US grid buyers outnumber Korea's equipment sellers four to one at this year's KOTRA event

Ten Korean suppliers met twelve American buyers on last year's version of this trip, roughly one apiece ^[4]. This year twelve suppliers face about fifty buyers, a little over four each, so the buyer count per supplier is up about three and a half times while the Korean side added two companies ^[3][1][2].

Whether those twelve are Korea's large listed makers or its small and mid-sized ones changes what a first order is worth, and the report does not name them ^[16]. The sponsors are KOTRA's Silicon Valley and Dallas offices, Korea Electric Power Corp and the Korea Electric Association of Cooperatives ^[2]. Last year the group went to Dallas only ^[7]. The Seoul Economic Daily reported that shortages of power equipment have prompted US buyers to court Korean suppliers ^[15].

The demand case rests on Wood Mackenzie's April analysis, which puts the US market for data center power equipment at $65 billion in 2030 against $20 billion in 2025 ^[9], 3.25 times in five years, or about 27 percent a year compounded ^[3]. The unit forecasts for the largest facilities run at nearly the same slope, pad-mounted transformers from 1,573 to 9,395 and switchboards from 5,111 to more than 30,500, both close to sixfold ^[11][4]. Capacity goes from 24 gigawatts in 2026 to 110 in 2030, 86 gigawatts of new load in four years, with data centers accounting for 68 percent of the increase in US power demand over that period ^[10][6].

The equipment market is small next to the money commissioning it. That $65 billion in 2030 is about 9 percent of the roughly $700 billion hyperscalers including Amazon, Microsoft and Google are spending this year ^[8][5]. For a buyer at that scale, switchgear is a small line item and a schedule risk at once. For a mid-sized Korean maker, one framework agreement is a year of output. I would expect delivery dates to decide more of these orders than price does.

The window may be shorter than the 2030 curve implies. Musk said the bottleneck in natural gas turbine production is the casting of blades and vanes, and that doing the casting in-house at SpaceX could bring turbines online up to 18 months earlier ^[13]. SpaceX has begun work on a casting plant, The Information reported ^[14]. Musk also said gas will be needed for the next several years to complement the 100 gigawatts a year of solar that SpaceX and Tesla are each building ^[12].

Three readings fit the evidence. Fifty meetings can produce no orders, in which case the headcount was a trade-promotion number. American buyers can use Korean quotes to squeeze incumbent suppliers and sign nothing. Or the shortage holds long enough to put unnamed mid-sized Korean firms onto the approved vendor lists of utilities the size of PG&E ^[6]. I lean to the third, mostly because about fifty procurement people giving up a day is a cost someone paid and the 2030 forecast is free ^[3][9]. If the 2027 trip goes back to Dallas alone with a dozen buyers, I was wrong.