Ford CFO Sherry House disclosed that memory chip costs driven by AI data center demand will add roughly $1 billion to the company's expenses in 2026, and the price shock is already working its way toward showroom floors.
The tariff conversation has dominated Detroit earnings calls all year, but Ford's chief financial officer quietly dropped a different kind of cost bomb in February. Sherry House told investors that DRAM memory chip prices, supercharged by AI infrastructure spending, would add about $1 billion to the company's cost sheet in 2026. It's not a supply disruption in the traditional sense. Ford says it has enough chips. The problem is what they now cost, and why.
Spot DRAM prices jumped roughly 450 percent between September 2025 and January 2026, according to industry data tracked by analysts and reported by Fortune. The cause isn't a factory fire or a pandemic lockdown. Samsung, SK Hynix, and Micron, the three companies that together control more than 90 percent of global DRAM production, have been converting their fabrication lines to make high-bandwidth memory, the specialized chip stacks that power AI accelerators. HBM generates three to five times the revenue per wafer of conventional DDR4. The economics aren't subtle. When Nvidia and its hyperscaler customers are willing to pay a premium that large, the conventional chip market gets starved of capacity.
The result is a shortage that has nothing to do with demand for cars falling off. Mid-to-high-end vehicles now use DRAM extensively: infotainment systems, advanced driver-assistance hardware, digital instrument clusters, over-the-air update architecture. Per-vehicle memory costs for those models have risen from a range of $40 to $90 to somewhere between $90 and $220, according to figures cited in industry reporting. That's before a single tariff enters the picture.
Automakers account for less than 10 percent of global DRAM demand. That's the core problem, and it's a structural one. When you're buying 8 percent of a market that's being radically repriced by buyers who need it far more urgently and can pay far more for it, your say in the matter is close to zero. Hyperscalers have been reserving HBM capacity years in advance, locking in supply that simply isn't available to carmakers. SK Hynix has warned publicly that shortages could persist well past 2030. That isn't a forecast about car chips specifically, but it describes the market automakers are now operating in.
Ford has been trying to get ahead of it. Micron signed a long-term supply agreement with Ford to lock in chip volumes for the next generation of vehicles, as reported by TheStreet. That kind of deal, once unthinkable for a commodity component, is now standard supply-chain management in Detroit. The memory chip is becoming the new steel: something you hedge, contract, and sweat over.
House said Ford expects material and warranty cost reductions to offset the DRAM increase, which is the right thing to say on an earnings call. Whether that holds if prices stay elevated through 2027 or beyond is a different question. Nobody knows. SK Hynix and Micron have both reported their entire 2026 HBM production is already sold out. New capacity takes years to build. There's no fast fix here.
What this actually means for car prices #
Ford didn't say it would raise sticker prices by a specific amount because of DRAM. Companies rarely do. But the math of absorbing a billion dollars in unplanned costs while simultaneously managing tariff exposure, an EV transition, and wage inflation has a way of showing up eventually. It shows up in trim-level cuts, in option-package repricing, in the quiet removal of features that used to come standard.
The chip industry has been signaling this for months. AI infrastructure spending isn't a contained phenomenon. Data centers have become the dominant buyer of DRAM globally, and when that share grows, everyone downstream, including the person shopping for a new pickup truck, pays through some mechanism or another. Ford's CFO just named the mechanism and attached a number to it. That number is $1 billion. It won't stay abstract for long.
Also read: AI data centers have quietly become the biggest threat to grid stability America has ever built • Monday.com cuts 630 jobs and calls it an AI pivot, but the math tells a harder story • Bank of America says the US deficit will hit $2 trillion in 2026 and the bill is coming due for startups