Air cargo traffic ‘defying gravity’ as AI sell-off leaves project forwarders unfazed

The cataclysmic sell-off in AI-related equipment that wiped out $1.3trn in chip stocks sent a powerful signal about the frothy AI scene, but it appears unlikely to dent the hot project forwarding market the AI boom has created. All the large chip makers saw their market caps drop by more than $100bn in the slump, led by Nvidia losing $238bn. A serious correction of AI firm valuations has loomed for some time, as rampant optimism around the technology had fuelled stratospheric revenue and profit projections of major actors. Still, market-watchers have shrugged off the sell-off as an event driven by sentiment rather than fundamentals, noting that the stock market showed a relatively moderate dip of 2% rather than the slump triggered by the end of the dot-com boom. “Today’s sell-off looks far more like late 1995 or late 1997 – violent recalibrations mid-boom, with years of upside still ahead,” one pundit commented on Investorplace. “Demand shows no sign of cracking,” he added. Still, the episode highlights the volatility of the sector and raises questions on how another, more pronounced slump would affect industries. Air cargo, for one, has been buoyed by this traffic, which has taken over from e-commerce as the chief growth engine. “Exceptional demand for semiconductors and AI-related hardware inspired 7% growth in global air demand in June,” Xeneta noted in its market report for the month and chief airfreight officer Niall van de Wouw commented that air cargo volumes were “defying gravity, thanks mostly to soaring AI-related shipments on Asia Pacific to North America corridors”. In the US the project logistics sector has been running “absolutely pedal to the metal” as one forwarder put it. “We are on several AI-related projects,” said Ragan Watson, project sales manager at Barnhart Crane & Rigging. “AI is the boom maker in the industry.” In addition to work on data centres, their immense power demand is keeping logistics providers on their toes. According to Mr Watson, that sector is undergoing an unprecedented growth pattern. “All major OEMS of power production are rapidly expanding operations and building new plants,” he said. Arguably, transformers have manifested themselves as the starkest indicator of the frantic demand from the AI side. Wait times for new units, which used to be 6-12 months before 2020, have stretched to 24-48 months, with large units taking over 36 months to deliver. Prices have soared 50% to 80% since 2020. The high volume of transformers moving to and across the US is stretching equipment availability, especially when it comes to rail cars. Capacity is “extremely tight”, one forwarder reported. Another noted that data centre electricity demand, as well as the large amounts of water required for cooling, had stirred up public opposition. The power grid was already under strain before the AI boom went into overdrive, and public worries over inflation have increased at the prospect of higher electricity bills. The White House has suggested that the large AI players should build their own power supplies. Mr Watson doubts that public opposition to data centres or a slump in AI build-out would quench the hunger for electricity, nor would it derail AI. The number of AI providers will thin out after the bubble, but demand will remain, he believes, likening the scenario to the dot-com bust, which wiped out many players but internet adoption continued to advance. Regarding power infrastructure needs, he anticipates demand to remain high for “quite some time”. “We were already at deficit before anybody could spell AI,” he said. A slowdown in data centre deployment could spell relief for industries that have struggled to get hold of memory chips, which the AI sector has gobbled up with a ferocious appetite.