🤔 AI and the early internet

What the early internet boom can tell us about AI

I’ll never forget my first visit to Harvard Business School back in 1997. I attended the annual Cyberposium conference organised by the school’s Tech Club. Apple’s former CEO John Sculley was there — he and I spoke on a panel together.

This came back to me earlier this week when I participated in another panel discussion at the HBS Leading with AI conference alongside EV member and HBS Professor Karim Lakhani,

Ethan Mollick

, Mary Erdoes from JPMorgan, investor Marcelo Claure, and others. During the panel, I shared how - as an investor - I think about AI as a general-purpose technology. Although the recording of the event isn’t available yet, I wanted to share with EV members some of the notes I made while preparing for the panel.

I caught up with Ethan Mollick

I’m flying back to London today, enjoy the notes!

---

Where are we?

Preparing for this event, I couldn’t help but draw parallels between the early days of the internet and the AI revolution we’re witnessing today. The lessons from the internet boom are invaluable as we navigate this new frontier.

I first attended the HBS Cyberposium in 1997, a time when the list of the largest US companies was dominated by the likes of Coca-Cola, Raytheon, Loews, PepsiCo, Walt Disney and AIG. The only tech firms present were Intel and IBM, even though Netscape had gone public just a year prior. Fast forward to 1998, four years into the internet wave, and telecommunications companies still overshadowed the emerging internet giants. Amazon’s market cap was a mere $7 billion compared to AT&T’s $100 billion, Cisco’s $78 billion, and Paramount's impressive $25 billion. Google, now a household name, was valued at less than $5 million.

By the turn of 1999, Microsoft had claimed the top spot, with Intel, IBM and Cisco also ranking among the tech elite. However, the top 20 list still included traditional giants like Home Depot, AIG and AT&T. Sun Microsystems, the epitome of the “dot in dotcom,” saw its stock price skyrocket from $25 to $310 between January 1997 and December 1999, reaching a peak market capitalization of around $200 billion in 2000 before eventually being sold for just $7 billion.

From 1996 to 2001, Paramount Global and Microsoft experienced similar growth at 44% per annum, while AT&T and Verizon saw 33% and 36% growth, respectively. Amazon’s stock grew by an impressive 50%, while Intel experienced a more modest 15% growth. Netscape, once a pioneer, had all but disappeared.

Now, let’s fast forward to the present day. In April 2023, well into the genAI boom, Nvidia’s PE ratio of 143 seemed expensive, but hindsight reveals it was a bargain. The stock price has since doubled, and the PE ratio has fallen to 75. Interestingly, the infrastructure companies of the early internet era, such as Cisco, Telcos and even Sun, either failed or substantially underperformed over the next two decades. Astonishingly, AT&T’s share price has risen by less than 15% in 30 years.

Today, revenues are primarily accruing to semiconductor manufacturers, hyperscalers and app vendors like OpenAI. But the question remains: is AI following a similar trajectory to the early days of the internet, or is it fundamentally different?

Is AI different from the early days of the internet? 

AI technology appears to be a disruptive innovation that sustains incumbents. We’ve seen incumbents, major tech companies, operate the cloud services necessary to train and deploy LLMs. They also have the resources and user base to integrate AI capabilities into their existing products and services, further strengthening their position. However, it’s important to note that the AI landscape is still in its early stages, and the long-term impact on incumbents remains to be seen. Four years into the boom, some of the big winners of the internet era such as Meta, eBay, Tencent, Netflix, Salesforce, Pinduoduo, PayPal and Zoom were nowhere to be seen.