When the internet revolution began in the 1990s, capital from around the world poured into dot-com companies. The bubble burst in 2000. Countless firms and astronomical sums of investment money evaporated. But the internet did not disappear, and on the fiber-optic cables, networks and telecommunications lines laid down at that time, new empires such as Google, Amazon and Meta were born.
Twenty-five years later, money from around the world is again being drawn into the vast black hole known as artificial intelligence. Debate is heated over whether generative AI is the prelude to a new civilization that will explosively expand human productivity, or a "dot-com bubble season two" created by excessive expectations.
The blaze of capital pouring out of global Big Tech is beyond imagination. Data released in July by the global investment platform moomoo shows at a glance just how fierce the "AI computing arms race" among U.S. Big Tech firms has become.
Meta is expected to sharply raise capital spending from $72 billion in 2025 to between $130 billion and $145 billion this year, based on its latest guidance. Microsoft is also expanding significantly, from $118 billion to $190 billion. Alphabet, Google's parent, is projected to more than double its spending, from $91.5 billion to between $195 billion and $205 billion. Amazon has likewise lifted its investment to $220 billion from $131 billion.
This year's investment alone reaches as much as $760 billion, an astronomical figure well above 1,000 trillion won in Korean currency. These enormous sums are flowing into AI data centers, high-performance servers, graphics processing units and power infrastructure.
Market views on this astronomical investment are sharply divided. Those who argue it marks a "sustainable beginning" see Big Tech's aggressive spending as an indispensable move to seize future dominance. Their logic is that AI is an infrastructure revolution that will change the operating system of every industry, and that the spending is a preemptive investment to gain an early lead in the era of artificial general intelligence. They expect returns on investment to become visible once cloud demand and enterprise AI adoption gain full traction.
The counterargument from those wary of a bubble is no less formidable. They point out that AI revenue is not growing fast enough to keep pace with the astronomical capital expenditure. The warning is that, as during the dot-com bubble, today's competition will also come back as an enormous cost burden and could shake the entire technology ecosystem.
Where, then, should Korea place its bet in this vast AI money game? If it goes head to head with U.S. Big Tech on financial firepower, its odds are low. We must choose a different path. We must pursue a multidimensional response strategy that competes on industrial depth rather than the size of capital.
First, rather than competing in general-purpose AI, we must build a niche ecosystem through demonstration of "K-vertical AI." Korea should concentrate on industry-specific business-to-business solutions that combine AI with industrial data in the fields where it holds world-class competitiveness — automobiles, semiconductors, biotechnology and Korean content. It is important to fully open high-quality data in the public sector and to substantively drive the AI transformation of domestic companies.
Second, special measures are needed for the supply of infrastructure such as electricity and data centers. AI infrastructure is an electricity guzzler. To secure a stable power grid, Korea should use the Distributed Energy Act to give preferential tax benefits and power supply to data centers outside the greater Seoul area, and should swiftly build an eco-friendly energy mix system that rationally links nuclear power, small modular reactors and renewable energy.
Third comes building an independent AI ecosystem based on domestic neural processing units and securing regulatory flexibility. Excessive dependence on specific foreign graphics processing units is directly tied to the issue of technological sovereignty. Korea must accelerate the "K-Cloud project," apply domestic NPUs and computing power to AI data centers and demonstration projects, and create a market where small and mid-sized firms and startups can use them cheaply and reliably. Outdated rules that block innovation must be boldly removed, and institutional buffers put in place.
No one can guarantee how long the AI investment boom will last. Some companies will fail, and there will be aftereffects from overinvestment. But the AI revolution will not stop. What we need now is to build a differentiated Korean-style AI strategy based on our advanced manufacturing, accumulated industrial data, solid infrastructure and dense technological capability. That is the path for Korea to end its history as a follower and leap forward as an "AI G3 nation" leading the world.
