GM is reconfiguring its North American production system in real time, shrinking excess EV capacity, expanding U.S. ICE output, and using that shift to structurally cut its exposure to tariffs and regulatory whiplash.
In Brief
- GM is treating tariffs as a permanent input cost and redesigning its manufacturing footprint and mix to engineer that cost out.
- The company has taken a deliberate step back from its original EV build plan, converting capacity back to ICE and resetting supplier commitments.
- U.S. onshoring is being funded and executed as a supply chain program: new plants, retooled lines, and automation deployed where truck and SUV margins are highest.
The Strategic Break: EV Pullback, ICE Onshore
The structural break in GM’s operating logic is clear. Over the past year it has abandoned the assumption that EV capacity is an unqualified asset and that tariffs are a short‑term distortion. Instead, it has made both explicit design parameters of the North American network.
Two decisions anchor the shift. First, GM has reversed part of its EV capacity build. It has sold its share in the Ultium Cells Lansing plant and is transitioning Orion Assembly in Michigan from EVs back to full‑size ICE pickups. This is not a minor trim; CFO Paul Jacobson quantified Q3 and Q4 2025 charges of 7.6 billion dollars, with 4.6 billion dollars to be settled in cash and most of that cash going out in 2026. Those charges settle commercial claims with suppliers and write down EV‑specific assets, and management expects 1 to 1.5 billion dollars of annual benefit from the resulting mix and cost change in 2026.
Second, GM has committed capital to a different kind of localisation. Around half of the planned 10 to 12 billion dollars of annual capex in 2026 and 2027 is earmarked for expanding U.S. manufacturing capacity for high‑demand vehicles and ‘further reduce our tariff exposure’, in Jacobson’s words. Mary Barra framed it more plainly: a regulatory environment ‘increasingly aligned with customer demand’ allows GM to onshore more ICE production to meet that demand.
Together, these moves mark a reset of the EV/ICE balance in the network and a pivot of capital towards U.S. truck and SUV capacity as a twin lever for growth and tariff resilience.
How The New Capacity Model Works
Operationally, the rewiring plays out through a sequence of footprint and capacity decisions that change where and what GM builds in North America.
At plant level, Orion Assembly is the most visible symbol. Originally slated for EV pickups, it is being retooled for next‑generation full‑size ICE pickups, with a planned deployment of 2,500 robots and cobots under GM‑designed control software. That automation program has direct implications for line balance, labor productivity, and ergonomics on high‑margin products rather than on low‑volume EVs.
On the battery side, the temporary downtime at the Ultium Cells joint venture and the Lansing exit slow cell output and pull down production tax credits in 2026. GM plans to offset some of that impact through inventory adjustments, running down cell stocks. The company positions this as a sequencing issue: lower credits in 2026 then become a tailwind in 2027 when normalized production resumes.
Network‑wide, GM is raising its forward U.S. production ambition to roughly 2 million units annually, explicitly linking that to the localization of specific nameplates. The Chevrolet Equinox is slated for Kansas, the Chevrolet Blazer for Tennessee, with incremental capacity for the Cadillac Escalade and new full‑size pickups at Orion in Michigan. This pushes key crossovers and trucks into domestic plants and shortens supply lines between body, powertrain, and final assembly.
In operational terms, this kind of shift requires a coordinated rework of sourcing volumes, inbound logistics lanes, and plant takt times. GM’s guidance that ICE volumes in 2026 will be constrained by model changeover downtime (for new Silverado and Sierra launches) shows how this is planned: build‑ahead where possible, accept short‑term loss of units during retooling, and then rely on higher run rates and pricing in 2027.
The EV side of the network is not simply abandoned. GM still plans for next‑generation battery chemistry (LMR) from 2028, with ‘several thousand dollars’ of cell and pack cost reduction per vehicle, and a second‑generation software‑defined architecture that will cover ICE and EV platforms alike. The capacity reset buys time for those economics to mature while limiting the capital tied up in under‑utilised EV lines.
Tariffs Treated as a Design Variable
GM’s tariff language is notable for its specificity. Through Q3 2025 the company incurred 2.4 billion dollars of gross tariff costs, with another 700 million dollars in Q4 taking the annual total to 3.1 billion dollars. That came in below earlier expectations of 3.5 to 4.5 billion dollars, and more than 40 percent of the gross cost was offset through go‑to‑market and cost actions.
For 2026, guidance assumes 3 to 4 billion dollars of gross tariffs again, with 750 million to 1 billion dollars landing in Q1 alone. Yet Jacobson stated that net tariff costs should be lower in 2026 than in 2025, implying offsetting measures beyond the 40 percent already achieved. He described the offset as a mix of pricing, manufacturing footprint changes and fixed cost reductions.
This aligns with a broader sector pattern. Ford, Mercedes‑Benz, Stellantis, Honda and Volkswagen have all disclosed multi‑billion dollar tariff burdens and made similar statements about treating them as structural. Where GM stands out is in the scale of deliberate U.S. onshoring as a mitigation program rather than just a compliance response. Approximately 5 billion dollars of the 2026–27 capex window is explicitly tagged to U.S. capacity that will, over time, move content inside the tariff wall.
At network level, this is implemented through re‑routing component flows to maximize USMCA benefits, shifting marginal volume from imports to domestic plants, and revisiting where future nameplates are allocated. The assumption of a 15 percent Korea tariff rather than 25 percent is built into GM’s 2026 guidance, and Barra acknowledged that any reversal would be a headwind the company would work to offset. That comment signals that tariff scenarios are now embedded in planning cadences rather than managed as ex‑post surprises.
Inventory Discipline as a Control System
The tariff and capacity narrative is underpinned by a tighter production‑to‑demand discipline than GM has historically maintained. Total company revenue fell 5 percent year‑on‑year in Q4 2025, and management attributed that primarily to a ‘disciplined approach to production and dealer inventory, including aligning EV production to demand’. In North America, GM ended 2025 with 48 days of dealer inventory, slightly below its 50‑ to 60‑day target range.
This inventory position is not incidental. It is part of a working‑capital and margin logic that deliberately trades some top‑line volume for better price realization and cash generation. Over the past five years, GM’s average annual free cash flow has risen from 3 billion dollars to 10 billion dollars, and nearly 25 billion dollars of free cash was generated across 2024–25. Jacobson explicitly linked that consistency to inventory and production discipline, and reiterated that no ‘big buildup’ is contemplated even as the truck changeovers occur.
In operational terms, this requires a more integrated planning cadence between sales, finance, and manufacturing: common views of order banks, dealer days‑of‑supply thresholds by segment, and clear escalation when production, tariff and mix constraints collide. It also depends on clean master data and SKU discipline so that buildable combinations match what can be sold without triggering incentive creep.
The approach mirrors moves seen at other large OEMs, but GM’s willingness to accept visible revenue decline while it resets EV capacity and absorbs tariffs suggests a more rigid adherence to inventory guardrails than in prior cycles.
Resilience Costs are Being Pulled Forward
GM’s supply chain rewiring carries its own cost headwinds. For 2026, Jacobson flagged 1 to 1.5 billion dollars of incremental costs from onshoring vehicle production to the U.S., investments in supply chain resiliency, and software initiatives. He split those roughly 50/50 between physical network changes and software investments.
On top of those, GM expects 1 to 1.5 billion dollars of headwinds from aluminum, copper, DRAM and foreign exchange. Semiconductor risk has not disappeared; the company spent 100 million dollars in Q4 2025 on alternative chip sourcing related to Nexperia and expects a similar cost in Q1 2026, explicitly choosing higher unit cost over production disruption.
These resilience expenditures sit alongside other structural shifts in the cost base: a projected 1 billion dollar improvement in warranty costs in 2026, 500 to 750 million dollars of savings from lower regulatory credit purchases as CAFE penalties are zeroed, and 400 million dollars of incremental high‑margin software and services revenue from OnStar and Super Cruise. The combined effect is a cost and revenue mix that is more complex but also more diversified, with a greater share of profitability coming from aftersales and digital layers atop the physical fleet.
An Operating Model Built Around North America
GM’s stated ambition is to return North America to 8 to 10 percent EBIT‑adjusted margins in 2026, and management claims that the midpoint of its 13 to 15 billion dollar EBIT guidance supports that. The way there runs through the supply chain.
The network is being tilted decisively toward domestically built, high‑margin ICE trucks and SUVs, with flexible EV capacity that can be throttled until pack economics improve. Tariffs are being treated as a permanent cost to be engineered around, not a transient policy risk. Inventory and production are being held tighter to demand, formulating working capital as a deliberate control variable. And automation, AI, and software‑defined architectures are being targeted at the plants and vehicles that matter most for margin contribution.
This operating model increases GM’s dependence on the robustness of its U.S. industrial base and the precision of its planning routines. It also reduces its sensitivity to one‑way EV forecasts and tariff shocks in a way that few of its peers have yet matched at this scale.
