[비즈한국] "AI is a bubble." As global stock markets have recently entered a period of consolidation, this sentiment is being voiced once again. However, looking at the moves of big tech companies like Meta, OpenAI, and NVIDIA, it suggests not just a "bubble," but rather the "direction of capital movement."
Meta has announced that it will invest $600 billion (approximately 880 trillion won) in the U.S. AI industry over the next three years. This is not merely about server expansion; it is a national-level project that includes data centers, power grids, energy, and community infrastructure. Meta emphasized, "AI data centers are the engines that power technology and the infrastructure that drives economic growth."
In fact, since 2020, Meta's investments have added 15GW of new power grid capacity and created tens of billions of dollars worth of local jobs.
OpenAI has also embarked on constructing a massive AI data center project worth up to $500 billion. The company has requested the White House to "expand tax credit benefits to include AI infrastructure as well." The proposal aims to broaden tax incentives currently focused on semiconductor manufacturing to include AI server manufacturing and data centers. This signifies that AI investment has now moved beyond private innovation to become a pillar of national industrial policy.
NVIDIA, the core of the AI frenzy, continues to expand. CEO Jensen Huang recently confirmed at TSMC’s annual sports day in Taiwan that "demand for Blackwell GPUs is explosive," officially confirming additional wafer orders. The chairman of wafer manufacturer TSMC also stated, "NVIDIA has requested more wafers." CEO Huang mentioned, "We have already received next-generation memory samples from SK Hynix000660, Samsung Electronics005930, and Micron," hinting that the AI semiconductor supply chain is expanding once again.
Kim Dong-won, Head of Research at KB Securities, drew a line by stating, "It is premature to compare the AI industry to the 1999 dot-com bubble." Kim analyzed, "In 1999, the U.S. was in a tightening phase, and dot-com companies had an average price-to-earnings ratio (PER) of 60, whereas current AI companies are at a PER of only around 30. AI is the third industrial revolution, following the PC and mobile eras." He added, "This year's KOSPI adjustment is just a comma in a major bull market," forecasting that the KOSPI could reach 5,000 points next year and 7,500 points in the long term.
The analysis suggests that semiconductors and power will drive performance in a "three-lows" (low interest rates, low oil prices, and low exchange rates) boom environment.
Park Seok-hyun, Deputy Manager at Woori Bank's WM Group, also predicted, "The foundation of the AI investment bubble debate lies in whether poor profits due to overinvestment will materialize. Ultimately, it is a question of whether corporate earnings, the source of investment, can maintain strong growth, and the outlook is positive."
The capital expenditure ratio of S&P 500 companies is still on an upward trend, and the ratio of capital expenditure to free cash flow is near its highest level since the pandemic. In other words, companies are still spending money for the future, and AI is connecting not to an "overheated story," but to capital-intensive real economy growth.
A report by DB Securities titled "Lessons from 1850s Railroad Stocks" offers an important hint for the current debate. The U.S. railroad bubble of that era was born under three conditions: optimism about new technology, abundant liquidity, and a robust real economy. The discovery of gold in California in 1848 created overflowing liquidity, and the Crimean War created a boom in the agricultural sector.
However, the bubble burst not because of technology, but when the economy collapsed. When the Crimean War ended in 1856, Europe resumed importing Russian grain; U.S. agriculture plummeted, leading to bad farmland mortgage loans, which caused a contraction in bank liquidity and ultimately a decline in railroad investment. In the end, the end of a bubble is always about the "everyday economy."
Kang Hyun-ki, a researcher at DB Securities, pointed out, "Recently, there has been a heated debate over the bubble in AI-related stocks, and one must remember that it is difficult to judge this through an analysis of the AI industry alone. Rather, what investors should watch is whether the general economy, excluding AI, is weakening."
Ultimately, what investors fearing an AI bubble need is not fear, but "distinction." The commonality of the PC (1980s), Internet (1990s), and mobile (2010s) revolutions is that the industries remained even after passing through their respective bubbles. For example, even after the dot-com bubble, internet penetration jumped from 38% in 1999 to 60% in 2002. Even after the froth was cleared, technology had deeply embedded itself into daily life. AI is following the same path.
An acquaintance who jumped into the AI industry earlier this year said, "AI is like the steam locomotive of the past." They added, "Those who secure AI first will be the ones to succeed in the future." Has their thinking changed now? The answer is "No."