Steve Case: What the AI Generation Can Learn from the Dotcom Bust

This March marked the 25th anniversary of the dotcom bust, a moment that serves as both a cautionary tale and a source of valuable lessons as we navigate the current artificial intelligence (AI) boom. AI, much like the internet in the 1990s, is being hailed as the next transformative technology, with massive investment and sky-high valuations fueling excitement—and speculation. But if history is any guide, it may take longer than many expect for AI to reach its full potential, and there will be both winners and losers along the way.

As the co-founder of AOL, I  had a front row seat to the Internet revolution. And I see several key lessons from that era that are relevant to the AI boom today. The lessons from the Internet and how they might inform the AI revolution are starting to become clear.  

Transformative technologies often take time to mature, but when the right conditions align, adoption can accelerate at an astonishing pace.  Policymakers must strike a delicate balance between fostering innovation and ensuring responsible oversight, as premature or excessive regulation can stifle progress. Meanwhile, market dynamics usually favor early adopters, but industry concentration in AI risks limiting opportunities for young startups. Finally, hype cycles are inevitable—overinvestment leads to corrections, but true breakthroughs endure. Together, these insights reinforce the need for strategic and surgical regulation, open access, and a steady hand in shaping AI’s future.

First, technology adoption takes time—but when it clicks, it moves fast. While AI may feel like an overnight success, its roots go back more than 70 years. The Internet too, has been in development for more than two decades when AOL became the first Internet company to go public, in 1992.

AI has arguably taken longer to fully develop than the internet did, but because the world is now more interconnected, when ChatGPT launched it was able to get 100 million downloads in just three months. By comparison, it took AOL nine years to reach 1 million subscribers. AOL’s first funding round in 1985 was just $1 million; today, some AI startups are raising $1 billion, with not much more than an intriguing idea and an experienced team.

Second, AI’s rapid adoption is forcing policymakers to engage earlier than they did with the Internet. Yes, the government played a critical role in enabling the Internet to flourish.  Key decisions—such as the breakup of Ma Bell, which spurred competition in telecommunications, and Congress’ decision to open access to the internet to everyone—helped unleash the digital age. But because consumer adoption of the Internet was relatively gradual, the government could initially adopt a “wait and see” attitude, and a “light touch” in terms of regulations, while they monitored how it evolved and impacted our lives. 

But now, there is pressure to move quickly, especially given the “doomsday” scenarios that AI experts are understandably concerned about. So coming to agreement sooner on appropriate guardrails is important. At the same time, we can’t let concerns about what might go wrong lead to stifling regulations that could hobble the many societal benefits of AI, or put at risk U.S. leadership of this highly strategic technology. It will be tricky, but we need to strike the right balance between responsible oversight and fostering continued rapid innovation.

Third, during the internet boom, major corporations experimented with the nascent technology, but few took it seriously at first. Many of America’s largest and most valuable companies at the time—like GE and AT&T—dabbled in online strategies, but were not sure it would have broad appeal. Their decision to largely stay on the sidelines created an amazing opportunity for a new generation of companies to emerge, including AOL, Yahoo, and many others. But AI is different. The biggest tech companies, Microsoft, Google, Meta. Amazon, and others—are all in, and there’s now an arms race to dominate AI. This raises concerns for startups. Unlike during the early days of the Internet, the development of foundational AI platforms is largely concentrated in the hands of a few dominant firms. Open-source AI could help increase the likelihood of broader access and innovation. Without it, we risk AI enabling Big Tech to get bigger, and a new generation of disruptive startups could be left behind.

Fourth, hype and FOMO will inevitably lead to market corrections. In the late 1990s, the belief that “the Internet will change everything” drove sky-high valuations. That belief wasn’t wrong—the Internet did change everything. But not overnight, and not for every company. When the bubble burst, many concluded that the Internet was a passing fad. Even within AOL Time Warner there were skeptics. In reality, the shakeout led to hundreds of companies falling by the wayside, but the strongest companies—Google and Amazon among them—really thrived.

AI is experiencing a similar hop and hype cycle. There’s a frenzy to invest early, reminiscent of the dotcom era, as investors don’t want to miss out on what is undoubtedly the next big thing. But as with the Internet, not every AI startup will thrive; indeed, most won’t even survive.

Still, the technological mega trend is real. The Internet led to a paradigm shift which changed all of our lives. AI will, as well. But fasten your seat belt, because if the dotcom bust is any indication, it will undoubtedly be a wild roller coaster ride.