Fact Check: Global AI Spending to Surpass 5,000 Trillion Won Next Year
A widely circulated claim that global AI spending will break through 5,000 trillion won next year holds up as an accurate reading of current industry forecasts, but it cannot be confirmed as fact until 2027 has actually run its course, according to a review of 12 primary sources published by this newspaper's fact-check desk on Sept. 18. The projection, associated with research firm Gartner, has become a reference point in Korean coverage of the AI infrastructure boom now driving demand for semiconductors and GPUs from suppliers such as SK hynix and Nvidia.
The claim
The claim under review states that worldwide spending on AI will exceed 5,000 trillion won in 2027, covering AI-related hardware, software and services. The figure has spread across Korean tech media and online communities, where it is frequently cited as evidence of an unprecedented investment cycle in AI data centers and the chips that power them.
What the sources show
์ถ์ฒ: ์กฐ์ฌ ์ถ์ฒ
The desk examined 12 primary sources, among them Gartner materials, and found no contradiction between the claim and the published outlook. The 5,000 trillion won figure matches the consensus of the forecasting documents reviewed, and nothing in the source material suggests the claim misstates or inflates the projection.
Why the figure cannot be confirmed โ yet
A forecast is a projection, not a record of spending. Because 2027 has not occurred, the accuracy of the prediction cannot be judged, and the desk treated that distinction as central to its assessment. Outlays could land above or below the mark depending on enterprise adoption, the supply of advanced semiconductors, and shifts in data center budgets over the coming quarters.
Why it matters for Seoul
The projected spending wave is closely watched in Korea, where the AI buildout rests on the tight coupling of GPU maker Nvidia and memory supplier SK hynix. Any sustained rise in global AI investment flows directly into orders for Korean-made semiconductors, making the accuracy of such forecasts a matter of near-term commercial interest, not just headline numbers.
Judged as a faithful representation of the current research consensus, while acknowledging that the actual spending has yet to take place, the claim is rated Mostly True.