Surging AI Power Demand Fuels Strategic Growth for Natural Gas Pipelines

2026-03-01 companies

Houston, Saturday, 28 February 2026.
With AI video generation requiring 100 times more power than images, the resulting energy surge is driving billions in natural gas infrastructure projects for Kinder Morgan and Energy Transfer.

The Physics of AI Demand

The energy intensity of artificial intelligence has fundamentally shifted the operational landscape for power generation. As of February 2026, new research indicates that generating AI video content consumes over 100 times more electricity than generating static images [3]. This spike is compounded by the fact that inference—the process where a trained model answers a user request—now accounts for 80 to 90 percent of total computing demand in the AI economy [3]. Consequently, the Electric Power Research Institute (EPRI) released findings on February 26, 2026, projecting that data centers could consume between 9% and 17% of total U.S. electricity by 2030, a figure that represents a doubling of their current share [5]. This trajectory aligns with forecasts from Goldman Sachs, which earlier this month predicted that power demand from data centers could surge by 50% by next year and by as much as 165% by the end of the decade [6].

Midstream Giants Capitalize on Infrastructure Needs

To meet this escalating demand, midstream energy operators are rapidly expanding their natural gas networks, which remain critical for providing reliable baseload power. Kinder Morgan (NYSE: KMI), which already transports 40% of the country’s natural gas production, has secured $10 billion in expansion projects, with $9.1 billion specifically allocated to natural gas infrastructure [1]. The company is actively pursuing an additional $10 billion in potential investments, with projects scheduled to come online through 2030 [1]. Financially, Kinder Morgan anticipates mid-single-digit earnings growth in 2026, with acceleration expected in 2027 as these large-scale pipeline projects enter service [1].

Grid Bottlenecks and Utility Responses

Despite the aggressive expansion by pipeline operators, the physical build-out of data centers faces significant headwinds due to power supply limitations. A report from Sightline Climate on February 24, 2026, revealed that approximately one-third of data centers scheduled to open in 2026 may face delays or cancellation [5]. Furthermore, as of February 26, 11 gigawatts of 2026 data center capacity remain in the “announced stage” with no signs of active construction [6]. This bottleneck has prompted U.S. utilities to boost capital expenditure plans to record heights; for instance, Duke Energy has increased its five-year capex plan to $103 billion, while Dominion Energy has raised its target to $65 billion to accommodate the concentration of data centers in northern Virginia [7].

Market Implications and Systemic Risk

The integration of AI into the broader economy has created a market environment reminiscent of previous technology booms, though with distinct structural differences. Morgan Stanley analysts note that the top ten names in the S&P 500 now comprise nearly 40% of the index, compared to about 25% during the peak of 1999 [2]. This concentration reflects an “arms race” in capital spending, where the largest tech companies could theoretically issue an extra $700 billion in debt without triggering a credit downgrade [2]. However, this spending surge has raised concerns about systemic risk, as the AI theme has spread beyond technology into industrials, energy, and commodities [2]. Investors are advised to monitor these developments closely, as Big Tech firms are expected to meet at the White House in early March 2026 to address the rising pressure to fund their own electricity infrastructure [6].

Sources

Data Centers Midstream Energy