On July 21, 2026, General Motors (GM) forecast that commodity market conditions, logistics, and rising DRAM prices would together create a $1.5 billion to $2.0 billion cost headwind for the full year. In China, BYD raised the price of its LiDAR-equipped driver-assistance option from 9,900 yuan to 12,000 yuan, citing higher storage costs. Neither company has explicitly named AI demand as the cause, but as memory markets shift to prioritize AI servers, memory costs are beginning to show up in automakers' profit plans and in the pricing of advanced features.
That said, GM's figure of up to $2 billion is not attributable to DRAM alone. And what BYD adjusted was not the price of the vehicle itself, but rather a driver-assistance system offered as an add-on for certain models. What we can confirm is limited to this: GM has named DRAM as one line item within a combined cost increase, BYD has moved the price of a limited option, and companies across the industry are expanding long-term supply contracts.
Breaking Down the Up-to-$2 Billion Figure and How GM Plans to Absorb It
On its Q2 earnings call, GM CFO Paul Jacobson put the combined 2026 full-year cost headwind from commodity markets, logistics, and DRAM at $1.5 billion to $2.0 billion. He noted that some of the DRAM-related inflation would be concentrated in the second half of the year. The company explained that in its North American operations in Q2, profit gains from pricing improvements, reduced EV losses, and lower warranty costs were partially offset by commodity costs—including logistics and DRAM—as well as costs tied to shifting production to the United States.
It's not possible to attribute the entire $2 billion figure to DRAM alone. GM has not disclosed a cost breakdown by commodity, nor has it revealed the contract prices for memory it sources from Micron or Samsung. On the earnings call, the company only noted that its strategic relationships with both companies date back to 2022 and have supported stable supply and joint roadmaps for next-generation memory.
Even so, DRAM was specifically called out as an earnings headwind. GM's Q2 revenue was $48.026 billion, with adjusted EBIT of $3.943 billion. Despite factoring in rising costs, the company raised its full-year 2026 adjusted EBIT guidance from a range of $13.5–15.5 billion to $14.0–16.0 billion. This absorption of DRAM-related and other burdens rests on assumptions including roughly a 0.5% pricing benefit in North America, along with $1.0–1.5 billion improvements each in warranty costs and EV losses.
In other words, memory prices have not immediately derailed GM's earnings plan. But DRAM has been added as a cost item of a scale that needs to be offset through pricing improvements and warranty cost reductions. Jacobson acknowledged that inflationary pressure could persist into 2027, and did not suggest that the burden would disappear by the end of 2026.
From 9,900 Yuan to 12,000 Yuan: BYD's Move on Driver-Assistance Pricing
In a price adjustment notice issued on April 28, BYD announced that starting May 1, the optional price of the LiDAR version of its "God's Eye B" (DiPilot 300) system—offered on select models under the Dynasty, Ocean, and Fangchengbao brands—would rise from 9,900 yuan to 12,000 yuan. That's an increase of 2,100 yuan, or 21.2%. BYD cited a "significant global increase in storage hardware costs" as the reason, and applied the old price to customers who had paid a deposit by April 30.
What's affected is an optional driver-assistance feature, not a blanket price increase across all BYD vehicles. Still, it shows that for a feature bundling a high-performance SoC, memory, and storage to process data streaming in from cameras and LiDAR, rising component costs could no longer be absorbed at the old price.
At BYD's Intelligent Driving Strategy launch event on May 28, Chairman and President Wang Chuanfu announced that the LiDAR version of God's Eye B would become available as an option across all BYD models. According to on-site coverage by the Chinese financial newspaper National Business Daily, Wang described the 12,000 yuan price as "cost price." BYD's official announcement stated that vehicles equipped with its driver-assistance systems now number over 3.15 million, generating more than 200 million kilometers of driving data per day. As the number of vehicles with driver-assistance features grows, fluctuations in memory costs stop being an issue confined to a handful of premium models.
The difference between GM and BYD lies in how the memory burden manifests. GM is absorbing it across the company's overall profit plan through a combination of pricing and warranty cost improvements. BYD, while limiting the scope, passed the 2,100 yuan increase directly onto the price of an add-on feature. Neither case points to a broad increase in base vehicle prices, but in both, the response to rising memory costs has surfaced in both earnings results and consumer-facing pricing.
Manufacturing Capacity Allocation Linking AI Demand and Conventional DRAM
According to TrendForce research, contract prices for conventional DRAM in Q1 2026 rose approximately 93–98% quarter-over-quarter, with a further 58–63% increase projected for Q2. These figures aren't specific to automotive-grade DRAM, but they reflect a situation in which the top three memory manufacturers are prioritizing production and shipments for high-price, high-margin server applications, tightening allocations for other uses such as PCs and smartphones.
AI and automotive applications don't necessarily use the same type of memory. HBM, used by AI accelerators, differs in both product and requirements from the LPDRAM, NOR, and UFS NAND used in automotive equipment. The transmission mechanism here isn't a story of AI companies directly buying up automotive components—it's about how DRAM manufacturers allocate their limited manufacturing capacity among customers and products. The expansion of AI inference is also pushing up demand for general-purpose server DRAM, and automakers have responded by moving toward long-term contracts to secure supply allocations.
At Micron, revenue from its Automotive and Embedded business unit in fiscal Q3 2026 reached $4.634 billion, roughly 4.11 times the $1.127 billion reported in the same period a year earlier. Gross margin for the segment also rose 53 percentage points, from 26% to 79%. However, this business unit includes embedded products beyond automotive applications, so the revenue growth cannot be attributed solely to rising automotive memory prices. Micron hasn't disclosed the respective contributions of volume, price, and product mix, so these results can't be used as a price indicator specific to automotive memory.
Usage on the automotive side is also trending upward. A 2023 summary of research commissioned by Micron from S&P Global Mobility put the average vehicle's RAM and NAND usage at 90GB at the time, projecting growth to roughly 278GB by 2026, and to around 2TB for high-end vehicles. That 278GB figure is not an actual measured value for 2026. Still, as infotainment, advanced driver-assistance, and in-cabin AI systems all stack up within the same vehicle, the trend toward higher memory capacity requirements per vehicle can be confirmed through the terms of the GM-Micron agreement as well.
Long-Term Supply Contracts Locking In Automotive Memory
On July 1, GM and Micron announced a strategic customer agreement to secure long-term supply of automotive memory, covering LPDRAM, NOR, and UFS NAND. The two companies will also collaborate on product definition, system optimization, and the validation and certification of next-generation memory for future vehicle architectures. Since neither company has disclosed procurement volumes, contract duration, or price-fixing mechanisms, it's still unclear how much this agreement will reduce GM's costs in 2027.
That same July, Micron also signed a long-term agreement with Ford, and announced strategic customer agreements with automotive Tier 1 suppliers including DENSO, Astemo, and Hyundai Mobis. The goal of these agreements is to give automakers visibility into supply and pricing, while giving Micron visibility into future demand—making it easier to plan certification processes and manufacturing capacity investments. Individual contract terms remain undisclosed, but a common thread across all of them is the early-stage alignment of long-term demand, supply capacity, and certification plans for next-generation products.
Additional manufacturing capacity is also coming online. Micron invested over $2 billion to modernize its Manassas, Virginia facility, announcing the start of 1α DRAM production there in May 2026. According to the company, this process is well-suited to DDR4 and LP4 for long-lifecycle applications. In discussing its agreement with GM, Micron explained that expanded and localized investment in automotive supply—including advanced DRAM manufacturing at Manassas—underpins the deal.
However, even as new manufacturing capacity comes online, that doesn't necessarily mean tightness in the automotive segment will ease immediately. Vehicles require components of consistent quality over long periods even after sale, and new-generation parts require validation and certification on specific vehicle platforms. The reason Micron is pursuing supply agreements and technical collaboration together is that simply swapping in different memory products isn't a viable option.
How Far Will the Costs Be Passed Through to Vehicle Prices?
What can be confirmed at this point is GM's combined cost increase and BYD's limited price hike on a driver-assistance option. The roughly 0.5% pricing benefit GM expects in North America includes multiple factors, such as product strength and prior-year model-year price adjustments, and no causal link to DRAM has been shown. Nor is there any figure indicating what percentage increase, if any, the auto industry as a whole is applying to base vehicle prices because of memory costs.
Even so, preparing for memory shortages has shifted from a footnote in earnings commentary to a genuine management issue. GM has explicitly named DRAM as a line item in its full-year cost outlook, BYD has adjusted pricing on an advanced feature, and Micron has expanded long-term contracts with both automakers and Tier 1 suppliers. The key indicators to watch will be conventional DRAM pricing from Q3 2026 onward, and how much DRAM-related burden GM factors into its outlook for 2027. If long-term contracts succeed not just in protecting supply but in tempering price volatility as well, that would give automakers more room to avoid passing costs broadly onto vehicle prices.