Morgan Stanley flags AI investment surge beyond dot-com peak levels

Thursday 11 - 08:34

By: Dakir Madiha

Morgan Stanley flags AI investment surge beyond dot-com peak levels

Artificial intelligence infrastructure spending by major technology companies has reached a scale that now exceeds the intensity seen during the peak of the dot-com era. The investment surge is driven by hyperscalers including Amazon, Alphabet, Meta, Microsoft, and Oracle as they expand data center capacity and computing power to support large-scale AI deployment. Analysts warn that the pace of capital deployment is reshaping the financial structure of the global technology sector.

Projected capital expenditure ratios show a steep upward trajectory. The combined cash capex to sales ratio for the five major AI investors is expected to reach 36 percent in 2026, rise to 44 percent in 2027, and hit 45 percent in 2028 when lease obligations are included. At the height of the dot-com bubble, the comparable ratio stood near 32 percent. Total combined spending is forecast at about 805 billion dollars in 2025, climbing to 1.12 trillion dollars in 2027, more than four times the level recorded in 2024.

The expansion comes with rising concerns over future amortization burdens. Several large technology firms could collectively record more than 680 billion dollars in depreciation charges in the coming years as newly built infrastructure begins to age and lose value on balance sheets. Oracle shows one of the sharpest projected increases, with amortization rising from about 7 percent of revenue in 2025 to 28 percent by 2028. Alphabet is expected to move from 4 percent to 11 percent over the same period.

Leverage risks are also increasing through off balance sheet commitments and leasing structures that amplify real capital exposure. Combined purchase obligations among leading hyperscalers are approaching 1 trillion dollars, while their share of total US large cap capital spending is projected to reach roughly 40 percent between 2026 and 2028. Broader financing needs are also expanding, with global AI related debt issuance expected to near 570 billion dollars in 2026 as companies rely more heavily on credit markets to sustain infrastructure expansion.