The Long Game: Can Big Tech’s AI Investments Pay Off?

By TMC Research Staff | The Milwaukee Company | tmcresearch@themilwaukeecompany.com

The artificial intelligence (AI) gold rush has been underway for a while now. From Amazon and Microsoft to Alphabet and Meta, the world’s largest tech companies have embarked on one of the most ambitious spending campaigns in recent history. Over $50 billion was poured into AI-related capital expenditures in a single quarter of 2024 by these tech behemoths, much of it channeled toward the construction of AI-optimized data centers and the procurement of cutting-edge hardware, like Nvidia’s famously costly chips. Yet, behind this spending bonanza lurks a looming question: Will all this capital ultimately deliver the returns that investors are hoping for? Or, like the infamous dot-com bubble of the late 1990s, will AI may prove to be more hype than substance?

The stakes are enormous, and the answer remains anything but clear.

The Temptation of Generative AI

Generative AI has captured the imagination of technologists, businesses, and consumers alike. The viral success of platforms like OpenAI's ChatGPT has demonstrated the technology’s potential, offering glimpses of a world where AI could transform everything from customer service to scientific research. Hundreds of millions of users have flocked to generative AI applications, and tech giants have wasted no time capitalizing on this momentum.

However, while user numbers may have surged, actual monetization remains elusive. Although businesses are testing how AI can increase productivity, premium offerings have yet to gain significant traction with consumers. The tech industry, optimistic about AI's future, appears to be betting on the long game. Sundar Pichai, CEO of Alphabet, captured this sentiment succinctly when he remarked that "the risk of underinvesting is dramatically greater than the risk of overinvesting." But many investors, wary of the sluggish pace of monetization, are left wondering whether AI will deliver tangible returns in the near term.

The Infrastructure Boom

The spending spree on AI has manifested most visibly in the construction of data centers. These centers, optimized to power AI models, require enormous resources—both financial and environmental. Unlike traditional data centers, AI centers house specialized chips, like Nvidia’s GPU units, which are essential for training complex AI models. These chips are not only expensive but also incredibly energy-intensive. For instance, Meta CEO Mark Zuckerberg has announced plans to procure 600,000 GPUs by the end of 2024, while Tesla’s Elon Musk has his sights set on 300,000 GPUs for his AI startup, xAi.

The scale of the infrastructure build-out is quite staggering. Microsoft has more than doubled its number of data centers since 2020, while Google has increased its count by 80%. Oracle, another significant player in the AI space, plans to add 100 more data centers to its portfolio. Yet, the rapid expansion comes at a cost: Data centers are notorious for their power demands, with even minor fluctuations in energy supply potentially derailing costly AI training sessions. Since 2015, the power demands of data centers in the U.S. and Canada have increased ninefold, a stark reminder of the environmental and economic pressures attached to AI's rapid growth.

Investor Patience Is Wearing Thin

While technology companies remain confident that AI will drive future growth, investors may be growing increasingly anxious. AI might be the future, but for many shareholders, the future needs to be more immediate. Even though tech companies have long viewed AI as a game-changer, Wall Street’s enthusiasm is showing signs of fatigue. Amidst recent market volatility, companies like Meta and Microsoft have already faced share price declines due to concerns over their AI spending.

This pattern is particularly stark when one considers the exorbitant figures that these companies are throwing around. Goldman Sachs predicts that tech giants could spend over $1 trillion on AI infrastructure over the next five years. To justify this investment, AI would need to generate an estimated $600 billion in revenue from data centers and chips alone. Yet, current AI-related revenues, while impressive, are estimated to only reach tens of billions—a far cry from the necessary targets.

Moreover, while AI spending has shot through the roof, the reality is that much of this capital is being funneled into infrastructure that may not yield immediate returns. In a particularly striking metaphor, Brent Thill of investment bank Jefferies likened the situation to “setting up for a day of surfing.” Tech companies, he argued, are still lugging their boards to the beach, but they haven’t caught the wave yet​.

A Longer Path to Profitability

Part of the issue is that the technology itself may be at its formative stages and appears to be still maturing. While AI enthusiasts predict that the technology will fundamentally reshape entire industries, from finance to healthcare, the timelines for these revolutions remain murky. Daron Acemoglu, an economist at the Massachusetts Institute of Technology, estimates that only a quarter of tasks currently exposed to AI will be cost-effective to automate within the next decade. This implies that the widespread economic impact many are predicting may still be several years away.

Executives, for their part, are trying to temper expectations. On a recent earnings call, Zuckerberg acknowledged that it would be years before AI applications were fully monetized. Google’s Pichai offered a similar perspective, emphasizing the “time curve” required to transform AI breakthroughs into meaningful revenue streams​.

Despite these reassurances, there’s no denying that AI presents significant risks. Unlike traditional tech infrastructure, which could often be adapted for new purposes, AI infrastructure is highly specialized. If the promised AI revolution fails to materialize, these expensive data centers and hardware could become the next generation's fiber-optic cable—a massive investment that takes decades to deliver returns.

The Lessons of the Dot-Com Bubble

The parallels to the dot-com era are hard to ignore. In the late 1990s, tech companies invested heavily in fiber-optic networks, convinced that the internet would soon become the backbone of the global economy. While that prediction was ultimately correct, it took far longer than expected for the infrastructure to pay off. Many companies folded in the intervening years, crushed under the weight of their investments.

AI, much like the internet, has the potential to transform the global economy. But whether it will do so on the timelines promised by today’s tech executives is far from certain. For now, the most visible beneficiaries of the AI spending boom are companies like Nvidia, whose chips power the AI models. But for Microsoft, Meta, Alphabet, and Amazon, the payoff appears to remain out of reach at least in the near future.

Conclusion: A Wager on the Future

The tech industry’s AI spending spree is a massive wager on the future. AI undoubtedly holds immense potential, and the investments made today could set the stage for tomorrow’s breakthroughs. Yet, the costs are immense, and the timeline for returns remains frustratingly uncertain. While AI optimists maintain that the risk of underinvesting is far greater than overinvesting, skeptical investors may be left wondering whether the industry has once again gotten ahead of itself.

In the end, the question is not just whether AI will change the world—it’s whether it will do so soon enough to justify the billions of dollars currently being spent. Investors, for now, seem to be watching and waiting, hoping that the wave they’ve bet on arrives sooner rather than later.

For enquiries contact Michael Willms at mwillms@themilwaukeecompany.com