News

Artificial intelligence is inspiring both excitement and unease. Investors are pouring billions into AI ventures, tech giants’ valuations have soared, and governments are racing to regulate. The parallels with earlier episodes of speculative exuberance—railways in the 1840s, dot-coms in the 1990s—are hard to ignore.

In a recent Economics Explored conversation, I spoke with Aman Verjee, General Partner at Practical Venture Capital in Silicon Valley and author of the forthcoming A Brief History of Financial Bubbles. We discussed whether AI is the next great bubble or simply the latest chapter in a long story of technological transformation.

Bubbles: Engines of Progress or Portents of Crisis?

Verjee distinguishes between bubbles that destroy and those that ultimately create. The 1880s Melbourne land boom—fuelled by debt and easy credit—ended in a banking collapse and deep recession. By contrast, the railway mania of mid-19th-century Britain and the dot-com boom of the late 1990s left behind some enduring infrastructure. Investors who bought at the top suffered, but society gained the rail network and the internet economy.

In each case, exuberance mispriced timing rather than potential. The technology did change the world—it just took longer to deliver profits than the market expected. For Verjee, that pattern may be repeating with AI. “It seems like there may be some reset coming where a lot of the companies won’t make it, but there are transformational technologies that are happening, and eventually we’ll get a lot of winners out of the space investing”, he notes.

Is AI the Next Dot-Com bubble?

Today’s “Magnificent Seven” tech firms trade at lofty multiples, and capital is flowing into AI start-ups promising to reinvent every industry. Yet unlike 1999, many leading players already generate substantial earnings. Nvidia sells real chips; Microsoft’s cloud business is profitable. The question is less whether AI is real—it clearly is—than whether valuations reflect sustainable demand.

According to MACROBOND, global artificial intelligence development is advancing rapidly, with the United States maintaining a dominant position. Many of the nation’s largest technology firms are deeply involved in AI innovation, positioning them as major contributors to overall equity market growth (see the chart below). 

Verjee compares AI to railways and the internet: both triggered speculative excess before transforming productivity. Short-term volatility, he argues, doesn’t negate long-term value. Investors may overestimate how quickly profits arrive, but society often underestimates how far the technology will ultimately reach.

AI and the Future of Work

Public anxiety about AI-driven job losses is widespread. Verjee is optimistic: history suggests automation reshapes work rather than eliminates it. “When 98 per cent of people worked on farms,” he reminds us, “the Industrial Revolution didn’t create 98 per cent unemployment.” Instead, labour moved into manufacturing and later into the services sector.

AI is likely to accelerate a similar process. Call-centre roles, data entry, and routine legal work will decline, while demand grows for engineers, analysts, and educators who can integrate new tools. The real challenge is adaptation—ensuring skills, education, and institutions adjust fast enough to capture the gains.

While AI will inevitably disrupt established roles, its deeper economic effect is to reshape decision-making itself. As Agrawal, Gans, and Goldfarb observe in Prediction Machines, AI doesn’t make “thought” cheap—it makes prediction cheap, increasing the value of uniquely human judgment and creativity. Rather than mass unemployment, we should expect the reconfiguration of tasks within jobs: some will be automated, while others will be newly created to complement machines. This transition, however, demands system-level change—redesigning workflows, education, and institutions to capture AI’s productivity potential rather than merely cutting costs.

Policy Matters

Verjee identifies several barriers to productivity growth familiar to Australian audiences. Over-regulation discourages hiring and risk-taking; high taxes and complex compliance sap energy from small firms. Healthcare costs in the United States add thousands of dollars per worker, and strict labour laws make employers reluctant to expand staff.

He also warns that well-intentioned AI regulation could entrench incumbents. State laws requiring small start-ups to document every potential algorithmic bias or “catastrophic risk” impose costs only big firms can bear. A moratorium or phased framework, akin to the hands-off approach that allowed the early internet to flourish under the U.S. Telecommunications Act’s Section 230 protections, might better balance innovation and accountability.

Education reform is another priority. The U.S. and Australia both face shortages of graduates in computer science, engineering, and healthcare. Verjee contrasts America’s laissez-faire attitude—“plenty of actors, not enough coders”—with Asia’s directive approach that channels students toward technical fields. Aligning curricula with labour-market needs is essential if AI is to lift productivity rather than deepen inequality.

Learning from the Past

According to Verjee, history shows that equity-driven bubbles, though messy, can seed lasting progress, while debt-fuelled booms often end in systemic crisis. That distinction offers guidance for policymakers and investors alike. Encouraging venture capital and equity finance spreads risk without imperilling the banking system. The greater danger lies in cheap credit and moral hazard, not in exuberant entrepreneurship.

For governments, the lesson is to foster an environment where innovation thrives but speculation does not distort credit markets. For investors, patience and diversification remain the best defences against hype cycles. And for societies, openness to technological change is key: productivity growth—the foundation of rising living standards—depends on it.

Conclusion

Every era of rapid innovation feels like a bubble while it unfolds. The railways and the internet each produced booms, busts, and eventual transformation. AI will likely follow the same path: excessive optimism now, enormous benefit later.

If policymakers learn from history—promoting flexible education, prudent regulation, and sound macroeconomic settings—the current enthusiasm for AI could mark not the peak of folly but the beginning of another great productivity surge.

Published on 4 November 2025. Adept Economics Director Gene Tunny prepared this article with assistance from Economist Arturo Espinoza. For further information, please get in touch with us via contact@adepteconomics.com.au or by calling us on 1300 169 870.