AI boom echoes the dotcom bubble, but not in the way you think

After the dotcom bubble burst, a glut of infrastructure looked like a mistake, but it laid the foundation for the internet’s eventual growth
Today’s AI investments may lower the cost of computing power and provide the basis for future breakthroughs, even if they precede a market correction

Jeffrey Wu

Published: 8:30pm, 5 Mar 2024

Why you can trust SCMP

As the sun rises on a new digital era, long shadows are being cast across a landscape dominated by giants – few more towering than Nvidia, whose market capitalisation has exceeded the entire Chinese stock market, as represented by H shares. This moment, emblematic of the burgeoning AI revolution, prompts us to ask: are we witnessing the inflation of a bubble, or the foundation of a transformative cycle?

The internet’s evolutionary leap from mere information retrieval to inference – as even Google seeks to provide direct answers, instead of offering lists of links – signals a profound shift in our digital interaction and underscores the metamorphosis of search technology from a passive tool into an active assistant in deciphering the world around us. Yet, as we marvel at today’s technological feats, we must also question where the future of value lies in the AI domain.

The audacity of tech luminaries like SoftBank’s Masayoshi Son, who is seeking up to US$100 billion for a chip venture, and OpenAI’s Sam Altman, who hopes to raise a staggering US$7 trillion for an AI initiative, illustrates the scale of the ambition driving the current wave of innovation. Alibaba’s latest US$1 billion bet on Moonshot AI, a large language model start-up, further reflects the belief in AI’s potential to redefine industries and societies.

These figures aren’t just investments; they are bets on a future where AI underpins every facet of our lives. Yet, as history shows, the timing of value capture in technological innovations is notoriously uncertain.

The dotcom bubble of the late 1990s is often remembered for the dizzying rises and falls, amid sky-high valuations of internet start-ups and a frenzied rush of venture capital investments. However, a critical aspect of this period that is often overshadowed by the spectacular demise of companies like Pets.com and Webvan is the phenomenon of infrastructure oversupply, particularly in the telecommunications sector.

02:38

Apple supplier Foxconn to build ‘AI factories’ using US hardware leader Nvidia’s chips and software

At the time, telecoms titans like Nortel, WorldCom and Cisco were at the forefront of building out internet infrastructure, in anticipation of an explosion of digital activity. Driven by the belief that the internet would radically transform every aspect of our lives, telecoms companies proceeded to collectively raise US$1.6 trillion on Wall Street.

The result? More than 128 million km of fibre optic cable was installed, representing 76 per cent of the total digital wiring base in the United States then, such that broadband capacity would be underused for years.

The current AI boom is not a perfect echo of the dotcom era. However, there still lies a possibility that this boom may balloon into a bubble, similarly outstripping immediate demand. From this perspective, Nvidia’s astronomical data centre revenue growth might not reflect genuine AI demand but rather the competition among governments and tech giants for strategic assets amid chip shortages.

Regardless, an AI bubble might not be a wholly negative development. Looking back on the aftermath of the burst dotcom bubble, although the overinvestment in fibre optic cable was initially seen as a folly, it actually laid the foundation for the internet’s explosive growth. Similarly, today’s AI investments will lower the cost of computing power and lay the groundwork for future breakthroughs, even if they precede a market correction.

As such, the true capture of value often trails technological advancements, for it emerges from the applications and services built on the infrastructure laid. Giants like Facebook and Netflix epitomise this trend, having amassed vast enterprise value by capitalising on the social networking and streaming landscapes made possible by the underlying internet infrastructure.

It’s a pattern seen throughout tech history: the lasting winners are those who best understand and serve their customers, not necessarily those who built the underlying technology. Tomorrow’s competitive edge will probably rest on mastering distribution and leveraging proprietary data to boost user experience and demand.

05:03

How does China’s AI stack up against ChatGPT?

Despite the sensational headlines and sky-high valuations, it’s imperative to recognise that we might not be at the peak of AI development. Compared with the dotcom era, when Cisco’s price-to-earnings ratio soared to 201 times, Nvidia’s current ratio of around 65 times suggests that, while enthusiasm is high, we are still in the early stages of AI’s market impact.

Yet, as we navigate the initial exuberance and approach the “trough of disillusionment”, where mismatches between supply and demand become apparent, it’s crucial to maintain a balanced perspective.

The timing of value capture remains elusive, even though the number of S&P 500 companies mentioning AI reached a new high during fourth-quarter conference calls, and artificial intelligence continues to hold promise for technological advancements across domains from healthcare and education to transport and entertainment.

In fact, some impact is already being felt. Swedish fintech company Klarna’s AI assistant, powered by OpenAI, managed to handle 2.3 million chats, or two-thirds of the company’s customer service chats, in a month, and is doing the work of 700 full-time agents. Meanwhile, Baidu’s Ernie Bot serves 26,000 enterprises monthly and handles more than 50 million queries daily – although net income dropped 48 per cent year on year partly due to the company’s hefty AI investments.

As we stand at this crossroads, it’s essential to understand that the hype surrounding AI could be both a beacon and a warning. Lessons from the past teach us that true innovation will endure, even if it may take time to emerge from the rubble of previous missteps.

Jeffrey Wu is a director at MindWorks Capital, a leading Hong Kong-headquartered venture capital firm specialising in technology investment across Greater China and Southeast Asia