Is the AI Trade a Bubble? History Has a Complicated Answer

Moody's stripped the US of its last AAA rating. Markets shrugged – mostly.

The quick scan: Tuesday was quieter than it had any right to be. Moody's downgraded US sovereign debt to Aa1 on Friday night – stripping the country of its last top-tier credit rating. Bond yields ticked higher, oil pushed above $102, and markets opened cautiously. By the close, the S&P 500 had barely moved and the session that could have been ugly turned out to be merely unsettled.

S&P 500: -0.07%, 7,403.05 – Essentially flat; the Moody's reaction was muted, though the 10-year Treasury yield climbed back toward 4.5% during the session
Dow Jones: +0.32%, 49,686.12 – Recovered some of Friday's losses; Home Depot beat on earnings but warned on full-year outlook amid oil-driven cost pressures
NASDAQ: -0.51%, 26,090.73 – Tech retreated modestly; Nvidia slipped ahead of its earnings report on Wednesday, which markets are treating as the week's main event.

What's driving it: This is the third US credit downgrade – after S&P in 2011 and Fitch in 2023 – and the market reaction followed the same pattern: initial alarm, then a fairly rapid return to business as usual. The 10-year yield rose to 4.46% before settling, well below the recent high of 4.59%. The more significant backdrop remains unchanged: oil above $100, two consecutive hot inflation prints, rate hikes back in the conversation, and Nvidia's earnings on Wednesday carrying outsized weight for the broader market. Iran negotiations remain at an impasse.

Bottom line: A US credit downgrade used to rattle portfolios for weeks. Now it's a Tuesday. That's either remarkable market resilience or investors becoming dangerously comfortable with the US fiscal trajectory – probably some of both. For L-Plate Retirees with bond holdings, rising long-term yields mean prices on existing bonds have fallen. If you're drawing down, it's worth checking whether your bond duration still fits this environment.

The AI Trade Just Entered Some Very Strange Company

The scoop: In the mid-1940s, the ballpoint pen was going to change everything. And it did – eventually. But first, it inflated a stock market bubble.

The Reynolds International Pen Company went public in late 1945 at $0.75 per share and quickly surged to $550 per share as investors piled in on the promise of a revolutionary new writing instrument. Within two years, the stock had collapsed to under $3. The ballpoint pen went on to become one of the most ubiquitous objects in human history. Reynolds International did not.

That distinction – between the technology being real and the investment being rational – is at the heart of what a new analysis of the Nasdaq 100's AI-driven surge is asking us to consider.

Where the AI trade sits in history

Ben Carlson of Ritholtz Wealth Management has been tracking the Nasdaq 100's 10-year return against the great market booms of the past century. The number is 640%-plus – and it has now surpassed Japan's Nikkei in the 1980s, the Dow Jones in the Roaring '20s, and the S&P 500's postwar surge in the 1950s. Only the 1990s Nasdaq run – the internet bubble – still sits ahead. And that one did not end with a soft landing.

A 2018 paper published in Marketing Science studied 51 major innovations between 1825 and 2000 and found bubbles in 37 of them – roughly 73%. The list is a museum of technologies that transformed everyday life: steam engines, telegraphs, automobiles, radio, television, personal computers, mobile phones, the internet. And yes, ballpoint pens.

The conclusion the paper draws is not comforting: markets have a long and consistent history of correctly identifying transformative technologies and then wildly mispricing the companies built around them.

Why this time might be different – and why it might not be

The bull case for AI is not hard to make. The technology is real. The productivity gains are measurable. The list of industries it will touch – finance, healthcare, logistics, legal services, education, software development – is genuinely comprehensive. Nvidia's revenue and earnings have not been works of fiction.

The bear case is more subtle. It isn't that AI doesn't work. It's that the market may have already priced in a version of the future that won't arrive on the expected timeline, at the expected scale, or with the expected winners.

The 1990s internet comparison is instructive here. The internet was not a bubble in the sense of being a fiction. It was real, it was transformative, and the companies that survived – Amazon, Google – became among the most valuable in history. But the vast majority of internet companies that IPO'd between 1998 and 2000 went to zero. The technology worked. Most of the investment did not.

The mechanism the Marketing Science paper identifies is useful: these technologies tend to be radical, highly visible, and capable of spawning other businesses – which makes them almost impossible to value in real time. Chips make models possible. Models make software tools possible. Software tools make automation possible. By the time investors try to price the whole chain, they are not analysing a company – they are betting on a civilisational shift.

The 1920s parallel

The Roaring '20s comparison tends to get deployed lazily – as shorthand for irrational exuberance. But the detail matters. The 1920s were a genuine technological transformation. Automobiles, motion pictures, refrigeration, airplanes, and radio all produced bubbles peaking around 1928–1929. The crash wasn't a verdict on the technologies. It was a verdict on the prices investors had paid for them relative to near-term earnings.

Japan's 1980s boom sits in a different category. That wasn't primarily a technology story – it was an asset bubble built on easy credit and the belief that Japan's industrial model was permanently superior. When that belief cracked, the Nikkei dropped 80% and took three decades to recover. The comparison to Japan is probably the one that should make investors most uncomfortable about elements of today's market that have less to do with AI and more to do with general asset price inflation.

What investors tend to get wrong

The mistake in the 1990s, as the article puts it, was not believing in the internet. The mistake was believing every internet stock could become the internet.

That framing applies directly to the current moment. The mistake is not believing in AI. The risk is assuming that every company claiming an AI strategy is worth the premium the market has given it – or that the handful of clear winners at the infrastructure level (chips, cloud, model providers) will sustain valuations that require a decade of perfect execution to justify.

The Nasdaq 100's 640% decade-long return is the product of genuine earnings growth. That's different from Japan in the 1980s, which was largely multiple expansion on flat-to-modest earnings. But at a certain point, even genuine earnings growth can be correctly anticipated and then overpriced. That's where historical pattern recognition becomes useful – not to call the exact top, but to stay honest about what's being paid for.

Actionable Takeaways for L-Plate Retirees:

• Distinguish between the technology and the investment. Ballpoint pens were real. So was the internet. So is AI. The question is not whether the technology works – it's whether the prices being paid today accurately reflect the earnings those companies will generate over the next decade. Those are different questions, and the answer to the first one tells you nothing about the answer to the second.

• Be wary of concentration. The Nasdaq 100's 640% return has been driven by a small number of very large companies. If your portfolio is heavily weighted toward that index – through ETFs, super/CPF investments, or directly held stocks – you have more concentration risk than the diversification numbers suggest.

• History doesn't say the bubble bursts tomorrow. The 1990s internet run went from 1995 to 2000 – five years of gains that rewarded people who stayed in for a long time before punishing them. Calling the top is not the point. The point is sizing your exposure appropriately so that a correction – whenever it arrives – is something your retirement plan can absorb rather than something that rewrites it.

• The companies that survive technology bubbles tend to be the ones with real earnings. Amazon and Google survived the dot-com bust because they had businesses that could generate cash. The companies that went to zero were mostly burning money in pursuit of scale. Apply the same filter today: which AI companies have real, growing, recurring revenue, and which are still largely a story?

• Don't time the market – but do rebalance. If your AI or tech exposure has grown significantly due to price appreciation, rebalancing to your target allocation is not a bet against AI. It's basic portfolio hygiene. Letting winners run until they dominate is how people end up with more risk than they planned for.

• The parallel that should concern retirees most is not the crash – it's Japan. A decade-long period of flat-to-negative returns from a market that was overvalued at its peak is more damaging to a retirement portfolio than a sharp crash followed by a recovery. Japan's Nikkei peaked in 1989 and didn't recover to that level for 35 years. For someone who retired at the peak, that time horizon was their entire retirement.

Your Turn:
When you look at your own portfolio today, do you have a clear sense of how much of your exposure is tied to AI and technology – directly or indirectly through index funds – and whether that's the allocation you'd have chosen deliberately?
The ballpoint pen story is a useful reminder that being right about the technology and being right about the investment are two completely different things. Has there been a moment in your own investing life where you experienced that gap?
History's technology bubbles tend to punish the latecomers more than the early adopters. Where do you think we are in the AI cycle – early innings, middle, or somewhere in the crowd that's arrived after the best seats were taken?

👉 Hit reply and share your thoughts – your answers could inspire fellow readers in future issues.

Writing this takes considerably more time than drinking coffee. If you'd like to help balance that equation, consider supporting L-Plate Retiree on Ko-fi.

Resources:

The Next Level Options (NLOMBA) course is a solid, all-in-one roadmap for mastering options investing. You’ll learn what options really are, how to invest in different market conditions, and how to pick strong companies using Buffett-inspired fundamentals.

Inside, the lessons walk you step-by-step through strategies like BOSS and Strategy X, so you’re not guessing – you’re following a proven structure that helps you invest with clarity and confidence.

What to see everything that’s included?
👉 Check out the Options Workshop