Kong CEO Augusto “Aghi” Marietti believes the current AI bubble might indeed pop, but he asserts that the massive hyperscale infrastructure being built now is not wasted capital; instead, it’s a foundational, long-term investment critical for the inevitable future of artificial intelligence, drawing parallels to the essential US railroad expansion of the 19th century.
The conversation surrounding artificial intelligence is often framed by extremes: unprecedented opportunity or an impending market correction. At the heart of this debate lies the enormous capital expenditure, or capex, being poured into building out the foundational infrastructure for AI. Augusto “Aghi” Marietti, cofounder and CEO of Kong, offers a nuanced perspective that acknowledges the potential for an “AI bubble” while staunchly defending the necessity and long-term value of the current hyperscaling efforts.
The Unavoidable Investment: A New Builders Era
Marietti characterizes the present moment as a “new builders era” where significant capital deployment is not just happening, but is genuinely needed to enable the full potential of the AI era. This isn’t just speculative spending; it’s laying the groundwork for what many believe will be a transformative technological revolution. The scale of investment is staggering, with major players like Amazon, Microsoft, Meta, and Google projected to spend an estimated $320 billion on capex, primarily for AI-related needs, according to a Business Insider analysis. These expenditures are constructing the physical and digital backbone—data centers, specialized hardware, and network capabilities—that future AI applications will demand.
Energy: The Looming Bottleneck for AI Growth
Despite the optimism surrounding AI’s future, Marietti points to a critical challenge: energy. He highlights that energy-related issues are likely to be the primary bottleneck stunting AI growth. The demand for power from large data centers, particularly for energy-intensive GPUs, is so immense that some companies are developing self-contained power supplies. Marietti bluntly states, “We don’t have the energy we need to power all the GPUs in the following year.”
This escalating demand for power underscores the physical constraints facing even the most ambitious technological advancements. According to the International Energy Agency, estimates suggest that data centers, largely driven by AI’s demands, could consume a significant and rapidly increasing share of global electricity, intensifying calls for energy-efficient hardware and renewable solutions. This dynamic signals that energy innovation or acquisition will be a key differentiator for AI leaders in the years to come.
Wall Street’s Jitters vs. The Optimist’s Historical Parallel
The massive capex spending craze by leading AI startups and Big Tech companies has undoubtedly fueled “bubble talk” on Wall Street. Even OpenAI CEO Sam Altman acknowledged in August that AI could be in a bubble phase, a sentiment echoed by economists who suggest current capex levels are propping up the entire US economy. Wall Street analysts have repeatedly questioned the pace and sustainability of these investments, drawing comparisons to past tech booms and busts, as noted in a recent report from Reuters.
However, Marietti, much like Altman, draws a compelling historical comparison to the 19th-century US railroad build-out. He argues that while some railroads were deployed “ahead of time,” they were all eventually used, fundamentally transforming the economy. Marietti posits a similar trajectory for AI infrastructure: “I think in AI, we’re just deploying ahead of time, and eventually something will blow up for a little bit, but we would eventually need the infrastructure that we’re deploying anyways.” This perspective suggests that any short-term market correction, or “down moment,” will not diminish the long-term strategic value of the infrastructure being established.
Key Takeaways from the Railroad Analogy:
- Early Deployment: Infrastructure is often built before its full utility is immediately apparent.
- Eventual Utilization: Over time, the economic value catches up to the infrastructure investment, proving its long-term necessity.
- Transformative Impact: Like railroads fundamentally reshaped commerce and society, AI is expected to revolutionize economic activity across sectors.
OpenAI President Greg Brockman further fuels the long-term demand narrative, suggesting that in the near future, every person might desire their own GPU. Such a scenario would necessitate an even greater expansion of the very infrastructure Marietti champions, solidifying its indispensable role.
What This Means for Investors: Long-Term Value Amidst Volatility
For investors navigating the volatile AI market, Marietti’s insights offer a crucial long-term perspective. While the immediate concerns about a potential bubble are valid, the underlying argument is that the infrastructure being built now represents indispensable capital assets. A market downturn might impact valuations, but the physical and digital foundations for AI will remain and become increasingly critical for future economic growth.
Consider the following for your investment strategy:
- Focus on Foundational Companies: Companies providing core AI infrastructure, such as chip manufacturers, data center operators, and specialized cloud providers, may represent more stable long-term plays.
- Monitor Energy Solutions: Given the energy bottleneck, companies innovating in or securing sustainable power for AI could see significant strategic advantage and become key investment targets.
- Historical Precedent: The railroad analogy encourages a patient view, recognizing that transformative technologies often involve initial over-investment followed by sustained utility and immense societal benefit.
This “builders era” demands foresight. While short-term market fluctuations are inevitable, the enduring need for robust AI infrastructure suggests that today’s significant investments, even if temporarily overvalued, are likely to yield substantial returns over the decades to come.
