The Arbitrage Window
AI infrastructure spend is outrunning your tariff anxiety
DHL's latest data shows the AI buildout is generating more freight volume than tariff policy is destroying. Your cost model may be obsolete.
DHL just said out loud what most freight teams are afraid to model. The AI infrastructure boom is generating more global logistics volume than tariff policy is destroying. That is not a narrative. That is a freight data point from one of the largest networks on earth. If your team is still running tariff-sensitivity scenarios as the dominant planning variable, you are optimizing for the wrong lever.
Who loses when the frame is wrong
Brands locked into a tariff-first cost model are making two mistakes. First, they are treating landed cost as fixed inside a volatile carrier market. Second, they are ignoring that demand shocks from AI infrastructure construction, data center buildouts, and industrial electrification are tightening capacity on specific lanes. That tightness is not uniform. It clusters by origin port, freight class, and modal mix. If your SKU velocity depends on a lane that now shares capacity with server rack shipments and copper cabling, your lead time assumptions are stale. Renegotiating rates on a static volume forecast while capacity is being absorbed by a different industry is how you end up with a NetPPM problem you cannot explain to your CFO.
Who wins in a volume-positive market
The brands positioned to win are the ones treating this as a cohort separation event. Not all freight demand is equal. AI infrastructure shipments are heavy, time-sensitive, and moving on contracts. That pressure pushes spot capacity toward consumer goods lanes in some corridors and away from them in others. The operators who mapped their lane exposure in the past 60 days, identified where AI-related industrial freight is competing with their own shipments, and locked in forward rate agreements on the exposed lanes are already ahead. The ones still waiting for tariff clarity are giving up margin while they wait.
The specific move your ops team should run this week
Pull your top 20 SKUs by 90-day velocity. Map their inbound lane by origin port and carrier. Cross-reference against known AI infrastructure project concentrations in those corridors. This is not a theoretical exercise. Industrial freight from chip fab expansion and data center construction is moving through the same ports as your goods. Where your lanes overlap, you have a capacity risk that tariff modeling will not surface. Flag those SKUs for a lead time buffer review. Add two to four days to your reorder point calculation on those items. If your 3PL or freight broker cannot give you lane-level capacity data segmented by freight type, that is a vendor performance issue, not a data availability issue. The data exists. Demand it.
Landed cost needs a new input
Your landed cost formula is missing a variable. Most models run: product cost plus duties plus freight plus drayage plus warehouse receipt. That is correct but incomplete in a market where freight rates on affected lanes are being pushed by demand from sectors your buyers have never tracked. Add a sector-competition multiplier to your landed cost model. It does not need to be precise on day one. A range-based estimate, built from your carrier's lane reports and cross-checked against public infrastructure project timelines, will get you within a margin your pricing team can work with. This is how top-decile operators are protecting sell-through. They are not waiting for the rate to hit the invoice. They are building the pressure into the buy.
Three questions to pressure-test
First: Can your freight broker name the top three non-retail freight types competing for capacity on your primary inbound lanes right now? If the answer is no, you are flying blind on one of your biggest cost inputs. Second: When did you last update the lead time buffer on your top 20 velocity SKUs, and was that update based on lane-level carrier data or a general market assumption? Third: If AI infrastructure freight demand keeps growing at its current pace through Q1 2027, what does that do to your landed cost on your highest-margin ASIN, and have you stress-tested that number against your next pricing window? Get those three answered before your next inventory buy. Then move.
Get The Beacon. Free, every Friday.
You're on the list.
The next edition lands in your inbox on Friday.