Key Takeaways
- Identifying the ultimate winners in the AI boom as events unfold is extraordinarily difficult.
- The DeepSeek selloff in 2025 offers useful lessons for investors navigating the AI theme.
- In the last period of technological change, the dot-com boom, there were few winners and many losers.
The latest AI turmoil is a reminder that investing through periods of rapid technological change is inherently messy. Being right about the direction of travel is not the same as being rewarded as an investor. Here patience and a long-term approach beat attempts to pick winners and losers in real time.
The current AI wave shares many of the same characteristics as previous technological revolutions such as the dot-com boom: Genuine innovation, exuberant expectations, and sharp market reactions to incremental developments. Yet it also differs in important ways, not least the sheer scale of capital being deployed and the speed at which new capabilities are emerging, this time around.
Recent market episodes underline these challenges. The software selloff following Anthropic’s latest model release, alongside the earlier DeepSeek volatility in 2025, offers useful lessons for investors navigating the AI theme.
Anthropic Launch Triggers Software Selloff
When Anthropic unveiled its newest model, software stocks came under immediate pressure as investors reassessed competitive dynamics and potential disruption. The iShares Expanded Tech-Software ETF IGM, for example, fell sharply. Many investors who were directionally correct in backing the long-term promise of AI were still caught out. To profit meaningfully from such moves would have required short selling the companies perceived to be most at risk, an approach that is complex, expensive, and unsuitable for most investors.
Looking at European AI-themed funds in the two weeks following the release of the Claude model, the performance gap between the best- and worst-performing strategies—Polar Capital Artificial Intelligence S Acc GBP and Fineco AM MarketVector Artificial Intelligence ESG ETF AI4U I—was 14 percentage points. That is a striking divergence over a very short period and highlights an uncomfortable truth: Even when the structural thesis is sound, implementation risk remains high.
Lessons for AI Investors From the Dot-com Boom
The most recent historical parallel is the late-1990s internet boom, when companies such as Amazon.com were propelled by sheer potential of the internet disrupting traditional bricks and mortar retail outlets. Hindsight can make the investment case appear seductively straightforward.
But it was not. Walmart WMT in the mid-1990s, for example, was one of the largest companies in the world and might have appeared an obvious casualty of the internet retail revolution. Instead, it adapted, invested heavily in logistics and technology, and remains today one of the largest constituents of the S&P 500 index. Structural change does not automatically eliminate incumbents and here adaptation matters.
The same may prove true in AI. Periods of uncertainty can make blanket selling of software companies feel correct. Ultimately, however, outcomes will depend on how businesses adapt—how effectively they integrate AI into their products, pricing models, and competitive positioning. Even amid volatility, there will be winners within software.
Picking AI winners, though, is not easy. Looking back, Amazon’s success can appear inevitable. At the time, it was anything but. For every Amazon.com AMZN, there were dozens early e-commerce pioneers like Boo.com and Kozmo.com that failed to survive. Structural change creates opportunity, but identifying the ultimate beneficiaries of cutting-edge tech as events unfold is extraordinarily difficult.
The DeepSeek episode offers a different but equally instructive lesson. The release of a competitive Chinese large language model prompted investors to question assumptions around brute-force computing power. Hardware producers such as Nvidia NVDA and Broadcom AVGO were hit hard, while many software names proved relatively insulated. The episode underscored how quickly narratives can shift—and how exposed even the largest companies are to changes in technological assumptions.
AI Is a Leveraged Bet With Uncertain Payoffs
Here is where the historical parallels begin to diverge. The dot-com bubble combined genuine technological progress with irrational exuberance. Today’s AI revolution is underpinned by extraordinary capital expenditure. Investment in AI infrastructure alone is expected to rise by around 60% this year, to more than half a trillion dollars. That represents a collective, leveraged bet by some of the world’s largest companies—with highly uncertain payoffs.
The result has been exceptional price volatility among mega-cap stocks—volatility that is unusual in the modern era for companies of this scale. Developments are arriving at dizzying speed, and markets are lurching with each technological advance.
It’s hard with to say confidence what this world-changing technology will ultimately deliver, nor how value will be distributed along the way. History does come to investors’ help here in showing that overconfidence is costly. Picking winners and losers in real time, however tempting, is difficult. A diversified and patient approach—one that acknowledges both the transformative potential of AI and the uncertainty surrounding its commercial outcomes—is likely to serve investors better over the long run.
