News / Airfreight race gets set as the AI data-centre boom gathers pace

© Voraphach Phaisri

AI: boom or bust? With much of the airfreight industry eyeing the financial benefits of the server and hi-tech market, concerns over a potential AI bust are ever-present. 

However, the threat of a significant sharp drop in demand seems overblown, for now. 

According to Goldman Sachs, data-centre occupancy rates, which are already high, will climb from around 85% in 2023 to more than 95% in late 2026, before easing in 2027 as additional facilities come online. Regions such as Northern Virginia, Beijing, Shanghai, and the San Francisco Bay area remain global hotspots, and by 2030, 122GW of data-centre capacity is expected to be operational.  

While North America has the most capacity scheduled to come online over the next five years, the expansion is unmistakably global. According to Clifford Chance, Europe has 1.7GW of space under development, with the region’s data-centre market projected to reach $64.5bn by 2029 at a CAGR of 7%.

Africa, meanwhile, is preparing for demand that will exceed supply by 300% over the next two years, with construction spending expected to reach $3.06bn by 2030. 

For logistics providers, this rapid scaling is creating both opportunity and uncertainty. DHL Global Forwarding CEO Oscar de Bok told The Loadstar: “This is huge growth in data centres… in the volumes, in the logistics.”

With “trillions of investments” flowing into data centres over the next three years, he pointed out that the impact on air capacity would be profound, creating a “huge opportunity” for logistics specialists able to offer not just lift, but technical expertise. AI-server movements are uniquely sensitive, requiring specialised handling, assembly-adjacent services, and a controlled environment, requiring premium logistics partners.  

Manufacturers are, similarly, bracing for a step-change: last month in the FT, Foxconn signalled that AI server production revenue would almost double quarter to quarter, its chair, Young Liu, calling expectations of industry-wide doubling in shipments next year a “conservative estimate”.  

The group’s AI server production is in Taiwan, Mexico and increasingly in the US.

“The capacity in the US is mainly for the US market, the capacity in Taiwan will be for most of the rest of the world, and we also are building up capacity in Japan – the trend we’ve seen is local capacity for local consumption,” Liu told the FT. With hyperscalers targeting $400bn in capital investment this year, Foxconn says it must eventually build enough US production to cover up to 50% of American cloud-provider demand. 

Yet logistics executives caution that supply chain alignment will be essential. Taiwan-based Morrison Express CEO Asok Kumar noted both the buoyancy and fragility of the semiconductor-linked segment. He said: “The semicon business is clearly in growth mode, especially with AI… sometimes you need to be lucky, and we are, since we’re headquartered in Taiwan.”

The sector has given Taiwan-US among the highest airfreight rates in the world. Flexport head of airfreight DACH, Arno Hausch, said yesterday: “It’s put on Taiwan on the map  –  the high tech and AI industry is requesting a lot of urgent air freight from Taiwan”, adding that it had bolstered transpacific rates, with high demand from the US AI industry. 

But forecasting remains difficult, added Mr Kumar. “We’re having negotiations with carriers… we haven’t quite mapped it out exactly.”  

One major strategic question is whether semiconductor and AI-server flows will shift meaningfully from air to ocean. Mr Kumar says the move is “certainly possible”, where planning horizons allow it, especially as falling sea freight rates and overcapacity make ocean transport more attractive. But he added: “At this point, I’m not seeing that substantively in terms of a mode shift.” 

As 2026 approaches, the logistics landscape around AI hardware will be shaped by rapid investment, tightening capacity, and evolving transport economics. It is now one of the world’s fastest-growing and most strategically important supply chains. But the pace of growth will, ultimately, be led by businesses, according to The Economist.

It said: “For the world to reap productivity gains from AI, normal businesses must incorporate the tech into their day-to-day operations. It is also the most important question in determining whether or not the world is in an AI bubble. From today until 2030 big tech firms will spend $5trn on infrastructure to supply AI services. To make those investments worthwhile, they will need on the order of $650bn a year in AI revenues, according to JPMorgan Chase, up from about $50bn a year today.”