[Into The World Of AI] Bubble Theory? Global AI Investment Is Still Booming

Samjong KPMG Report
"AI Accounted for More Than Half of Last Year's Venture Investments"
"8 of the Top 10 Largest Deals Were Related to AI and Machine Learning"

"The current speed of AI development is 10 times faster than previous industrial revolutions. We are now standing at a moment of great transformation."

Sundar Pichai, Chief Executive Officer (CEO) of Google, made this remark in his opening speech at the AI Impact Summit underway in India. His comment was intended to dismiss the 'bubble theory' emerging in the industry and instead argue that investment should be increased.

In his speech on the 19th (local time), CEO Pichai announced the 'America-India Connect' project, a subsea optical cable route connecting the United States, India, and the Global South, and unveiled an infrastructure investment plan worth 15 billion dollars (about 21.68 trillion won). Previously, Alphabet, Google's parent company, also stated that it is planning this year's capital expenditures (CAPEX) in a range between at least 175 billion dollars (about 253 trillion won) and up to 185 billion dollars (about 268 trillion won).

Although concerns about an AI bubble are emerging, expectations that the global AI market can translate into real outcomes through two pillars, Physical AI and Agentic AI, are leading to expanded investment.

According to the report "Global Venture Investment: 2025 Q4 Trend Analysis and Outlook" recently published by the Samjong KPMG Economic Research Institute, last year's global venture investment reached 512.1 billion dollars (about 740.5 trillion won), up 30.7% from the previous year. This is the third-largest amount in the past 10 years, and the analysis is that large-scale venture investments in the AI sector drove this growth trend.

The report stated that AI investments accounted for more than half of last year's venture investments, totaling 270.2 billion dollars (about 390.8 trillion won). Even when looking only at individual investment deals, 8 out of the top 10 largest deals concluded last year were related to AI and machine learning. In particular, the 40 billion dollar investment that OpenAI attracted from SoftBank and others was recorded as the largest private investment deal ever.

This year as well, AI is expected to remain the top investment target for global venture capital (VC) firms. The research institute assessed, "AI is the key means for operational innovation and productivity improvement for companies and industries, and it will maintain its position as the top field for global VC investment," adding that this is "because AI is regarded as the core means of operational innovation and productivity improvement for industries."

Investment in AI infrastructure and AI companies across the industry has continued up to very recently. On the 18th, major foreign media outlets reported that Humane, a Saudi Arabian state-owned AI company, invested 3 billion dollars (about 4.34 trillion won) in xAI. Supported by Saudi Arabia's sovereign wealth fund, the Public Investment Fund (PIF), Humane was established last year under the initiative of Crown Prince Mohammed bin Salman, the de facto ruler of Saudi Arabia. At the 4th AI Impact Summit being held in New Delhi, India, OpenAI also decided to build a data center in India. This data center, which OpenAI will build together with Tata Group, a major Indian conglomerate, will start at a capacity of 100 megawatts (MW) and is planned to scale up to 1 gigawatt (GW).

The reason the industry draws a line against the AI bubble theory is that demand for real-world services such as AI chatbots is very strong. On top of this, demand is so high for AI infrastructure, including graphics processing units (GPUs), that supply cannot keep up. Jensen Huang, Chief Executive Officer (CEO) of Nvidia, said in an interview with U.S. business news channel CNBC earlier this month, "Demand (for AI) is going through the roof, and there is a fundamental reason for this," stressing, "We are in a once-in-a-generation period of infrastructure build-out. It is the largest infrastructure build-out in human history."

CEO Huang also expressed a negative view of the fact that the 'dot-com bubble' of the early 2000s is frequently compared to the recent AI bubble theory. Unlike that time, when many companies received investment but did not engage in real business activities, GPUs are now being used for actual work. He said, "Back then (during the dot-com bubble), there was 'dark fiber' (unused fiber-optic networks) that was laid but not used," declaring, "Now, there is no 'dark GPU'."