The power buildout is short 20,000 apprentices a year, not short money

Take the pipeline figure literally and 2030 stops being a forecast. The deficit between the 2024 intake and the annual entry rate Goldman says is required is 20,000 people a year ^[21]. Run that from 2025 through 2030 and 120,000 apprentices who were supposed to be in training never start ^[22]. Measured against a single year of national intake, the full requirement is 11.3 years of apprentices at the 2024 rate, or 7.8 years at the rate the bank says is needed ^[23].

What makes that arithmetic binding rather than merely awkward is the training cycle. Electricians and lineworkers need three to four years ^[3], so someone who starts in 2027 certifies in 2030 or 2031 ^[26]. The 2030 crew is already enrolled or already absent. Capital carries no such lag, which is the whole tension: the planned utility spend works out to roughly $871,000 of committed capex standing behind each additional worker the report says is missing ^[24]. Goldman's own wording is plainer than the coverage of it. "Power is a critical bottleneck," the report says, "but increasingly, the requisite labor presents a structural constraint of its own," and "training cannot happen at the pace capital is being committed" ^[13].

Two details make the headcount worse than the headline. More than half the existing utility workforce has under a decade of experience ^[9], so the sector is short of the journeymen who supervise apprentices, not only of apprentices. And solar, wind and battery installations need more than 2.5 times the lifecycle workforce of fossil equivalents ^[10], so a cleaner build mix raises the requirement instead of relieving it.

The wage numbers deserve more attention than they got. The sector employed about 8.5 million people in 2024 at a median of $58,810, according to the Department of Energy, while traditional fuel production averages $65,400 and power plant operators take about $103,600 ^[14][15]. If the target is 20,000 more entrants annually, that spread is the price signal that has to move, and neither account reports any sign that it has.

Fortune puts humanoid robots in the gap instead. Goldman's unit forecast runs from 20,000 in 2025 to 1.4 million in 2035 ^[16], with widespread commercial deployment expected between 2027 and 2029 ^[17]. The 1.4 million lands five years after the shortfall it is offered to relieve ^[27]. Barclays' Zornitza Todorova told CNBC the market goes from roughly $3 billion now to $200 billion by 2035 ^[18]; cross that value against Goldman's volume and the implied price is about $143,000 a unit ^[28], which is a capital good on a depreciation schedule rather than a hire. Fortune also notes that banks and private equity have no historical performance data to underwrite these projects ^[20]. US manufacturing already carries more than a million unfilled materials-handling roles ^[19], and machines have not cleared those.

One note on bookkeeping: Fortune reports the requirement as 500,000, Cryptobriefing as 510,000, and the components it itemises sum to 507,000 ^[2][25]. The rounding is noise. The consequence is that the next wave of slipped in-service dates gets explained by crew availability, not by the cost of money.