • At its peak in March 2000, Cisco made up ~3-4% of the S&P 500 • Tech stocks (internet/networking) drove most Nasdaq gains - the index rose ~400% from 1995 to March 2000 peak. • Telecom/internet infrastructure capex hit ~$100–200 billion annually by late 1990s, fueling massive overbuild. • VC funding saw ~40-50% of total dollars into internet/tech startups by 1999, concentrated in e-commerce and networking. 𝗧𝗵𝗲 𝗽𝗮𝗿𝗮𝗹𝗹𝗲𝗹𝘀: • Rapid value creation concentrated in a single infrastructure provider (Cisco then, Nvidia now). • Equity market gains increasingly dependent on a narrow group of megacap stocks tied to one technological narrative. • Massive capex commitments ahead of proven monetization. • VC and private capital flows clustering in one technology theme and consolidating around a small set of perceived winners. • Infrastructure capacity being built faster than adoption can absorb. • Investor expectations extrapolating current growth rates far into the future. 𝗔𝗻𝗱 𝘁𝗵𝗲 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝗰𝗲𝘀: • AI has measurable productivity gains, unlike many dot-com models without revenue. • Nvidia has hardware-software lock-in, where Cisco’s networking gear commoditized quickly. • AI compute demand is concentrated in a few hyperscalers, not fragmented carriers. • Chip shortages are keeping GPU prices high, whereas networking gear saw prices collapse quickly after 2000. • AI adoption spans multiple industries, unlike dot-com’s narrow online commerce focus. • Today’s leaders are profitable incumbents, not unproven startups. • Consumer AI adoption scaled exponentially faster than early internet.   𝗦𝗼, 𝘄𝗵𝗮𝘁 𝗱𝗼 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸 - is this an AI bubble, or are markets simply discounting future returns? Opinions my: own, Graphic sources: WSJ, IMF, P. Kriaris 𝐒𝐮𝐛𝐬𝐜𝐫𝐢𝐛𝐞 𝐭𝐨 𝐦𝐲 𝐧𝐞𝐰𝐬𝐥𝐞𝐭𝐭𝐞𝐫: https://lnkd.in/dkqhnxdg

Some thoughts on the state of the AI Industry today: Hype is omnipresent in the rapidly evolving world of Artificial Intelligence (AI). Every day, new breakthroughs and advancements are touted as the next big thing, promising to revolutionize industries and solve complex problems. However, amidst this excitement lies a significant danger: the risk of being misled by the noise and falling victim to inflated expectations. One of the primary dangers of AI hype is the potential for misallocation of resources. Companies and individuals, driven by the fear of missing out, often invest heavily in AI technologies without fully understanding their capabilities and limitations. This can lead to wasted resources and failed projects. For instance, the AI bubble of the 1980s, known as the "AI Winter," saw massive investments in AI technologies that were not yet mature. Many investors suffered significant financial losses when these technologies failed to deliver on their promises. To avoid falling prey to the hype, it is crucial to filter out the noise and focus on the signal – the true, sustainable advancements in AI. Here are some practical steps to help navigate this landscape: - Do Your Research: Before investing in or adopting any AI technology, conduct thorough research. Understand the technology's underlying principles, its current state of development, and its realistic applications. Be wary of exaggerated claims and seek information from reputable sources. - Look for Proven Use Cases: Focus on AI solutions that have demonstrated success in real-world applications. Case studies and testimonials from credible organizations can provide valuable insights into the technology's effectiveness. - Adopt a Skeptical Mindset: Approach AI innovations with a healthy dose of skepticism. Question the feasibility of grand promises and seek out expert opinions. Remember that if something sounds too good to be true, it probably is. - Learn from History: Historical examples, such as the Dot-Com Bubble and the AI Winter, serve as cautionary tales. During the Dot-Com Bubble of the late 1990s, many internet companies with unsustainable business models received exorbitant valuations, leading to a market crash when reality set in. Similarly, the AI Winter reminds us of the importance of aligning expectations with technological realities. In conclusion, while the potential of AI is immense, it is essential to navigate its landscape with caution. By filtering out the noise and focusing on substantiated advancements, we can harness the true power of AI without falling victim to the dangers of hype. Let's learn from the past and approach the future of AI with informed optimism and strategic discernment.

Irrational Exuberance: It was the Golden Era of Credit with money flowing freely. Direct Lending fund managers took in more and more capital to make loans in increasing volume in greater size. The Golden Era of Credit resulted in Irrational Exuberance. Irrational exuberance describes financial markets where asset prices and underlying fundamentals reflect excessive investor optimism fueled by herd behavior, speculation, and FOMO, rather than rational analysis. The phrase popularized by then-Fed Chair Alan Greenspan referred to the dot.com era, but could have been applied to Direct Lending willingness to leverage software companies. The competitive dynamics in private credit have led to a troubling race to the bottom. Managers are offering additional turns of leverage, tightening spreads by 25 to 50 basis points to win lead arranger status, and structuring PIK-toggle features that defer cash interest when borrowers can’t service their debt. The goal is simple — win the deal, deploy capital, and generate fees. The result is loans underwritten at 10x EBITDA with little to no free cash flow remaining after debt service. Risk management and portfolio management guidelines that should cap single industry concentration limits at ~15% did not apply. Software was the "it-thing.” ARR made software the safest sector ever. Whereas a direct lending manager might limit leverage to 5x debt-to-EBITDA, the software lender was willing to provide 10x or lend on recurring revenues instead. He reasoned that 10x was not excessive, especially since the lender was willing to structure the loan based on forward guidance (pro-forma projections) or LTV on inflated valuations. In fact, 10x could equal 8x or even lower when the investment team made re-occurring expenses look like a one-time adjustment. Yes, it was the Golden Era of Credit. Direct Lenders, on average, have 23%-plus exposure to software. However, these funds on average take 1x of back-leverage, so the investors have 2x the risk or ~46%-plus of their capital invested in highly leveraged software companies. The software finance bubble is bursting before our very eyes; the default cycle will play out in the next 2 years. On CNBC Closing Bell yesterday with Scott Wapner, I was asked very pressing questions. There are no easy answers to his questions, only truthful, insightful, and painful ones that many must now confront (link below). Opportunistic Credit is the flipside to Irrational Exuberance with dislocation, distressed and capital solutions will create a strong vintage of investment opportunities.

Owning companies who struggle to adapt to new competition / technology can hurt but so can paying too much for the winners of tomorrow. It's even possible to do both at the same time. See for example the chart showing the percentage decline in market cap from the dot com bubble peak of Amazon and its competitor electronics retailer Circuit City, who went bust. Also consider: The landline telephone and the ‘brick’ mobile phone. The Polaroid camera and the digital camera. The Walkman, CD player, MP3 player. Look at what Apple eventually did to Nokia (and many others) but Apple still fell sharply when the dot com bubble burst. The typewriter, Lotus 1-2-3, onsite servers. Microsoft still fell sharply during the dot com bust despite word, excel and eventually cloud computing. Newspapers, radio and TV advertising, Yahoo. Google and Meta weren't even listed in 2000. As technology evolves, it creates both winners AND losers. Shares in the losers struggle, shares in the winners often experience bubbles. I think the average investor should be thinking harder about: Who the losers might be from AI powered search and the potential effect on existing online advertising and online shopping business models. It's also worth considering how AI could affect how we book things like holidays and influence where we decide to put our savings or how we choose things like insurance or mortgages. The effect AI could have on office software. Which device(s) will become the main way we interact with AI. The effect self driving electric cars could have. The effect home robots could have. The effect a digital euro etc could have on existing payment providers. Tomorrow's world (remember that show?) might look quite different to today's. Owning the companies who lose out tends to hurt but the eventual winners could also experience bubbles and the accompanying busts. History suggests it's probably wise not to assume today's winners will be tomorrow's winners but also not to ignore valuations on tomorrow's potential winners. One thing that seems clear is that it’s very unlikely that every big company today will win from the new technologies that could disrupt how things are currently done. It's probably also worth imagining that you're a dragon on Dragons Den and asking yourself some of the questions you might ask a relatively new business that came in to pitch for your money. Questions like: "Who are your competitors and what is your competitive advantage?" "How much do you charge compared with your competitors, is it affordable and how much does it cost for you to provide the service?" “What’s the rate of depreciation of your assets?” "How much revenue and profit do you make from the specific product / service you’re pitching today and how do you justify your revenue and profit growth expectations for the next 5 years?” “How do you justify your valuation?” Always ask the uncomfortable questions, no matter the size of the company.

I always laugh when a VC tries to convince people that something is not a bubble. VCs make risky investments in untested, cash-burning companies, hoping for a big payoff when they go public. Thus, VCs have good reason to convince others that something like AI isn't a bubble. A VC founder telling you why something in which they've invested is absolutely not a bubble is like your friend telling you to share in their can't-miss opportunity with an "investment" in Labubus (or Pogs, Beanie Babies, or NFTs). Maybe this fellow is right; after all, none of us has a crystal ball. But there is certainly reason for concern. By one estimate, by the end of 2025, "Meta, Amazon, Microsoft, Google and Tesla will have spent over $560 billion in capital expenditures on AI in the last two years, all to make around $35 billion." (https://lnkd.in/gstGag7G) That wouldn't be such a bad thing in a nascent market if, in fact, these investments were paying off for the companies spending that money, but it's been suggested that as many as 95% of AI projects show no return (https://lnkd.in/gXcJ84dw). None of this means AI won't be big; after all, the internet changed the world, yet nascent web companies still suffered a painful bubble burst because of too much early investment and optimism. Just look at Amazon, if you want the perfect example. The company's stock fell 94% between 1999 and 2001, and it took almost a decade to recover. And for every Amazon that thrived after the dot-com bubble, dozens vanished. As tech companies race to invest more in AI to try to gain early market share (and bump competitors out of the market), we're seeing an unsustainable level of investment in the hope each company will be one of the last ones standing. For example, OpenAI currently has a valuation of around $500 billion, a 67% increase from earlier in 2025 (https://lnkd.in/gHUzHkEf). The company is private, so public financials are not available, but CEO Sam Altman had a rather testy response to being asked how a company with $13 billion in revenue can make $1.4 trillion of spend commitments (https://lnkd.in/gp96YT-9). Perhaps OpenAI will be a long-term winner, but the very definition of a bubble is when valuations explode due to investors pouring vast sums of cash into companies with very little revenue and massive losses, in the hope that they will grow astronomically and deliver profits in the future. AI will be big. Some AI companies will be massive success stories. That does not mean we're not in a bubble. I think we are, but time will prove whether I or the VC founder is correct. https://lnkd.in/gWWaHZCX

Almost every great infrastructure revolution in history was funded by people who lost their shirts. This is not a footnote. It is the mechanism. Carlota Perez, in Technological Revolutions and Financial Capital, identified the pattern across two centuries: a transformative technology emerges, financial capital floods in, overbuilding follows, a crash clears the field and then the real economy quietly gets to work on the rubble. The hype was never wrong about the technology. It was wrong about who would profit from it. The railway mania of the 1840s funded the physical backbone of industrial Britain and America. American railroads went bankrupt repeatedly across the nineteenth century, their capital written down in successive reorganisations. JP Morgan consolidated the wreckage. The dot-com bubble funded the fibre optic backbone of the modern internet. Over 90 percent of it sat dark in 2002. Within a few years it was the cheapest and most consequential infrastructure on earth, bought from bankruptcy for cents on the dollar. Others built empires on it. The original investors did not. Hyman Minsky spent his career answering the question: Why does this keep happening? His insight was not that markets go mad. It’s that each stage of the cycle feels entirely rational to the people inside it. Stability breeds instability: good returns attract capital, inflated capital inflates prices, inflated prices attract more capital until the structure collapses. The mania doesn’t announce itself. It is only visible in retrospect, which is precisely why it recurs. Charles Kindleberger, building on Minsky, showed that this is not a quirk of particular circumstances. He documented the same dynamic across five centuries of commercial history. Infrastructure is uniquely vulnerable: high fixed costs, low marginal costs and network effects make every entrant believe they will be the last one standing. The anatomy never changes because human nature never changes. Which brings us to today. The AI infrastructure buildout including data centres, chips, power capacity and cooling systems is running at a scale that makes the fibre bubble look modest. Billions are being spent on this infrastructure, eating large chunks out of the cash flows of the companies that are investing. Every individual decision looks defensible. The infrastructure, once built, is permanent. William Janeway, in Doing Capitalism in the Innovation Economy, makes the argument that speculative excess is not a flaw in the system. It is the only mechanism capitalism has reliably produced for funding infrastructure at the scale transformation demands. It takes a mania. History suggests that the wreckage always comes. The question for the investing companies is not whether you are in a mania. The question is whether you will be positioned to build on what it leaves behind.

If every investor in your WhatsApp group sounds like a genius... ...you may be looking at a market cycle rather than investing skill. A few years ago, capital felt almost limitless. Funding rounds were closing faster, valuations kept rising, and nearly every portfolio seemed to have a breakout success story. In that environment, it's easy to confuse momentum with judgment. The numbers are telling. Global venture funding crossed $680 billion in 2021, making it one of the strongest years in the industry's history. Capital was abundant, optimism was high, and growth often mattered more than efficiency. Then the cycle changed. Funding slowed. Liquidity tightened. Investors became selective. Suddenly, many assumptions that looked obvious in a rising market were being tested in real time. That's why I often come back to a line from Naval Ravikant: "Smart money is just dumb money that's been through a crash." Not because crashes create smarter investors. But because they reveal what bull markets often hide. The dot-com crash erased nearly 78% of the Nasdaq's value. During the 2008 financial crisis, Sequoia Capital's message to founders was simple: "R.I.P. Good Times." Different cycles. Same lesson. The investors who emerged stronger weren't necessarily the most intelligent. They were the ones who developed pattern recognition through uncertainty, mistakes, and changing market conditions. What's particularly interesting is that the same applies to founders. Growth markets reward ambition. Tough markets reward discipline. The best investors aren't asking, "What worked last cycle?" They're asking, "What still works when the cycle changes?" And increasingly, that distinction may be the real definition of smart money. Do you think experience is still the biggest edge in investing? #VentureCapital #PrivateMarkets #Fundraising #CapitalMarkets #Investing #StartupFunding #InvestorRelations

Twenty-five years ago, I gave a lecture warning that the Internet hadn’t changed business; it had merely exposed its fundamentals. The title was “Naked Business Models.” It was August 2000. The dot-com bubble had just begun to burst. I argued that companies were spending vast sums to attract customers with no credible path to profit. Now, a quarter century later, I’m seeing the same pattern unfold with artificial intelligence. The metrics have shifted from “eyeballs” and “clicks” to “tokens” and “model queries,” but the mindset remains familiar: growth first, value later. Cheap capital, boundless optimism, and seductive storytelling are once again distorting our perception of technology. The lesson from 2000 still applies: innovation doesn’t repeal economic laws. In this article, I reflect on what the dot-com crash taught us, and why AI’s extraordinary promise will only endure if we remember what truly creates value. #AI #Innovation #Entrepreneurship #Technology #Startups #BusinessStrategy #DigitalTransformation #HigherEducation

I launched my first company during the 2000 dot-com crash. With no funding, no safety net, and tons of naiveté, we built a profitable 8-figure business that thrived while others were still collapsing. That trial-by-fire experience shaped how I've led every PE-backed turnaround since. Many household names were born in bad times: Microsoft (1975 recession), FedEx (1973 oil crisis), Netflix, Airbnb, Uber, etc. As brands try to navigate the current economic chaos, a few timely lessons and opportunities: Efficiency DNA When we acquired struggling businesses during my PE days, we'd always find bloated cost structures that had accumulated during boom times. Teams had lost their hunger. Expenses lacked scrutiny. Contrast this with companies born during downturns. They are forced to develop an almost spiritual relationship with capital efficiency. Build lean operations into your DNA. It's less likely (but certainly not impossible) to develop the expensive habits that later require painful corrections. Value Proposition Clarity In prosperous times, marginal business ideas can gain traction through sheer market momentum. During downturns, only truly compelling value propositions survive. (I always loved the Buffet quote "Only when the tide goes out do you discover who's been swimming naked.") You can't hide behind vague platitudes when consumers scrutinize every dollar. Your value proposition must be crystal clear and irrefutable. Talent Arbitrage A byproduct of downturns can be exceptional people being on the market. I've watched startups founded during economic contractions assemble dream teams that would have normally been unattainable. Companies with the vision and resources can build big advantages. The Counterintuitive Marketing Advantage Economic downturns often deliver lower CACs. Why? As competitors pull back marketing spend (which I don't recommend, as it often is an opportunity to take share), advertising costs can drop. Media consumption typically increases. And consumers become more deliberate about discovering value-oriented offerings. Warby Parker exploited this dynamic during the Great Recession. When consumers were particularly price-sensitive, they offered stylish glasses at a fraction of incumbent prices and efficiently acquired customers. Constraints breed creativity. The brands that survive during this period of tariff uncertainty won't succeed despite the challenges – they'll succeed because of them.